How to modificate Ultralinear to triode?
sv3eao
each chanel at Push-Pull Ultralinear. The anode voltage is 432V and the
screens take the voltage at the transformer's Ultralinear point from a 220R
resistor. The cathode voltage is setting for 400mv (45mA each, if i remember
corectly). How can i modificate to run as triode and what will be the
avantage and disavantage???
Many thanks.
Fotis.
Choky
--
Choky,Prodanovic Aleksandar,YU
my reply address is tweaked!!
sv3eao <sv3...@yahoo.com> wrote in message
news:a2bdnp$9un$1...@usenet.otenet.gr...
Wiz
Choky,
mind decoding this for us?
cheers/Carron
"LABOR SVGIT"
Ross Matheson
<a2bdnp$9un$1...@usenet.otenet.gr>,
"sv3eao" <sv3...@yahoo.com> wrote:
If it isn't broken don't try to fix it.
If you have to ask the price you can't afford it.
If you want to understand what a modification will do, learn some theory.
Consider why you have this idea that you can outguess the manufacturer of
your amp as to performance with the existing components with no knowledge.
Because you heard that "Triode Is Better" ?
However, now you have asked, others will no doubt try to impart some
theory or pointers to theory, or illustrations and comparisons. But you
will likely find that if you change it to triode mode you must change
several other parameters in the circuit to optimse it for friode mode, and
the output transformer plate to plate load for the ultralinear transformer
is fixed, so you have to work with that, unless you want to tear it apart
completely, and it may not be optimum for triode mode, so a best
compromise with the given voltages has to be found, and you will need to
alter the working point of the output valves against this load, changing
the grid bias, and ... all in all, you would be better off to start anew.
Don't mess with it on some sort of idea until you know what you are doing
and what the risks are. Every circuit is unique, however similiar. Really,
you are messing with design points in a commercial product and you need
the circuit to begin with, to understand the situation and consequences.
Having tried to indicate some of the disadvatages, and zero advantages, I
can tell you that it is as simple as disconnecting the screen 220R
resistors from the UL taps and connecting them to where the anodes connect
instead. Voila! Triode mode! But unless you know the OPT impedance and
what is best you are just hacking without knowledge on some sort of whim.
Further, you may damage your amp. Are you really so dis-satisfied with it?
Copland is a Scandinavian (Or Danish?) amp isn't it? Don't make it worse!
In addition, you'd get a lot less power, so the speakers are important.
RdM
Ultra Cautionary Tales, 2002.
My NZD 0.05c worth.
"You have to try!" [from 'Here I Am' - Lyle Lovett and his large band]
*Treat everything with skepticsm. Including this. Check it out yourself.*
*Context Is All*
"Only Connect!"
E.M Forster
Ross Matheson
<20020119063904...@mb-mh.aol.com>,
too...@aol.comSPAMKILL (Wiz) wrote:
This is happening all too frequently now, IMO.
Could we please ask people to actually type or cut&paste these out?
alt.binaries.pictures.radio
alt.binaries.radio.pictures
alt.binaries.schematics.electronic
As an example of the confusion familiarity has already spread, check
http://groups.google.com/groups?q=ABPR++ABRP++ABSE&num=50&hl=en
In any case, they're newsgroups where one can post binaries. However,
they're also imperfect and unreliable in reception, and not everyone gets
all the posts or messages, yet are still popular and useful if you don't
have a web page or one of those free photo archive sites to upload images
to ... maybe these should be made more use of. The binary newsgroups are
archived on the web though so you can still access missed posts.
Finding them is left as an exercise for the student:-
(Translation: I'm tired and hungry and want to leave off and cook dinner!)
http://www.usenet-replayer.com/
RdM
Dennis Moore
I don't know about the Copland. However, having done the
switch to triode, imo it was an improvement on Conrad-Johnson,
Dynaco, VTL, and Quicksilver amps. And none of them gave
any problem whatsoever.
Dennis
"Ross Matheson" <ste...@ihug.co.zz> wrote in message
news:3c494e1a...@news.akl.ihug.co.nz...
Michael Mueller
I consider myself a highly intelligent individual, who has been reading
RAT now for 4 years. I have never gotten involved in a flame before.
However, not all of us are hip to "ALL" abbreviations. I asked this same
question 2 weeks ago after hours of trying to find the answer on line. I
found that this news group is normally very patient with people who are
not as knowledgeable. I believe the responses to this question were
however very rude.
brett
Ross Matheson <ste...@ihug.co.zz> wrote:
> If it isn't broken don't try to fix it.
> If you have to ask the price you can't afford it.
> If you want to understand what a modification will do, learn some theory.
[bla bla bla]
what an arrogant rant that was, ross! wow.
once you get the instructions on how to do the mod, just do it and see
how it works. if you like it, great, if not then return it back. that's
the fun of it. if you think learning theory bla bla is fun, then do it,
if not, then don't. you don't need to do anything but have fun and enjoy
your hobby.
brett
--
brett
ple...@sympat.spam.no.ico.ca
(remove the obvious in the above address to email back)
Robert Casey
>
> what an arrogant rant that was, ross! wow.
>
> once you get the instructions on how to do the mod, just do it and see
> how it works. if you like it, great, if not then return it back. that's
> the fun of it.
Wouldn't the mod require a change in the negative frrdback loop
to really optimize it?
Ross Matheson
> Ultra cautionary indeed Mr. Matheson.
>
> I don't know about the Copland. However, having done the
> switch to triode, imo it was an improvement on Conrad-Johnson,
> Dynaco, VTL, and Quicksilver amps. And none of them gave
> any problem whatsoever.
brett <do...@email.me.com> wrote:
> [bla bla bla]
>
> what an arrogant rant that was, ross! wow.
Gentlemen, I apologise unreservedly. Thank you for your frank feedback.
Once again I made the mistake of posting when "tired and emotional" and
having had a few too many drinks without dinner. I took off on entirely
the wrong note and got too wrapped up in it. I shall caution myself not to
post without serious and sober reflection in the future. I'm embarrassed.
I see that it came across as patronising, pompous, arrogant. I'm sorry.
It serves me right for trying to be too smart, and being stupid instead!
My apologies also to Fontis. I had no right to write so personally and
rudely in reply to a stranger who had a fairly straightforward question.
That said, I do believe there is sometimes a temptation to fiddle without
understanding, based on hearsay, that might lead to unexpected results,
but I accept now, looking at a discussion of EL34 in triode, UL, and
pentode in one of the Philips books, that no great harm should come of
trying it in this case. The drive voltage required will be about the same;
the a-a load for UL might be somewhat less than ideal for triode mode for
lowest distortion and class A working, and if it is cathode biased (which
isn't clear) he may get away with the same value resistor or resistors
(the Phillips book mentions 470ohm/tube at 400Va in UL & triode, with 6.6k
and 10k loads respectively) although since it is unknown whether this is a
fixed bias or cathode bias amp, and what the current OPT a-a load is, and
how far into Class AB vs A it is biased, and how much feedback it uses, I
would still suggest that results will be uncertain, or at least could be
better optimised with some more knowledge of the circuit parameters.
One certain result is as I mentioned that P out will be less; maybe half.
As I say, I do accept now that no harm should occur;- try it and listen.
Even better, make some measurements in comparing the two if possible.
Also, could he try changing the 220R resistors to 100R or even leave them
out and tie the screens directly to the plates? 45mA per tube? seems maybe
a little low to me for very much Class A working, but the a-a load is
unknown. The 400mV mentioned seems like a typing error or else it is
perhaps a measurement across a low value resistor in the cathode circuit?
This may indicate that it is a fixed bias design, but not necessarily!
I'm really no expert though; nearly all of my knowledge is from reading
rather than doing, and I have not actually played with EL34's so I shall
now pull my head back in and go and attend to work I should be doing.
Sufficiently contrite and suitably ashamed of myself, I hope, in apology.
(I'll certainly try to resist future temptations to comment so stupidly!)
My intention was to be helpful, however stupid and smartarse it seemed.
As they say, though, "The road to Hell is paved with good intentions!"
Please do correct any errors I may have made in my guesswork above.
Ross Matheson
Ross Matheson
> 19 Jan 2002 11:39:04 GMT, in rec.audio.tubes
> <20020119063904...@mb-mh.aol.com>,
> too...@aol.comSPAMKILL (Wiz) wrote:
>
> > >Go to ABPR or ABRP or ABSE
> >
> > Choky,
> >
> > mind decoding this for us?
> >
> > cheers/Carron
> > "LABOR SVGIT"
>
> This is happening all too frequently now, IMO.
> Could we please ask people to actually type or cut&paste these out?
>
> alt.binaries.pictures.radio
> alt.binaries.radio.pictures
> alt.binaries.schematics.electronic
By this I meant, in case it wasn't clear, to suggest that when people
announce a binary post in one of the above groups, they use the full group
name rather than an acronym, to save others having to ask every time.
It'd be easy enough to cut & paste it from the header in the binary post,
and it might save people a lot of confusion. They're not common acronyms!
Cheers,
Ross Matheson
Tim
Hey guys, this reminds me -
Why do triodes _*always*_ have less power than a tetrode or pentode
equivalent?
Does it have something (everything) to do with the greatly lower plate resistance?
Tim
--
"WOOHOO! Who would've guessed reading and writing would pay off!"
- Homer Simpson
Patrick Turner
Tim wrote:
> > How can i modificate to run as triode and what will be the avantage and disavantage???
>
> Hey guys, this reminds me -
>
> Why do triodes _*always*_ have less power than a tetrode or pentode
> equivalent?
>
> Does it have something (everything) to do with the greatly lower plate resistance?
>
> Tim
The ability of triodes to change the voltage at the plateis determined by the grid voltage
change, but
regardless of load, there is a limitation of grid voltage
change when the grid starts to go positive when turning on the tube.
In many power tubes, the grid loses control over the plate current
at this point.
Perhaps you should try to understand load lines and plate curves, and all
will become clear.
Some tubes, say a 6L6, can have their grids driven into the positive voltage region
when in triode, pentode, or UL.
But grids driven positive require a sudden huge
current ability, and normally cause
a drive amp to cease producing a linear drive voltage,
so a direct cathode follower should be used for getting more
triode power from tubes like a 6L6 operating
in class AB2 mode.
A negative grid does not attract electrons, but soon as it goes
positive, electrons are attracted.
Patrick Turner.
sv3eao
I am very sorry for the confuse that i made to the group...
For me the audio is ONLY a hobby, not my job. I am not electronic but
audio-radio amateur. I know enough theory about semiconductors but not for
tubes. I think that is not a good idea to read the Philips or Mullard
Handbook or make PSpice emulations for "Disconnect the 220R screen resistor
at the transformer end and resolder it
to the anode (plate). Do that for each EL-34." that Mr Andre Jute wrote
me...
But no problem at all. I receive every note or suggestion with pleasure. I
like theory but i haven't the time to learn it, so there are the newsgroups
with people who know more than me.
Here is the schematic of my amplifier that i have copy with my pencil from
the pcboard. http://www.qsl.net/sv3eao/cta402.gif
So, for Mr Matheson and the group, the bias psu has -46V. The dealer's Fax
write to adjust bias for 400mv (45mA per tube) at cathode check point. The
resistor from the cathode to earth is 10R. Is a good idea to work it hoter
for better linearity ? For examle 450 or 500mv (at cathode's check point)
will be better? Any other trick for even better linearity? About the HV psu
there is no problem, it is big enough. The feedback is maybe about 6db and
no more than 10db.
I asked for the triode mode because i saw at another Copland's model a
triode/pentode front panel switch and i want to make it to test if i like
it. If no i will back again at the first potition.
My regards to all...
Fotis from Greece.
Paul D. Spiegel
> triode/pentode front panel switch and i want to make it to test if i like
> it. If no i will back again at the first potition.
That's the right approach. Give it a try. Andre's instructions tell
you how to do it.
Ross Matheson
<a2eb8v$a9t$1...@usenet.otenet.gr>,
"sv3eao" <sv3...@yahoo.com> wrote:
> Fisrt, Thank you all for your suggestions.
> I am very sorry for the confuse that i made to the group...
Not a problem. OK. Understood. Sorry to take so long to reply ...
> For me the audio is ONLY a hobby, not my job. I am not electronic but
> audio-radio amateur. I know enough theory about semiconductors but not for
So it seems! I was impressed by your amateur equipment DIY projects.
I'm the opposite;- I picked up valves/tubes in my teens but NOT
semiconductors, and returned to the interest again in my 40's, intending
also now to revive the amateur operator certificate I got back then and
build a valve 2m transmitter and a few other things ...
> tubes. I think that is not a good idea to read the Philips or Mullard
> Handbook or make PSpice emulations for "Disconnect the 220R screen resistor
> at the transformer end and resolder it
> to the anode (plate). Do that for each EL-34." that Mr Andre Jute wrote
> me...
Well, I said, too, toward the end of my piece, first off - :
: Having tried to indicate some of the disadvatages, and zero advantages, I
: can tell you that it is as simple as disconnecting the screen 220R
: resistors from the UL taps and connecting them to where the anodes connect
: instead. Voila! Triode mode!
That really *is* the process. Find the wire from the transformer that
connects to the 220R resistor (your R1 in cta402_Out.gif) that connects to
pin 5 (screen/G2). I imagine the amp is built using a PCB. Ideally that
resistor is quite close to, or connected directly to pin 5. IF R1 is large
enough, and its leads are long enough, you MAY be able to unsolder it at
that transformer wire end and attach it instead to pin 3 (plate/anode).
If R1 isn't long enough or you think it may stress it too much to bend it
around then do this instead. Leave it where it is and instead unsolder the
transformer lead. Put insulating tape on the exposed end and tuck it or
tie it out of harms way. Then cut a small piece of spare wire to go from
that hole to pin3. This is probably a MUCH better way to do it, in fact.
So now the U/L tap wire is disconnected, and pin 5 connects through R1 to
pin 3. If you were to contemplate a switch you'd take that wire to it, and
have it switched between pin 3 and the original transformer U/L tap wire.
Best to leave the 220R in circuit. Others may use 100R, or connect it
directly, but it is there to stop potential oscillation, and does no harm.
Mullard used 1k once, "for increased linearity", but HiFi World's version
designers thought it unecessary on their measurements of the 5-20.
That is all there is to it. Be SURE that you have let the capacitors
discharge fully before you switch it - you can check that the HT voltage
has diminished to zero at your HV test points, and time it for reference.
NEVER switch over while the amp is turned on; you may damage it.
[Observe safety precautions!] [There are LETHAL voltages in a tube amp!]
I was being cautionary precisely because your question indicated that you
did not know a lot about tubes, and also because I saw someone mention in
an EL34 shootout that the Copland amps are "highly optimised" designs ...
& so I said it may not be optimised for (or, say, optimal in) triode mode.
I still hold to that view. It may sound different, but may not be ideal.
The feedback network values may not be as suitable, either, in triode.
{Yes, I was being discouraging, as in a kind of counterpoint to the
happy-go-lucky "she'll be right mate" approach that some promote, but you
have shown enough "nous" that you can probably grasp the complexities:-}
As always, you proceed at your own risk. Without knowing the impedance
ratio of the OPT it is hard to predict a best bias point for triode mode.
You could disconnect it and measure* the turns ratio: square it for the
impedance ratio. This might give some insight into operating conditions,
Measuring the voltage drop across your R2 in the HV PSU between the test
points at both idle and near full power (before any distortion) could be a
useful indicator of current at full power, though it includes preamp mA.
(Use music, don't worry about a signal generator - just an indication)
Actually, there is an UL mode discussed and used in the Philips book
called "low loading", where the tubes are run at low dissipation and can
provide full power on music but would self destruct on a sine wave beyond
a few watts, so play it safe! Yours may not be so. Try it on music anyway!
(In Other Words: Don't worry if you don't have an audio signal generator!)
The "low loading" approach uses auto (cathode) bias set as though for grid
bias, IIRC, and yours is already grid ("fixed") bias, albeit the cathode
10R is bypassed for AC by the 100uF cap, so it would seem ~ "fixed" bias.
(Which of course also means "fixable", as you can set the value.)
However, I am a bit puzzled by the -46V mentioned, as well as the 45mA, as
this seems radically different from the classic Mullard UL circuit for 2 x
EL34 in UL or triode, where the figures per tube seem more like 25 Vin and
80mA per tube at >30W, maybe 70mA and 17V in for P out=20W.
(I'm just looking at characteristic charts, not claiming to know;- Ideas?)
Could this be a misinterpretation of 46V pk to pk signal input to grids?
(I'd really like to try to understand what is going on in this circuit! -)
If it is biased at -46V per tube, for 45mA Ik, what does this imply?
**
The Mullard and Phillips books in question were books on ultralinear
amplifiers with the EL34 and EL84 (and an ECL82 3W SE & 7W PP), and there
is some introductory detail on UL vs Pentode & Triode conditions for both.
If anyone is interested, they are:
HI-FI AMPLIFIER CIRCUITS By E. Rodenhuis 1960 Philips Technical Library
Mullard CIRCUITS FOR AUDIO AMPLIFIERS
The Mullard is like an earlier version and gives precise physical
construction details, unlike the Philips which goes into seeming more
theory detail and adds more amplifiers, including a p-pp EL84.
There's a currently sold book which must be similiar,
Mullard Tube Circuits for Audio Amplifiers <looks it up...>
I see it can be found on www.google.com , and just now am looking at a
sample Contents page from Amazon and it seems identical to the chapters of
the early Mullard above. (I uh, photocopied the bulk of my two, years ago)
**
Anyway, back to the subject at hand ...
> But no problem at all. I receive every note or suggestion with pleasure. I
> like theory but i haven't the time to learn it, so there are the newsgroups
> with people who know more than me.
(You'll probably like the quote I just found, below my signature:-)
> Here is the schematic of my amplifier that i have copy with my pencil from
> the pcboard. http://www.qsl.net/sv3eao/cta402.gif
And the others in the PSU thread. It has been interesting looking at them.
> So, for Mr Matheson and the group, the bias psu has -46V. The dealer's Fax
> write to adjust bias for 400mv (45mA per tube) at cathode check point. The
45mA would be 450mV over 10R. There is a small error here I think.
> resistor from the cathode to earth is 10R. Is a good idea to work it hoter
Hotter? Maybe, maybe not.Try it and listen, play with the bias if you
want. However, the bias may be set where it is for a reason at present;-
if you alter it experimentally in triode mode you may need to put it back
where it was when returning to UL mode, IF the designer knew best! (:-))
Without measuring, you cannot know what it is doing ... only listen.
Some may say that is enough, but you can only guess at distortion, and you
don't know what the actual low and high power conditions are or will be.
> for better linearity ? For examle 450 or 500mv (at cathode's check point)
> will be better? Any other trick for even better linearity? About the HV psu
Trick? Treat? Hmm. Hopefully the PSU has some capability for current.
Unless you are able to measure linearity, how will you evaluate "tricks"?
> there is no problem, it is big enough. The feedback is maybe about 6db and
> no more than 10db.
Here is the page for CTA402. (It has a typo, 53W instead of 35W later)
http://www.datateam.hu/pages/copland/cta402.html
There is no mention of how many dB feedback. (Is this hearsay or known?)
6-10dB is a quite low figure.The OPT descriptions sound reasonable.
They mention a 210W AC input to the amp only. Could be constant or idle.
> I asked for the triode mode because i saw at another Copland's model a
> triode/pentode front panel switch and i want to make it to test if i like
> it. If no i will back again at the first potition.
(Back at the start position!) (Oh well, if that's the worst it isn't bad!)
I assume that it is this model?
http://www.datateam.hu/pages/copland/cta505.html
6550 PP. 67 W <1%D ultralinear, 35 W 1% triode
"The power supply is drawn from a new 450
voltage ampere custom build torodial
transformer which feeds a 6x470 mikroFarad
capacitor reservoir. The innate quality of the
amplifier perits the adjustable bias to be set
for minimum variation of plate and screen
current in the output stage, whilst a minimal
18 dB negative feedback produces the
optimum performance, achieving an
agreeable parity between the dynamics and
the smooth character of the amplifier"
"Power Consomption (sic) 250 W (350 W at max output) "
This is a totally different amp & perhaps designed to be switchable.
It may be that yours is perfectly capable of a corresponding treatment.
On the other hand, results may be variable. Who knows? Try it and see, but
be intelligent about it and realise there is more to it, yet, to be known.
(Which could be fun to learn!!) (I certainly hope so, for myself!)
Cheers and good experimenting!! (and intelligent data acquisitions:-)
> My regards to all...
> Fotis from Greece.
Regards to you all as well! (Hey, where *are* the EL34 loadline experts?)
[Chris Merren?]
Ross Matheson
[Learner Licence :-]
New Zealand
(All the above is simply my own thought or opinion at this moment.)
((Honestly !!:-)) (I haven't any further clues!) (I'm even just guessing!)
"There is no answer. There is no question. There is only the road, which
stretches out in front of you, an endless night of coffee, technical
manuals, snack food wrappers, disgruntled family members,
and hundreds of hours spent struggling to become adept.
Or you can hire a consultant." -Tom Oehser.
Fotis Visilias
Two days before your answer, i made the modification from UL to "triode" but
i disapointed.
The amplifier is pcb constructed (no point to point wired) so, it was so
easy to disconect the small wire from UL' tap and then to solder it to
anode's transformer tap. All about my and amplifier's safety was allready
known.
The sound after the mod wasn't so good......
The mid and high frequencies was clear and transparent ("maybe" litle better
from UL) but the lower end was just unacceptable!!! The men voices was
sounding more thin and the bottom end of drums set was sounding like
plastic!!
I tried without negative feedback, too but i didn't like it at all.
So i came back to UL's mode with PLEASURE!!!
About the bias infos, i said that the bias PSU has -46V out but i haven't
measure what is the excactly voltage after the drop registors and trimmers
to the EL34. I think it will be much lower than -46V. Before a year i wrote
a letter to Copland's design team for bias adjustment information and they
answer me, to check across R2 cathode's resistor for 400mV=45mA for each
tube. You can read the answer at:
http://www.qsl.net/sv3eao/cta402_biasInfo1.gif and at
http://www.qsl.net/sv3eao/cta402_biasInfo.gif
I wrote to Copland's team beacause a dealer say me to check for 450mV the
other dealer for 480mV, the other 500mV, e.t.c...
The power of CTA 402 is 35W not 53W and you are right about CTA 505.
And for the final i think that you are right and experienced. Every mod that
i tried about UL/triode or with/without feedback the best results was the
designer's prefered...
I am sure that your question will be: " why you are looking to modificate
this sofisticated design amplifier?"
I tell you why. Before CTA 402 i has the CTA401 for more than 5 years.
http://www.qsl.net/sv3eao/cta401.jpg .
I selled it and i bought the CTA 402. It sounds much more extended with
transistor like bass, beautifull open mid and extended wonderfull tremple
and has 35W instead of 25W of CTA 401, BUT i think that is a bit off,
coherence and harmonic. It has an all better bandwidth, clarity and power
than CTA401, but is a bit "colder" than CTA 401, too.
So i will try tomorrow to bias it little "harder" (maybe for 450-500mV) to
hear the results. If i will take a bit of honey (harmonics pleasure at my
ears) instead of power, this will be the best amp for me...
Another point that i think is to start changing tubes ...but isn't so good
idea to buy all of them to test how they sound.
Until now my amp has Svetlana output and all the others are Edicron.
My regards
Fotis Visilias http://www.qsl.net/sv3eao
"Ross Matheson" <ste...@ihug.co.zz> wrote in message
news:3c4fc36a...@news.akl.ihug.co.nz...
Fred Nachbaur
I'm glad there's at least one other person in the world who can't stand
the sound of triode-connected pentodes.
UL is still, in my not always humble opinion, the best output stage
topology by a long shot.
I also like global NFB in judicious amounts, which makes me doubly
unpopular in some circles. That's ok, I know what I like. ;-)
Cheers,
Fred N
Choky
--
Choky,Prodanovic Aleksandar,YU
my reply address is tweaked!!
Fred Nachbaur <fnac...@netscape.net> wrote in message
news:3C50794D...@netscape.net...
Fred Nachbaur
Fred N
Choky wrote:
> How dare you to have your own opinnion? ;))
>
> --
> Choky,Prodanovic Aleksandar,YU
> my reply address is tweaked!!
> Fred Nachbaur <fnac...@netscape.net> wrote in message
> news:3C50794D...@netscape.net...
>
>>Hi Fotis,
>>
>>I'm glad there's at least one other person in the world who can't stand
>>the sound of triode-connected pentodes.
>>
>>UL is still, in my not always humble opinion, the best output stage
>>topology by a long shot.
>>
>>I also like global NFB in judicious amounts, which makes me doubly
>>unpopular in some circles. That's ok, I know what I like. ;-)
>>
>>Cheers,
>>Fred N
>>
>>>[...]
--
+---------------------------------------+
| Music: http://www.netidea.com/~fredn/ |
| Vacuum Tube projects & other stuff: |
| http://www.fna.muohio.edu/dogstar/ |
+---------------------------------------+
Tim
Screw you! Sometimes I eat it for all 3! :^))
> > How dare you to have your own opinnion? ;))
BTW Fred, you're not alone, last time I tried triode on my pentodes
(namely 6L6 SE I think), I ended up leaving it in pentode...
Tim
Ross Matheson
in rec.audio.tubes<a2prmd$dpq$1...@usenet.otenet.gr>,
on Thu, 24 Jan 2002 22:42:57 +0200,
> Two days before your answer, i made the modification from UL to "triode" but
> i disapointed.
> The amplifier is pcb constructed (no point to point wired) so, it was so
> easy to disconect the small wire from UL' tap and then to solder it to
> anode's transformer tap. All about my and amplifier's safety was allready
> known.
> The sound after the mod wasn't so good......
Ah, a small mistake!? If you have disconnected the UL wire and soldered *it*
to the anode, then you have just shorted out that part (43% usually!) of the
transformer and presented the amplifier with a much smaller load!
You are meant to disconnect the UL tap and leave it disconnected!
Then, connect the screen/G2 (through the 220R resistor - ie take the other
end of the resistor that USED to connect to the UL tap and connect it) to
the anode, instead of the UL tap. So the screen and anode are effectively
joined together (by the 220R) and both join the transformer only at the
anode connection. Insulate the UL tap wire because it has HV on it.
Make sure that the tubes have exactly equal DC bias each!
These DC currents are meant to exactly cancel out in the output transformer,
in terms of (not!) creating a magnetic field.
> So i will try tomorrow to bias it little "harder" (maybe for 450-500mV) to
> hear the results. If i will take a bit of honey (harmonics pleasure at my
> ears) instead of power, this will be the best amp for me...
> Another point that i think is to start changing tubes ...but isn't so good
> idea to buy all of them to test how they sound.
> Until now my amp has Svetlana output and all the others are Edicron.
>
> My regards
> Fotis Visilias http://www.qsl.net/sv3eao
Another possibility might be to try different speakers?
But try the "triode" mode correctly, as above, first!
Regards, Ross.
Ross Matheson
> Another point that i think is to start changing tubes ...but isn't so good
> idea to buy all of them to test how they sound.
> Until now my amp has Svetlana output and all the others are Edicron.
>
> My regards
> Fotis Visilias http://www.qsl.net/sv3eao
This might be interesting for you then. (It's all opinion of course!)
EL34s Comparisons
http://www.audioasylum.com/audio/tubes/messages/73365.html
Ross Matheson
John Byrns
<ste...@ihug.co.zz> wrote:
> Ah, a small mistake!? If you have disconnected the UL wire and soldered *it*
> to the anode, then you have just shorted out that part (43% usually!) of the
> transformer and presented the amplifier with a much smaller load!
It isn't clear whether Fotis made "a small mistake" or not. You may be
right about the mistake, and indeed the amplifier would be presented with
a much smaller load indeed, just the DC resistance of the shorted section
of the winding, as reflected to the remaining active portion of the
primary. However it is possible that Fotis made the connections
correctly, which is how I read it first time around. When I first read
the post, I assumed the tube sockets and screen resistor were on a PC
board, with leads running from the PC board to an output transformer that
had terminals, rather than leads. If that is the case with this
amplifier, then removing the screen lead from the transformer, and then
soldering it to the anode terminal would be the correct way to do the
modification. Fotis will have to clarify how his amp is built and how
exactly he did the mod?
> You are meant to disconnect the UL tap and leave it disconnected!
> Then, connect the screen/G2 (through the 220R resistor - ie take the other
> end of the resistor that USED to connect to the UL tap and connect it) to
> the anode, instead of the UL tap. So the screen and anode are effectively
> joined together (by the 220R) and both join the transformer only at the
> anode connection. Insulate the UL tap wire because it has HV on it.
There is the possibility that this is exactly what Fotis did, at least
that's how his description seems to read.
When people talk about triode moding an amplifier, I always wonder about
the operating point and load line of the tube. As a pentode, or in UL
mode, most tubes are run into load impedance's that are too low for proper
triode operation. It may be worth while trying a higher load impedance by
connecting the speakers to the next lower tap on the transformer, to raise
the primary impedance. For example try connecting an 8 Ohm speaker to the
4 Ohm tap. The bias that was used for UL operation may also not be right
for triode operation, although this is a complicated issue, and depends
among other things on how the UL amp was originally biased, if it was
biased deep into class AB, then the bias may be OK for triode operation as
is, and this may be the case with Fotis's amp.
Regards,
John Byrns
Surf my web pages at, http://www.enteract.com/~jbyrns/index.html
Fotis Visilias
*it*
> to the anode, then you have just shorted out that part (43% usually!) of
the
> transformer and presented the amplifier with a much smaller load!
>
> You are meant to disconnect the UL tap and leave it disconnected!
> Then, connect the screen/G2 (through the 220R resistor - ie take the other
> end of the resistor that USED to connect to the UL tap and connect it) to
> the anode, instead of the UL tap. So the screen and anode are effectively
> joined together (by the 220R) and both join the transformer only at the
> anode connection. Insulate the UL tap wire because it has HV on it.
No mistake at all. I means that i soldered the UL's wire to anode point but
NEVER sortened the transformer's UL point.
> Make sure that the tubes have exactly equal DC bias each!
> These DC currents are meant to exactly cancel out in the output
transformer,
> in terms of (not!) creating a magnetic field.
It was the same value for all tubes...(400mv)
> Another possibility might be to try different speakers?
Noway, i use these speakers (Triangle Zays) for more than 3 years and i
enjoy their sound.
Thank you for your very nice link about tube comparisons, Ross.
Fotis Visilias
http://www.qsl.net/sv3eao
Bob Groschen
"Fotis Visilias" <sv3...@yahoo.com> wrote in message
news:a2s371$8ol$1...@usenet.otenet.gr...
> No mistake at all. I means that i soldered the UL's wire to anode point
but
> NEVER sortened the transformer's UL point.
I think what we have here is a failure to communicate....sigh.
Fotis, to *correctly* convert to triode mode, there must be *NO*
connections
to the UL taps on the transformer. None, nada, zip, zilch. So you must
unsolder the UL taps from the transformer, fold over the ends and slip a
piece of heat-shrink over the end to prevent shorting.
Then you can tie the screens to the plates (either directly or through
series resistors
the latter being the better idea).
*Then* try the listening test again.....
--
Best Regards,
Bob Groschen
Bob Groschen
"John Byrns" <jby...@enteract.com> wrote in message
news:jbyrns-2501...@216-80-98-79.d.enteract.com...
> In article <3c51552...@news.akl.ihug.co.nz>, Ross Matheson
> <ste...@ihug.co.zz> wrote:
>
> > Ah, a small mistake!? If you have disconnected the UL wire and soldered
*it*
> > to the anode, then you have just shorted out that part (43% usually!) of
the
> > transformer and presented the amplifier with a much smaller load!
>
> It isn't clear whether Fotis made "a small mistake" or not. You may be
> right about the mistake, and indeed the amplifier would be presented with
> a much smaller load indeed, just the DC resistance of the shorted section
> of the winding, as reflected to the remaining active portion of the
> primary. However it is possible that Fotis made the connections
> correctly, which is how I read it first time around.
I read it the same way Ross did and I *was* going to poin tthis out, but
saw several posts on this thread so I assumed that someone else (like Ross)
would point this out.
> When I first read
> the post, I assumed the tube sockets and screen resistor were on a PC
> board, with leads running from the PC board to an output transformer that
> had terminals, rather than leads. If that is the case with this
> amplifier, then removing the screen lead from the transformer, and then
> soldering it to the anode terminal would be the correct way to do the
> modification. Fotis will have to clarify how his amp is built and how
> exactly he did the mod?
Agreed.
> There is the possibility that this is exactly what Fotis did, at least
> that's how his description seems to read.
Sorry but I read it 180 out from this. We'll just have to wait for Fotis to
fill us in....
Fotis Visilias
GGGGRRRRRRRRRRRRRR!!!!!!!!!!!!!!!!!!!!!
It was CCOOORRRRREEEEEECCCCCCCCCTTTTTTTT........
Fotis.
Fotis Visilias
GGGGRRRRRRRRRRRRRR!!!!!!!!!!!!!!!!!!!!!
It was CCORRREEEECCCCTTTT........
And i mean that : there are 2 pieces of wire coming from the pcb to Output
transformer. The one is going from 220R (screen resistor) to transformer's
UL point and the other from the EL34's adode to the anode's transformer
point. I disconected the wire of screen , but from the transformer's side
not the pcb's side and then soldered to the other wire (anode). So the UL
points of transformer was disconected.
Thank you
Fotis Visilias
Ross Matheson
on Sat, 26 Jan 2002 09:25:55 +0200,
> GGGGRRRRRRRRRRRRRR!!!!!!!!!!!!!!!!!!!!!
> It was CCOOORRRRREEEEEECCCCCCCCCTTTTTTTT........
>
> Fotis.
(8=}) !!
Big Grin!)
OK, we get the message;-)
Sorry it didn't work out to sound better for you.
Still, it does show that people were concerned that you got it right!
I was a little uncertain when I posted (nearly 2am and I had just got home)
but felt that there just may have been a possibility of a misunderstanding,
so I went ahead... Ah, well, where to from here?
(Maybe this is how manufacturers entice you into the upgrade path;- if you
can't improve on their midrange product yet want more, they have an even
bigger and better model to show you!)
Having gained some familiarity with tubes now, do you think you might ever
consider building one yourself, whether from a kit or a known good design
(schematic) with quality components? Otherwise, you look at other commercial
products until you find one that suits you even better than this one.
Given that you are happy with your speakers, and presumably your source
devices, CD or whatever, I guess you either live with the amp, or upgrade.
Best of Luck!
Ross
Fotis Visilias
> OK, we get the message;-)
> Sorry it didn't work out to sound better for you.
> Still, it does show that people were concerned that you got it right!
>
> I was a little uncertain when I posted (nearly 2am and I had just got
home)
> but felt that there just may have been a possibility of a
misunderstanding,
> so I went ahead... Ah, well, where to from here?
>
> (Maybe this is how manufacturers entice you into the upgrade path;- if you
> can't improve on their midrange product yet want more, they have an even
> bigger and better model to show you!)
But i don't want to play this game...
> Having gained some familiarity with tubes now, do you think you might ever
> consider building one yourself, whether from a kit or a known good design
> (schematic) with quality components? Otherwise, you look at other
commercial
> products until you find one that suits you even better than this one.
> Given that you are happy with your speakers, and presumably your source
> devices, CD or whatever, I guess you either live with the amp, or upgrade.
I think that the perfect way is to upgrade some critical components with
much better for the final tuning.
I just need a bit ( and i mean only a bit ) of harmonics.
So, thanking all the people on the group that answered or suggested to me,
my next questions on the newsgroup will be about the components replacing.
( i hope to make a much smaller confussion than i made at this group. OK i
am joking...)
>
> Best of Luck!
>
> Ross
Good luck to all and thank you for your patience!
Fotis Visilias
http://www.qsl.net/sv3eao
John Byrns
<sv3...@yahoo.com> wrote:
> GGGGRRRRRRRRRRRRRR!!!!!!!!!!!!!!!!!!!!!
> It was CCOOORRRRREEEEEECCCCCCCCCTTTTTTTT........
They are just not going to believe you until you say that it sounded much
better in triode mode. Everyone knows pentodes sound best connected as
triodes, and if you don't report this result, you must have done something
wrong.
John Byrns
<sv3...@yahoo.com> wrote:
That is exactly what I heard you say you did the first time, I don't know
why people have such a hard time reading plain English, and or want to
change the meaning of a statement around? Hopefully you have made it so
clear this time that no one can say you did it wrong.
Choky
--
Choky,Prodanovic Aleksandar,YU
my reply address is tweaked!!
Fred Nachbaur <fr...@netidea.com> wrote in message
news:3C50C2D0...@netidea.com...
Preamp
wrote:
>In article <a2tlnv$ol2$1...@usenet.otenet.gr>, "Fotis Visilias"
><sv3...@yahoo.com> wrote:
>
>> GGGGRRRRRRRRRRRRRR!!!!!!!!!!!!!!!!!!!!!
>> It was CCOOORRRRREEEEEECCCCCCCCCTTTTTTTT........
>
>
>They are just not going to believe you until you say that it sounded much
>better in triode mode. Everyone knows pentodes sound best connected as
>triodes, and if you don't report this result, you must have done something
>wrong.
>
>
<grin> My SE EL34's sound best in UL (I've tried pentode and triode
mode). BUT, they still don't sound as good as my 6B4G's.
Patrick Turner
Preamp wrote:
I have heard too many deleriously sublime UL amps,
both PP, and SE, to be able to say that UL is not too good,
and worse than triodes.
Nearly everytime I have tried to hear a difference, I couldn't.
Blokes I know have tried UL, and then tried triode, and
I was there when they did it, and there was not a not to hear.
As for the measurements, well, they don't mean much, but
most UL amps are able to make more power than triode
at the SAME level of distortion.
So a ten watt triode amp might have 1% THD, but the UL
will make 15 watts at 1%, while the same two beam power or tetrode
tubes, with the same amount of NFB applied, will have
more THD at 10 watts than either triode or UL.
It is important how we compare these things, lest our
comparisons and reasonings be silly.
Then there is the question of Z out,
and the % of class A in the operating conditions.
One thing is for sure, and that is SET amps will have a very different
distortion profile than PP.
When that THD is a high figure, my ears don't like it much,
but to some folks, it is very pleasant.
It depends on how you use the amp, and the sensitivity
of speakers, and the load matching to some extent.
With Intermodulation distortion, a single triode will
have more tube gain when there's more tube current flow,
as the tube is in the negative going plate voltage part of the cycle,
and less in the positive part of the cycle.
This means the higher frequencies become AM modulated by
the lower frequencies, and the change of amplitude of the HF
is at the rate of the LF.
But with PP, especially in class AB amps,
changes in output tube gain occur on both positive
and negative peaks of a wave, as both parts are produced by tubes
in the output stage turning on. So the frequency of the
modulation of a HF wave by a LF wave is twice that of SE.
I would have thought therefore that the mix of distortion
products is of greater number in PP, than in SE,
although the PP amp will have its distortion at a higher level
than a PP amp.
It seems highly probable that people prefer the
distortions of SE more than those of PP.
That is because the second harmonic transfer action
creates IMD products which are less unpleasant than
the same amount produced by third harmonic
transfer curve, ie, the famed S bend, of most PP amps.
It all probably has an irrelevent importance when the distortion
is less than 0.1% with normal listening, and the amp didn't have
much distortion to begin with, before the NFB was added.
I think SE is for special occasions, with special speakers,
and PP for when you need / want a fair bit of grunt,
although it depends on the power of one's SE.
I find my 22 watt SEUL are just fine, for 90% of the time.
But I don't want to wear out rare tubes, so I use
PP most of the time, which is heavily class A,
and the sound is clear, well defined, natural,
and with that touch of warmth and creaminess
that simply must have been in the original recording
or else how does it get there?
Patrick Turner.
Choky
me well)) with modest 40mA per tube via 410 volts-switching between triode and UL mode is nothing
dramatic.
That means that UL is not much worse than triode; in any case-I spotted difference only with
power hungry mini monitors,when I really can't decide what to like more-sweeter triode or more
ooohmp from UL.
I tell ya-Zen-almost that is not important-which config is in use;
If someone's amp is already good enough,in that case any change not must be big step
,especially in forward direction.
Note-accent is on -"good enough"
--
Choky
Prodanovic Aleksandar
choky*remove*@eunet.yu
YU
remove *remove* to reply!!
"Patrick Turner" <in...@turneraudio.com.au> wrote in message
news:3C53FF83...@turneraudio.com.au...
Patrick Turner
Choky wrote:
> ZM like class A,but in case of my modest PP EL34 m-blocks- (Raa 'round 4K3 (if my memory serves
> me well)) with modest 40mA per tube via 410 volts-switching between triode and UL mode is nothing
> dramatic.
> That means that UL is not much worse than triode; in any case-I spotted difference only with
> power hungry mini monitors,when I really can't decide what to like more-sweeter triode or more
> ooohmp from UL.
> I tell ya-Zen-almost that is not important-which config is in use;
> If someone's amp is already good enough,in that case any change not must be big step
> ,especially in forward direction.
> Note-accent is on -"good enough"
Then Choky, when one gets to "good enough"one can then focus on other aspects of the sound system,
instead of being completely possessed by the demons
of doubt about one's amp.
One can move to speakers, preamps, turntables,
maybe tackle CD transports, and DtoA converters,
if only one had the time all this takes.
Sometimes it is necessary to get the hammer and saw out,
and do some serious building work, to improve room accoustics.
I am glad I did, all those years ago,
when youth allowed me my dreams.
Patrick Turner.
Choky
dream above dreams?
ya know- Choky studied (and finished) Construction (Building) College (is that proper term for
School just before University?).
After that I was one year on Construction University,after that I switch to Technical and almost
finished it............
So,maybe we are bros not via tubes only..... ;))
--
Choky
Prodanovic Aleksandar
choky*remove*@eunet.yu
YU
remove *remove* to reply!!
"Patrick Turner" <in...@turneraudio.com.au> wrote in message
news:3C54243D...@turneraudio.com.au...
Preamp
<in...@turneraudio.com.au> wrote:
>
>Then Choky, when one gets to "good enough"one can then focus on other aspects of the sound system,
>instead of being completely possessed by the demons
>of doubt about one's amp.
>One can move to speakers, preamps, turntables,
>maybe tackle CD transports, and DtoA converters,
>if only one had the time all this takes.
Couldn't agree more.
>
>Sometimes it is necessary to get the hammer and saw out,
>and do some serious building work, to improve room accoustics.
>I am glad I did, all those years ago,
>when youth allowed me my dreams.
>
Sadly, that will have to wait until the kids move out and I can
reclaim the family room. :-)
Paul D. Spiegel
> mode). BUT, they still don't sound as good as my 6B4G's.
I had a similar experience with PP EL34's. I preferred UL to
triode connected. Best for me in pentode mode was a separate regulated screen
supply.
I later made a switch to 6B4G's and have never looked back.
Denis
> Everyone knows pentodes sound best connected as
> triodes, and if you don't report this result, you must have done something
> wrong.
>
This is almost entirely wrong. Indeed, the full exploitation of unique
features of the pentodes requires extensive engineereng considering
both the schematic topology and output transformers. It is a pity, 90%
of known pentode amps share the same design errors. There is similar
situation with triode amps too, and so much overhyping with SE
designs.
If done properly, the triode connection of some power pentodes is
helpful in some driver circuits.
Regards,
Denis Afanassyev
Trip Rifton
triode or ultralinear, the triode curves are much more linear. Ultralinear
curves are more linear than pentode. More linear = more accurate. I find
it odd that someone would prefer the sound of something inherently less
accurate, but then lots of people like the sound of solid state devices.
- Trip
"Denis" <denis_af...@mail.ru> wrote in message
news:32f1783d.02012...@posting.google.com...
Fotis Visilias
redesign a triode mode el34 amplifier from the beginning, "maybe" it will
sound better than UL , but at what speakers? BUT, my suggestion was that my
UL amplifier was sounding bad, when i just modificated the transformer's
connection from UL to triode and nothing else. Maybe was better if i was
going to rebias it or change the feedback but.....
About transistors and tubes, I have heard terrible sound from transistors
and valves, too. Also i have heard triode amplifiers with oceanic soundstage
and plenty of harmonics to play Jazz Bands or Rock without the drummer. Is
it accurate? I like tubes but i think that most of designs with
semiconductors are more accurate. Is this good at all? NO! I like a sweet
blondie girl a few hours at my bed, but i prefer my little blackhair wife
for the rest of my life.So...
I have a friend of mine with a SE triode amplifier and a small quality pair
of speakers with a very relaxable and soft, dry sound. He is dying for this
kind of sound but i hate it at all. I hate it because it softs any kind of
music. I play percussions and drums for almost 20 years and i never heard
my cymbals or my snake drum "soft". When i play drums my base set isn't soft
or dry and i haven't play on a drum set hearing my base set, sounds like
spool or my cumbals sounds like "paper in oil", yet. I like some tubes and
some transistor, too. I like to hear from triode SE 5W amplifier, when a man
plays a sax or a quitar solo, but when a full group (trio, quarteto, e.t.c)
is going to play i prefer to enjoy it from my UL or a good transistor.
Because this way make me imaging that i play with the groop. That is what is
enjoyable with a good PP , UL or semiconductor, instead of small power
triode amlifiers.
For the final, how many people could hear the BB king, John Campell or
thousand Jazz- Blues groups, imaging that they are "here" from a SE triode
5-10W amplifier? Me... anyway.
Many thanks
Fotis Visilias
Trip Rifton
use high sensitivity speakers with triodes and silver wiring, and the sound
is anything but "soft". A rim shot on any of our systems will blow your
hair back man. A lot of amps (usually high feedback types) overshoot
horribly on the leading edge of a transient, and many people take this to
mean it is more dynamic, high definition, etc. Well yeah, but they are
adding stuff to the signal that isn't supposed to be there.
Sounds like you didn't rebias your amp after switching to triode, which
could cause a problem. Keep in mind that when you switch a pentode or
ultralinear amp to triode, you are giving up half the power if you don't
change anything else. If you are using the same inefficient speakers and
listen at the same level as before, you will probably just hear more
distortion afterwards, as you go into (earlier) clipping.
- Trip
"Fotis Visilias" <sv3...@yahoo.com> wrote in message
news:a343pf$rep$1...@usenet.otenet.gr...
Choky
Then use appropriate amp for your room volume.
Don't talk about SET in conjunction with little speakers-SET need horn speaker for proper work.
So,if you don't like horns-don't even think about SE.
hehe-but I think that Snare Drum is best from 12" placed in 4 feet wide horn ;)
--
Choky
Prodanovic Aleksandar
choky*remove*@eunet.yu
YU
remove *remove* to reply!!
"Fotis Visilias" <sv3...@yahoo.com> wrote in message news:a343pf$rep$1...@usenet.otenet.gr...
Denis
> Why is this wrong. If you look at the curves for a EL34 for example in
> triode or ultralinear, the triode curves are much more linear. Ultralinear
> curves are more linear than pentode.
This is entirely wrong.
First we are trying to compare two totally different things.
The triode is a voltage amplifier, i.e., the device with low internal
resistance. This is because the triode tube possess an internal
voltage feedback. The physics is simple: the plate current of a triode
depends on plate voltage as well as on the grid one. The control grid
serves two functions, it directly controls plate current and
suppresses the influence of the plate voltage to some degree.
Depending on the grid position in respect to the cathode and plate,
and the grid winding pitch, this suppression may vary over quite broad
range. The return ratio of the internal voltage feedback in a triode
is often named as 'transparency' and is nothing but inverse gain
factor, 1/mu. The bad thing is that this factor is considerably
nonlinear, therefore the internal feedback not only establishes
finite voltage gain and output resistance, but introduces some
additional nonlinearity.
The degree of this nonlinearity depends on the particular valve type.
The output I-V curves of an ideal triode for equal increments in grid
voltage must be parallel and equally spaced in respect to each other.
The best approximation to this ideal are the direct-heated triodes
like 300B, and also such indirectly heated triodes as 6SN7. The worst
cases are the large voltage regulator triodes like 6C33, and.... the
triode-connected pentodes.
Just compare their I-V curves, and all will become clear.
The screen grid of a pentode suppresses this internal feedback
making the plate current almost insensitive to the plate voltage.
Therefore the pentode has much higher output impedance and is
considered a current amplifier. To achieve necessary current damping
one must implement an external feedback of some kind. The triode or
ultralinear connection is just a possible way to introduce this
voltage feedback locally. Because this feedback acts exactly the same
way as the internal feedback of a triode, it inroduces some additional
(voltage-dependent) nonlinearity. This nonlinearity is usually much
higher than that of a good triode like 300B. The UL connection rapidly
reduces nonlinear distortion when used up to an amount of 15 - 20%,
then THD soars and Ri continues decreasing. Another drawback is that
UL or triode connection means that the screen voltage is equal to
voltage one, whily it is beneficial to reduce the Vg2 in respect to
Vp.
There is an ultimate solution: the cathode feedback circuit such as
one used in Quad II.
This circuit applies external voltage feedback by means of small
primary section connected to cathodes. This feedback is linear and
adjustable. Also there is a freedom in choosing the screen voltage.
The results of optimal implementation of this circuit are simply above
everything. The pentode stage with CFB can exceed DHT triodes, other
triodes, triode or UL connected pentodes, etc. The distortion spectrum
is also remarkably soft.
I extensively use the CFB output stage in my designs tohether with
very wideband output transformers. Look at www.cortmi.com.ua/omak.
> More linear = more accurate. I find
> it odd that someone would prefer the sound of something inherently less
> accurate, but then lots of people like the sound of solid state devices.
>
Quite right!
Denis
Patrick Turner
Trip Rifton wrote:
> Why is this wrong. If you look at the curves for a EL34 for example in
> triode or ultralinear, the triode curves are much more linear. Ultralinear
> curves are more linear than pentode. More linear = more accurate. I find
> it odd that someone would prefer the sound of something inherently less
> accurate, but then lots of people like the sound of solid state devices.
I think what Denis was trying to say was that with a decentoutput transformer,
and decent drive amp, a pentode amp
will equal any other type, and provide more power.
I don't know if this is a universal truth.
However, if the open loop character of the amp has a
wide bandwidth, then slightly more NFB can be applied,
than in the case of triodes, and the result is supposed to be
indistinguishable to triodes, or UL.
People have only been arguing about this for 65 years,
ever since the BCD pentoad was invented for power uses.
( BCD = bean counter's delight ).
What I do know is that class AB pentode amps are not so wonderful,
in terms of open loop distortion, and at any level,
and up to the maximum power of the same tubes used as triodes,
have a lot more distortion.
In fact, with careful thought and design, we can get away without any global
NFB
with triodes.
But with pentodes, you cannot, because the output Z will
be far too high, and some NVFB someplace must be used,
as in the case of transistors and mosfets.
McIntosh is an example of a lotta NFB applied.
There is another way to get Zout low without NFB.
And its the use of positive current negative feedback.
Nobody uses it anymore, as ther is no reduction of distortion,
just a reduction of Z out, but perhaps better than nothing.
But one can use a combination of PCFB with NVFB,
and getaway with a smaller total amount of NVFB.
Class A use of pentodes is much better, as there is no crossover distortion,
which tends to occur when either of the output tubes cuts off sharply,
during AB action, which does not happen as badly with triode or UL.
On the other hand, because pentoads have so much gain,
a little VNFB goes a long way, and to equal triodes
with zero NFB, only 10 to 12 dB is needed.
( One must not suppose that triodes have inherent VNFB
included, as the last time I looked, there were no
feedback networks of resistors within the 2A3 I looked at,
and so any inherent feedback, if it is there, was heavenly sent,
totally allowable, isn't cheating, and isn't fake,
and was put there by the God Of Triodes,
in His infinite wisdom, and never to be questioned, OK. )
I have nothing against pentoads, but I must use feedback
effectively when I use them.
When the GOT said we could have pentoads,
He said well, OK, but there's a price,
and it was 2 more electrodes, a supressor, and a screen grid,
and extra labour to line it all up right, high plate resistance,
and a necessity to apply manmade networks of correction,
ie, for NFB.
More clients of mine prefer triode amps, than pentoad amps.
Patrick Turner. http://www.turneraudio.com.au
Patrick Turner
Fotis Visilias wrote:
Well said Fotis.It sounds like the SET low power amps simply run out of headroom
too easily.
But a 25 watt SE amp with some NFB can be quite excellent,
and none of this softening up of transients occurs, and in fact
the sound is never so good, but you need a really decent OPT.
People with Lowther horn speakers would not agree, of course,
and they would say that a watt will do them fine.
Patrick Turner.
Fotis Visilias
> Sounds like your friend has a speaker/amplifier mismatch. My friends and
I
> use high sensitivity speakers with triodes and silver wiring, and the
sound
> is anything but "soft". A rim shot on any of our systems will blow your
> hair back man. A lot of amps (usually high feedback types) overshoot
> horribly on the leading edge of a transient, and many people take this to
> mean it is more dynamic, high definition, etc. Well yeah, but they are
> adding stuff to the signal that isn't supposed to be there.
>
> Sounds like you didn't rebias your amp after switching to triode, which
> could cause a problem. Keep in mind that when you switch a pentode or
> ultralinear amp to triode, you are giving up half the power if you don't
> change anything else. If you are using the same inefficient speakers and
> listen at the same level as before, you will probably just hear more
> distortion afterwards, as you go into (earlier) clipping.
I never heard any kind of horns speakers, so i can't tell anything about
them. I just say that i prefer to hear my favorite (Jazz,Rock,Blues and no
classic) music from a good PP or UL pentode amplifier instead of a small
triode amplifier from my 93db Triangle speakers. Maybe there is a small
triode amplifier outside there to make me change my mind, but i haven't
heard it yet. When i changed the UL mode to triode mode at my Copland, it
sounded bad enough... The linearity was OK but i lost the punch and energy
of bass even at lower volume level. After of the modification the bass
wasn't so extended and sounded like someone taped the bass-reflex hole of my
speakers. I didn't rebias it after the modification, so it was setting to
450mv to cathode's 10R resistor check point that manufacture prefers to be
for normal UL use.
Fotis.
Patrick Turner
Fotis Visilias wrote:
> I never heard any kind of horns speakers, so i can't tell anything about
> them. I just say that i prefer to hear my favorite (Jazz,Rock,Blues and no
> classic) music from a good PP or UL pentode amplifier instead of a small
> triode amplifier from my 93db Triangle speakers. Maybe there is a small
> triode amplifier outside there to make me change my mind, but i haven't
> heard it yet. When i changed the UL mode to triode mode at my Copland, it
> sounded bad enough... The linearity was OK but i lost the punch and energy
> of bass even at lower volume level. After of the modification the bass
> wasn't so extended and sounded like someone taped the bass-reflex hole of my
> speakers. I didn't rebias it after the modification, so it was setting to
> 450mv to cathode's 10R resistor check point that manufacture prefers to be
> for normal UL use.
>
> Fotis.
It does sound like you may enjoy high levels of sound.The bias point should be
fine at 450mv across the 10 ohm resistors.
But when you change to triode, from UL, the load seen by the tubes is
not optimal for triode, and should be a bit higher.
But that would mean even less power than you are getting.
Triode amps are not for everyone, as they do have power restrictions,
and when you get them going so there is just occasional clipping,
which to me is far too loud, expect the sound to go a bit soggy.
UL and pentode amps do the same thing, but at higher power.
So why don't you just get a more powerful amp?
What output tubes are in your Copeland?
If they are EL34, why not change to KT88, or 6550,
and use the same bias current setting.
Then see how it sounds.
Patrick Turner.
Trip Rifton
I had nearly the same experience as you awhile back when I first switched my
Marantz 8b (EL34) from ultralinear to triode mode. I lost some dynamic
punch on 93 db speakers, which is what I was using at the time. But, the
sound had a MUCH more realistic tone quality. Vocals, piano and string
instuments just took on a natural ease that I found impossible to give up.
I had the same experience switching a set of highly modified Dyna Mark III's
(KT88). So, from there it was on to horns and DHT's. It's is the only way
to get both punch and ultimate delicacy that I know of. As an ex-hi-end
dealer, I've had much of the best of solid state and tube gear (including
OTL) through my shop over the years, and finally found a sound to live with.
Other peoples experience could be very different. One friend of mine swears
by electrostatic speakers, which are horribly inefficient. I wouldn't want
to hear DHT's on them, but they sound great with high powered pentodes that
smack the panels around with authority ;-)
- Trip
"Fotis Visilias" <sv3...@yahoo.com> wrote in message
news:a35te6$op1$1...@usenet.otenet.gr...
Denis
>
> What I do know is that class AB pentode amps are not so wonderful,
> in terms of open loop distortion, and at any level,
> and up to the maximum power of the same tubes used as triodes,
> have a lot more distortion.
Actually one can get 18W in class A with THD<2% or even 25W in class
AB still without any sign of crossover distortion from a pair of 6L6.
> In fact, with careful thought and design, we can get away without any global
> NFB
> with triodes.
Moreover, almost all trodes are better to be used without global NFB
if someone wants to get some good soung.
> But with pentodes, you cannot, because the output Z will
> be far too high, and some NVFB someplace must be used,
> as in the case of transistors and mosfets.
> McIntosh is an example of a lotta NFB applied.
>
A pair of KT88 will give 30W for 4 Ohm load in class A with Rout<1Ohm
in cathode feedback circuit with proper OPT.
>
> Class A use of pentodes is much better, as there is no crossover distortion,
> which tends to occur when either of the output tubes cuts off sharply,
> during AB action, which does not happen as badly with triode or UL.
There is no considerable manifestation of crossover distortion if Io
is at least 15% of Imax.
> ( One must not suppose that triodes have inherent VNFB
> included, as the last time I looked, there were no
> feedback networks of resistors within the 2A3 I looked at,
> and so any inherent feedback, if it is there, was heavenly sent,
> totally allowable, isn't cheating, and isn't fake,
> and was put there by the God Of Triodes,
> in His infinite wisdom, and never to be questioned, OK. )
>
The feedback network does not necessarily contain resistors.
The partial transfer of the output signal to the input, what is a
physical definition of the feedback takes place in a triode by means
of the internal electric fields.
Denis Afanassyev
Henry Pasternack
> The partial transfer of the output signal to the input, what is a
> physical definition of the feedback takes place in a triode by means
> of the internal electric fields.
Would you not agree that if the low plate impedance of a triode is due
to internal negative feedback, then the same can be said as well for a
diode? After all, a diode is very much like a triode with grid tied to
cathode. And if the plate resistance of a diode is due to internal
negative feedback, can you not go one step further and say that *any*
resistor has internal negative feedback? After all, in both cases,
the change in voltage with current flow (in other words, resistance)
is directly due to the effect of internal electric fields operating
inside the triode, diode, or resistor, as the case may be.
Now, if you can say a resistor operates by the principle of negative
feedback, then you can say any ordinary piece of wire does so, too.
In fact, any electrical component whatsoever is, by this definition,
a negative feedback device.
I've asked this before and never received an intelligent answer, so
I don't expect one now. I'm convinced it's possible to model a
triode as a negative feedback device. I just don't think it's very
useful. And I'm sure most of the people who insist triodes have
internal negative feedback aren't really sure why, in the sense that
they have not looked at the math behind the physical mechanisms at
work inside the tube and expressed them in a form recognizable as
a feedback equation.
If you look at how the plate curves for a triode are derived, they
come from a differential equation relating the mass, acceleration,
density, and electrical field of electrons within the tube. There
is nothing in the math that remotely resembles feedback.
I'm firmly in the no-internal-feedback-triode camp. Insisting
triodes have internal negative feedback seems like an affectation
to me.
But what do I know? I'm not a triode. I don't even play one on
TV.
-Henry
John Byrns
"Henry Pasternack" <eat....@spam.com> wrote:
>I'm convinced it's possible to model a
> triode as a negative feedback device. I just don't think it's very
> useful.
Why isn't it useful? The common SPICE model for a triode is a negative
feedback formulation, and it's fairly useful.
Henry Pasternack
news:jbyrns-3101...@216-80-74-252.d.enteract.com...
> Why isn't it useful? The common SPICE model for a triode is a negative
> feedback formulation, and it's fairly useful.
I don't use SPICE, so I wouldn't know.
-Henry
Fred Nachbaur
I would answer this "no." In a diode, there is essentially only one
parameter controlling plate current: plate voltage.
In a triode there are two mechanisms at work; plate voltage and grid
voltage. Furthermore, a change in grid voltage causes a change in plate
current, which in turn causes a change in plate voltage, which (as I
think we all agree) causes a change in plate current that tends to buck
the original change imposed by the grid.
Now whether one chooses to view this mechanism as a negative feedback
system is, as I've said before, up the the individual. You can choose to
view it that way, and perhaps gain some more understanding on how a
triode works; or you can choose the alternate approach, and settle on
"That's just the way it is."
I'm not a triode either, but for myself it helps to understand them by
considering the *mechanism* of a triode as a NFB system.
Fred N
Henry Pasternack wrote:
Chris Hornbeck
<eat....@spam.com> wrote:
>Would you not agree that if the low plate impedance of a triode is due
>to internal negative feedback,
No, actually. What am I missing here? Or is it all too subtle
for me?
>Now, if you can say a resistor operates by the principle of negative
>feedback, then you can say any ordinary piece of wire does so, too.
>In fact, any electrical component whatsoever is, by this definition,
>a negative feedback device.
Gentlemen (and -women), as much as I respect your good selves and
thoughts, THIS is The Strangest (and perhaps, ultimately, the most
interesting), Strange Thing i've heard lately.
More, please!
Chris Hornbeck, Guyville
guyville{at}aristotle{dot}net
Patrick Turner
Denis wrote:
> Patrick Turner <in...@turneraudio.com.au> wrote in message news:<3C56666E...@turneraudio.com.au>...
> >
> > What I do know is that class AB pentode amps are not so wonderful,
> > in terms of open loop distortion, and at any level,
> > and up to the maximum power of the same tubes used as triodes,
> > have a lot more distortion.
>
> Actually one can get 18W in class A with THD<2% or even 25W in class
> AB still without any sign of crossover distortion from a pair of 6L6.
And as soon as you go away from the one and onlyload value to give these splendid figures, the distortion is
not to good.
And when you go for 30 or more watts, things are worse....
807 in tetrode with 600 volts can make about 80 watts,
mainly all class B, and distn is about 13%.
>
>
> > In fact, with careful thought and design, we can get away without any global
> > NFB
> > with triodes.
>
> Moreover, almost all trodes are better to be used without global NFB
> if someone wants to get some good soung.
Not so.I get better sound with global NFB.
I have proved time after time that a bit of global feedback
is a darn good technique, providing the OPT and other factors are
wide bandwidth to start with.
> > But with pentodes, you cannot, because the output Z will
> > be far too high, and some NVFB someplace must be used,
> > as in the case of transistors and mosfets.
> > McIntosh is an example of a lotta NFB applied.
> >
>
> A pair of KT88 will give 30W for 4 Ohm load in class A with Rout<1Ohm
> in cathode feedback circuit with proper OPT.
Thre has to be a lot of cathode feedback, and if you makea decent driver, and OPT, all is well.
Mc Intosh amps use the cathode feedback in the output stage
to get Ro lower than triodes, but drive voltage is very high.
Out of the frypan, and into the fire.
> >
> > Class A use of pentodes is much better, as there is no crossover distortion,
> > which tends to occur when either of the output tubes cuts off sharply,
> > during AB action, which does not happen as badly with triode or UL.
>
> There is no considerable manifestation of crossover distortion if Io
> is at least 15% of Imax.
>
> > ( One must not suppose that triodes have inherent VNFB
> > included, as the last time I looked, there were no
> > feedback networks of resistors within the 2A3 I looked at,
> > and so any inherent feedback, if it is there, was heavenly sent,
> > totally allowable, isn't cheating, and isn't fake,
> > and was put there by the God Of Triodes,
> > in His infinite wisdom, and never to be questioned, OK. )
> >
>
> The feedback network does not necessarily contain resistors.
> The partial transfer of the output signal to the input, what is a
> physical definition of the feedback takes place in a triode by means
> of the internal electric fields.
Fine by me. It isn't a man made phenomena,it is GOT made, as He made the laws of physics, nobody else.
No resistors are used for feedback,
hence zero negative feedback, by mankind, that is.
Triodes are natural, pentodes are contrived.
It ain't what you got, but the way you use it,
that makes good sound,
and there are more ways than one.
Patrick Turner.
>
>
>
> Denis Afanassyev
>
> www.cortmi.com.ua/omak
Patrick Turner
Henry Pasternack wrote:
Henry, its all about envy, purity, shame, pride, and inferiority
complexes.People are envious about what they havn't got.
99% of the world's amplifiers must use NFB
to get anywhere with sound signals.
But 1% get really excellent sound without any.
And they are the tribe of triode worshippers.
They have shirtloads of disposable resources,
and spend up on the 1% in a big way sometimes,
and they don't give a toss for efficiency, which, when it is about
worst that you could have, gives the best triode sound.
This state of affairs is obnoxious to the sensibilities
of the constant current tribe, as although they dominate
99% of world thinking, 1% refuses to be converted to
the one true faith.
99% of the world's amps use constant current devices for their
amps, in the form of pentodes, transistors, mosfets,
( and a few are not sure so they partially cheat, in a way called
ultralinear.)
Anyhow, a bit of NFB gets over these problems of
the CCS tribe, but they still have this envy over their lack
of natural purity, that the 1% have.
So in order to make themselves feel happy if not smug,
they invent all this stuff about
inherent triode NFB, and deny the existance of the God Of Triodes.
> But what do I know? I'm not a triode. I don't even play one on
> TV.
>
> -Henry
But if you drape yourself in clear plastic, and stand still with a
lightedcandle in your hand, then you may get confused with a transmitting
triode.
Not for long, as when they connect up the voltage, you'll
move out of the way.....
Patrick Turner.
Patrick Turner
Chris Hornbeck wrote:
The magazine, Wireless Engineer, in April 1953,carried articles about
feedback in triodes with a lotta math,
and explanation, that is difficult to understand.
It isn't as if the pentode and transistor tribe thought it up last week.
I see a triode as a resistor with value Ra ohms in series with
a low impedance voltage generator with three terminals,
the output to Ra, the common ground, and the input, and
the generator has makes an output voltage = mu x input voltage.
Very simple, and free of any feedback equivalent network.
But I guess some dude with smarty pants education from some university
someplace can make the simple model all so much more complex.
John L Stewart sent me the articles.
Maybe he understands them better than I do.
May he comment, and explain......
Anyone like to quote the RDH4 on the NFB in triodes????
Patrick Turner.
Henry Pasternack
operation of a vacuum tube.
Take as a working hypothesis the claim that the low plate resistance
of a triode is due to an internal negative feedback mechanism.
If you wish to determine the plate resistance of the triode, you can
hold the grid at a fixed potential and observe the change in plate
current with plate plate voltage. This is a standard measurement
technique, one that is taught in engineering schools. It depends
only on externally observable properties of the tube and is independent
of the internal mechanism responsible for the observed behavior.
I will claim that if you connect the grid to the cathode, it is as
if the grid is not there at all. This follows because the electric
field due to the grid-cathode potential is zero, and therefore the
grid has no influence on the space charge density, or the rate at
which electrons are ejected from the cathode. Under this condition,
the plate current is purely a function of the plate voltage.
In other words, the triode with grid shorted to cathode is equivalent
to a diode. This is evident from the electrostatic model used to
derive the plate current equation. It's evident from the triode
plate current equation. And it's evident from comparing the triode
Vg = 0 plate curves with the plate curve of a diode.
Let's call this special case the "diode connected" triode. There's
nothing special about the diode connected triode. In particular,
there is nothing that changes our working hypothesis.
Since the behavior of the diode connected triode is identical to that
of a true diode, and since the internal mechanisms are the same, it
logically follows that a diode's plate resistance is also due to the
effect of negative feedback.
I understand this may be a controversial conclusion. Let's for the
moment set aside our skepticism and agree that the the diode does have
internal negative feedback. Connect the cathode of the diode to ground
and the plate to a B+ supply through a resistor. Increase the value of
the B+ voltage. The plate voltage will increase. This will increase
the electric field at the cathode. Through a series of electrostatic
interactions, the plate current will increase. This will tend to draw
down the plate voltage, partially canceling the original increase.
I think you will agree this is exactly the same behavior we would see
with a triode, the behavior that we are calling negative feedback.
Now, replace the diode with a resistor. You will see that the very
same steps I have just described for the diode apply as well in this
case. Internally, the mechanism giving rise to resistance is different,
but the chain of cause and effect that we've used to justify the claim
concerning negative feedback (an increase in terminal voltage causes a
change in the internal electrostatic field which raises the current,
dropping the terminal voltage in opposition to the original increase)
remains the same.
Therefore, a resistor also has internal negative feedback, according
to our definition. Since a wire is just a very low-value resistor, we
can extend the same definition to it as well.
If you tie the plate of the tube to fixed potential and put the load
resistor in the cathode circuit, you will have exactly the same sort
of situation as before. An increase in plate voltage causes an increase
in cathode current which raises the cathode voltage, opposing the change
in current. This is familiar as negative current feedback. But we don't
use this as an argument to claim the tube has internal voltage negative
feedback. So why do we make this argument when the resistor is connected
to the plate?
Actually, I'm very conscious of the rationale for the negative feedback
point of view. The plate current is determined by the rate at which
electrons are emitted from the cathode. This, in turn, is a function
of the space charge density surrounding the cathode. If the cathode is
hot enough, the space charge density settles at a value that results in
exactly zero electric field at the cathode. A change in either plate or
grid potential will cause the space charge density to vary in order to
re-establish the zero-field condition at the cathode. And this, in turn,
changes the number of electrons available to be collected by the plate,
modulating the plate current. Since both the plate and the grid voltages
can influence the charge density and hence the plate current, we can view
them as two input terminals for the tube. But the plate is also the output
terminal, so it is as if the output terminal is connected to the input
terminal, i.e., voltage negative feedback. It's an attractive argument,
but one that I think, if carried to its logical conclusion, leads to some
contradictory and counter-intuitive results.
Cheers.
-Henry
"Fred Nachbaur" <fr...@netidea.com> wrote in message
news:3C57731C...@netidea.com...
Chris Hornbeck
<eat....@spam.com> wrote:
>The following discussion presumes an understanding of the physical
>operation of a vacuum tube.
I've clipped out most of an extremely careful, abstruse, thoughtful,
and still (for my poor abilities) puzzling chain of thought. Leaving:
>Since the behavior of the diode connected triode is identical to that
>of a true diode, and since the internal mechanisms are the same, it
>logically follows that a diode's plate resistance is also due to the
>effect of negative feedback.
Ah!, here i have trouble following the argument. Could you restate
for an untutored layman the "given" part about the internal
feedback mechanism inherent in a triode? If i've misrepresented
your argument, also please correct me.
>Internally, the mechanism giving rise to resistance is different,
>but the chain of cause and effect that we've used to justify the claim
>concerning negative feedback (an increase in terminal voltage causes a
>change in the internal electrostatic field which raises the current,
>dropping the terminal voltage in opposition to the original increase)
>remains the same.
>
>Therefore, a resistor also has internal negative feedback, according
>to our definition. Since a wire is just a very low-value resistor, we
>can extend the same definition to it as well.
OK. Let's say that feedback is conceptually a few spinning spoons
on a fossile-fuel engine, the faster the spoons spin, the less air
is allowed into the the engine, IOW, a governor. Is *this* definition
of feedback in the same ballpark (or even the same continent?) as
in this discussion?
Thanks for your always topical and thought-provocing posts.
Fred Nachbaur
right, you're wrong" flame wars.
Just for fun, let's look at it from the other direction. Let's take a
tetrode, with a fixed screen voltage. This device will exhibit a
relatively high plate resistance, as measured by your described
methodology of keeping grid voltages constant and measuring the plate
current vs. plate voltage "as seen by" an external device. The plate
current will be largely independant of plate voltage, because the
electric field gradient is relatively constant between cathode and
screen. It's as if the plate's potential is masked by the presence of
the screen, and the cathode doesn't "see" variations in plate voltage to
a significant degree. An increase in grid voltage causes an increase in
plate current (and screen current, albeit to a much lesser degree,
because of the screen's limited ability to capture electrons, most of
them falling through toward the plate).
We now vary the voltage on the control grid, and establish that the
device exhibits a certain transconductance (delta Vg/ delta Ip). Again,
this effect is relatively independant of plate voltage due to the
presence of the screen, which imposes a relatively constant voltage
gradient between cathode and screen.
Now let's return to the triode. The cathode will no longer be "looking
into" a constant voltage gradient. An increase in grid voltage causes an
increase in plate current, as before, but this causes a drop in plate
voltage because of the plate load resistance. The voltage gradient
between cathode and plate is therefore decreased, resulting in *less*
plate current, and opposing the original tendency to increasing plate
current.
So, is it the plate load which is "responsible" for the operation of the
circuit "as if" it were a negative feedback system? Or is it the nature
of the triode itself, with its variable cathode-to-anode voltage
gradient? Or does the circuit *as a circuit* behave as if negative
feedback were applied?
Another aspect: if you look at the plate curves of a triode, the first
glance would suggest that the triode is a highly non-linear device, with
all those curved Ip vs Vp lines for various grid voltages. It would seem
that pentodes are much more linear, with all those nice, almost
straight, almost horizontal lines. However, what matters in an amplfier
circuit is linearity *along the load line.* In other words, what happens
when we put this apparently highly non-linear device into a circuit
containing a plate resistor? It doesn't take long playing with rulers
and graph paper to ascertain that the triode is, indeed, more "linear"
along the load line than a comparable tetrode.
We could just as easily argue about whether adding an unbypassed cathode
resistor adds feedback to a system, or not. I say it does, because it's
a convenient, predictable, and *useful* way of looking at it. Similarly,
when it comes to the operation of plate circuits, I still say "If it
quacks like a duck, maybe it's a platypus but I'll call it a duck."
Cheers to you too,
Fred N
Henry Pasternack wrote:
Denis
Thanks a lot for very good post!
This is an excellent plain English explanation of the internal
feedback mechanism.
Indeed, let me add something on triode vs. tetrode linearity.
This internal triode feedback is highly nonlinear. When a triode
operates at typical loadind conditions, i.e., the plate load is 3 - 4
times the internal resistance, more than 50% of total harmonic
distortion originates from this nonlinear feedback. If the triode sees
an infinite load like the current source, 100% of the nonlinearity is
attributed to the internal feedback, and the depth of this feedback in
this case is infinity. This phenomenon could be the explanation of the
fact that the SE triode amplifiers are extremely choosy as to speaker
impedance. The best sound could be achieved only if the load impedance
varies over a quite narrow range around some definite value.
By other words, the internal feedback in the triode does two things:
It reduces the basic distortions originating from the dependence of
plate current on Vg^3/2, and at the same time it adds the distortions
originating from the nonlinearity of it's return ratio. These two
contributions vary in respect to each other depending on the plate
load, and there is certain value of load resistance at which one
obtains most favourable results. The internal feedback at this point
also will be moderate. If the triode is used as the resistance-loaded
gain stage it is possible to operate it at very this optimal point.
Indeed, if it is the power stage, the triode will see the complex
impedance of loudspeaker together with the output transformer, and
such optimal loading conditions may exist only at several frequencies.
As to the harmonics spectrum, the nonlinearity of this internal
feedback is of even order, therefore it adds mostly the even harmonics
increasind their percentage in THD. Indeed, the distortion spectrum of
a triode always contains a series of both even and odd harmonics with
their amplitudes gradually decreasing with order.
Certainly, it is better than the distortions of a commercial
solid-state class AB amplifier and also than that of any power-abused
PP tube design.
Now, what the pentodes can do if someone wants an ultimate amplifier.
The screen grid suppresses to large degeree the above internal
feedback making the pentode the current source, defined by a
transconductance, gm. The cathode current is roughly proportional to
Vg^3/2 the same way as in the triode. Note that there are certain
deviations from this fractional power law, especially at extremely
negative biases. Here we'll stress that the control grid nonlinearity
is much the same in both triode and pentode, and the distortions
introduced by this physical mechanism are mostly defined by the ratio
between the AC and idle components of cathode current, and are much
the same in tubes of different types.
Due to the fractional power law, these distortions always contain both
even and odd harmonics in gradually decaying spectrum.
There is another significant source of nonlinearity in the pentodes,
the screen grid. It induces the distortions mostly due to the
redistribution of the total cathode current between the plate and
screen grid.
It is important that these distortions are mostly of odd order, so
this current redistribution generates mostly the odd harmonics,
especially the 3rd one.
And what is much more important these odd harmonics are out of phase
in respect to those contained in the cathode current due to the
nonlinear action of control grid! This is evident from the shape of
screen current I-V curves.
As the screen current in respect to the plate one depends very much on
the screen voltage it is possible to alter the amount of these
nonlinear products in the plate current simply by varying the screen
voltage.
This opens the possibility to totally eliminate the third harmonic
from the distortion spectrum of the pentode stage for quite broad
portion of it's dynamic range. The odd harmonics of higher order will
be suppressed too. Actuallly, the odd harmonics will appear only
approaching the clipping point, and in soft manner. I've perfected
this mode of operation for such fantastic tube as EL509 in OMAK Jewel
Three monoblock. What is more interesting, such cancellation does not
obviously require class A operation of PP stage. When Jewel Three
comes to class AB at, say, 2 Ohms load, the distortion behaviour is
much the same. There is no sign of often mentioned crossover
distortions, the harmonics decay in the noise when the signal level
decreases. The output impedance as referred to the secondary does not
measurably depend on signal level, and is almost constant over the
whole bandwidth.
Another fine physics come from the small UL modulation at the screen
grid. Because the UL configuration introduces the local feedback much
the same way as in triodes, it introduces that damned feedback-induced
nonlinearity. Indeed, the screen current is added to the plate load
only partially, according to the UL tap percentage. Therefore, when
introduced in quite small amounts, usually below 20%, the UL
modulation effectively softens distortion spectrum, especially
approaching the clipping point. The optimal percentage reduced further
with reduced screen voltage and at certain point comes close to the
optimal percentage of the CFB winding. Here the positive effects of
odd harmonic cancellation, UL modulation and linear cathode feedback
come together.
Let me note here that nothing equal is possible with triodes.
Most pentode PP class AB amplifiers are designed bearing in ming just
the highest output power, so they use always too high screen
voltages, low idle current and lots of global NFB. The audiophile
products are not the exceptions. From other hand, the output power of
a CFB stage optimised the above way is still higher than that of the
same pentodes in triode connection.
Finally, the great problem with such power stage may be the
driver/phase splitter. The distortions of the power stage may be
reduced to such extent that those produced by driver may appear to be
higher.
Regards
Denis.
Fred Nachbaur <fr...@netidea.com> wrote in message news:<3C58D9A8...@netidea.com>...
Patrick Turner
Fred Nachbaur wrote:
> Good post, Henry. It's nice to see different viewpoints without the "I'm
> right, you're wrong" flame wars.
>
> Just for fun, let's look at it from the other direction. Let's take a
> tetrode, with a fixed screen voltage. This device will exhibit a
> relatively high plate resistance, as measured by your described
> methodology of keeping grid voltages constant and measuring the plate
> current vs. plate voltage "as seen by" an external device. The plate
> current will be largely independant of plate voltage, because the
> electric field gradient is relatively constant between cathode and
> screen. It's as if the plate's potential is masked by the presence of
> the screen, and the cathode doesn't "see" variations in plate voltage to
> a significant degree.
Yes, you have got it.
What is happening is that electrons at the cathode are experiencing the
tug of the electric field of the screen wires, and these screen off the effect of
the plate
electric field.
So, wild with excitement, these electrons stream off on in search of their
master, positivity, and get a fair rate of speed up.
Some hit the screen wires, and are absorbed, and this forms the screen
current, about 10 to 30% of the plate current.
Most electrons miss the screen, and overshoot it,
and its too late to slow down much, and go for any darn thing which is at all
positive,
even a plate which is at the bottom of a signal voltage wave, or the top,
it makes little difference.
So they attempt to get into the metal of the plate, but some bounce off,
and head back to the screen, to have another attempt at landing.
They have slowed a bit,
and they encounter this supressor grid,other 3rd grid, of less spiral turns,
which is kept at cathode potential, between the plate and screen.
This is a similar charge to the electrons, and slows them down
even more, and they mostly return to the plate.
So regardless of the plate voltage, the stream of electrons
which miss the screen, which is at a fixed potential, is about constant,
if the control grid voltage and cathode voltage is constant.
> An increase in grid voltage causes an increase in
> plate current (and screen current, albeit to a much lesser degree,
> because of the screen's limited ability to capture electrons, most of
> them falling through toward the plate).
>
> We now vary the voltage on the control grid, and establish that the
> device exhibits a certain transconductance (delta Vg/ delta Ip). Again,
> this effect is relatively independant of plate voltage due to the
> presence of the screen, which imposes a relatively constant voltage
> gradient between cathode and screen.
>
> Now let's return to the triode. The cathode will no longer be "looking
> into" a constant voltage gradient. An increase in grid voltage causes an
> increase in plate current, as before, but this causes a drop in plate
> voltage because of the plate load resistance. The voltage gradient
> between cathode and plate is therefore decreased, resulting in *less*
> plate current, and opposing the original tendency to increasing plate
> current.
So you have to compensate by using more grid voltagethan in a pentode, to get the
same current to flow.
But you have not used any resistors in any network, have you.
> So, is it the plate load which is "responsible" for the operation of the
> circuit "as if" it were a negative feedback system? Or is it the nature
> of the triode itself, with its variable cathode-to-anode voltage
> gradient? Or does the circuit *as a circuit* behave as if negative
> feedback were applied?
>
> Another aspect: if you look at the plate curves of a triode, the first
> glance would suggest that the triode is a highly non-linear device, with
> all those curved Ip vs Vp lines for various grid voltages. It would seem
> that pentodes are much more linear, with all those nice, almost
> straight, almost horizontal lines. However, what matters in an amplfier
> circuit is linearity *along the load line.* In other words, what happens
> when we put this apparently highly non-linear device into a circuit
> containing a plate resistor? It doesn't take long playing with rulers
> and graph paper to ascertain that the triode is, indeed, more "linear"
> along the load line than a comparable tetrode.
>
> We could just as easily argue about whether adding an unbypassed cathode
> resistor adds feedback to a system, or not. I say it does, because it's
> a convenient, predictable, and *useful* way of looking at it. Similarly,
> when it comes to the operation of plate circuits, I still say "If it
> quacks like a duck, maybe it's a platypus but I'll call it a duck."
For practical purposes, there is only plate resistance to worry aboutin triodes,
and any internal feedback equivalent circuit isn't necessary
to consider. We just got what wev'e got, with a triode.
But with a pentode, we have a range of options, and external loops of
feedback can be arranged.
If you feed platypus bread crumbs, they'll probably ignore them,
as they are very shy creatures, with an entirely different
way of life to a duck, which will go for bread, to our
eternal amusement, and wonderment.
Patrick Turner.
Patrick Turner
Denis wrote:
In every single pentode amp that I have ever built or repaired,the load tolerance of pentodes
is so peaked around just one load,
that their useful power ability is somewhat overstated.
All sorts of different combinations of harmonic products are produced
in a pentode at different loads.
Speakers vary widely in their impedance, and most triodes,
without any connected loops of feedback, cope much better,
and all frequencies, with less distortion.
And the simple way to get more triode power to equal that of pentodes
is to use a bigger tube, such as a KT88/6550 in lieu of the 6L6/EL34/KT66
that you presently use,
or use a quad of tubes instead of a pair.
It is not unusual for a pentode amp to have a more complex set of
distortion components than a triode.
When we connect a loop of feedback to get the pentode
circuit to behave like a triode circuit, in terms of output impedance
of the triodes with no feedback loops, the THD of the pentodes
is still generally worse.
This is because that amount of applied loop feedback would only be about 8 to 10
dB with most pentodes.
Then you have what you feed back intermodulate with the input signal,
and the mix of products gets worse, rather than better.
In an SE pentode amp, the 2H is reduced a bit but the 3H may increase by this mechanism,
and much more 9H is present.
And SE pentodes have appalling distortion at their recomended loads and powers,
many are 12%, very high number of components, not much to recommend them.
UL over comes a lot of these shortcomings.
Cathode feedback coils do too, but the distortion in most normal pentodes,
like the EL34, is just reduced in %, and still has high order THD in it.
I find that when screen feedback is used as well, the outcome
is better than either triode, or plain UL, let alone pure pentode.
I havn't used EL509, but I bet it is not entirely unique in that
there is less distortion than if you connected it as a triode.
Pentode gain is about equal to RL x Gm, so when two are in class
AB , and one cuts off, sharply, as pentodes tend to,
there is then a kink in the transfer curve going from A to AB,
as the one tube which is conducting is using only half the OPT winding, and its load has
suddenly changed from half RLa-a, to a quarter.
Triodes are much less distortion productive during this process in class AB,
as their cut off is much more gradual, so less odd order upper harmonics
are produced compared to pentodes.
And since Triode Ra is about a tenth of a pentode, in say an EL34,
then the OPT is driven by a lower source impedance,
and iron distortion is lower.
Class A pentodes are livable withable, but you need plenty feedback,
and that means good OPTs.
It is rare to ever see distortion in a driver amp for pentodes exceed the
output stage THD, unless the output stage feedback requires a high drive voltage
due to that feedback as in the case of the McIntosh.
So I never use more than 12% of CFB from the OPT to the cathodes.
I am rebuilding an amp with 4 x 6550 in each channel, set up like this,
and it also has 25% of the primary voltage fed to the screens, and
so it is a "superlinear" connection.
It requires 70 vrms drive, and any more drive, due to more CFB would have the
benefit of the lower output stage THD negated by the rise in driver THD.
I found this arrangement of tappings and windings to give the lowest
amount of high order THD, and like all good triode and UL amps,
produces 0.2% of 3H THD at near full power, with a TOTAL
amount of UL, CFB and global NFB of about 17 dB, at 50 watts,
mainly all class A.
The other THD products are down 20 dB or more.
I am using a pair of EL84 tubes strapped as triodes in a LTP
to drive the quad of output tubes, which each have 100 kOhm
grid bias resistors, to keep reverse grid current voltage low
in the 6550 tubes I have, which are now about 5 years old,
but only 500 hours, max.
These were bought as NOS JAN GE6550, made about 1980??.
Of the total of 8 in 2 channels,
2 develop 3 volts reverse voltage across a 250 kOhm
bias resistor, hence the need to keep bias resistors to
the MINIMUM practical value, lest a tube thermal away.
This means the DC load resistor for the driver tubes has to be a third or less
of the grid resistor, and hence to get a wide voltage swing, at low THD,
I have used EL84 in triode, with 20 watt 18 kOhm DC supply loads,
with 16 ma of plate current.
The added benefit is the low output Z of the driver.
If any capacitance effects might tend to slow down what we hear,
it won't be by very much, because the Cin of the output stage is so low
due to the local feedback used there from a well interleaved transformer.
At normal listening levels, THD is an academic quantity,
less than 0.02%, and 20 times less than an equivalent powered SE amp,
even a triode one, with the same loop feedback.
Some folks still prefer an SET, and in some cicumstances,
they are all that is needed.
Patrick Turner.
Choky
adequate circular cell ;)
That's nature;
In most cases ,even tetrodes and pentodes are happier strapped in triode mode.
So,who I am to argue with Laws of Nature? ;)
--
Choky
Prodanovic Aleksandar
choky*remove*@eunet.yu
YU
remove *remove* to reply!!
===================
Fred Nachbaur
be useful just for fun. :-/
Fred N
Choky
Choky never talk about rules ;) .........
--
Choky
Prodanovic Aleksandar
choky*remove*@eunet.yu
YU
remove *remove* to reply!!
===================
"Fred Nachbaur" <fnac...@netscape.net> wrote in message news:3C5C58F2...@netscape.net...
Denis
>
> In every single pentode amp that I have ever built or repaired,the load tolerance of pentodes
> is so peaked around just one load,
> that their useful power ability is somewhat overstated.
This is not the whole truth. If a pentode without any kind of external
NFB is driven by the same AC voltage at control grid, the THD will
have a minimum, and power - a maximum at some definite load. Indeed,
the properly feedbacked pentode stage is far more load - tolerable
than any triode, especially when driving the reactive loads.
> All sorts of different combinations of harmonic products are produced
> in a pentode at different loads.
The pentodes can be put to such a mode of operation, when their
distortion spectrum is much softer than that of any triode. They are
most flexible amplification devices ever made.
Sorry, but these unique features of the pentodes are seldom properly
exploited.
> Speakers vary widely in their impedance, and most triodes,
> without any connected loops of feedback, cope much better,
> and all frequencies, with less distortion.
Triodes already contain feedback, beleive or not. Therefore the only
correct comparison could be done fo a triode without external feedback
and a pentode with it.
>
> It is not unusual for a pentode amp to have a more complex set of
> distortion components than a triode.
This is simply an example of wrongly designed pentode amplifier (in
respect of lowering the distortion). Too much pentode amps are
designed for maximal power, not the sound quality. The triodes also
can be uase in such a way, and they will sound in this case equally
dirty.
In contrary, I have an amplifier with EL509s in output stage, which
generates only 2nd hadmonic for most of it's dynamic range, and
behaves much the same from open output to 2 Ohms load. I'm sure this
is also possible with other power pentodes, especially 6550.
> And SE pentodes have appalling distortion at their recomended loads and powers,
> many are 12%, very high number of components, not much to recommend them.
> UL over comes a lot of these shortcomings.
Proper operating point optimisation and application of cathode
feedback makes the pentode an ideal tube for a SE stage too.
Everything I said about the harmonics spectrum is applicable.
Moreover, the meschanism of 3rd harmonic suppression was described as
early as in 1950s.
>
> I havn't used EL509, but I bet it is not entirely unique in that
> there is less distortion than if you connected it as a triode.
Any pentode can work much better in a CFB circuit with optimal CFB
percentage and screen grid voltage against the same pentode being
triode connected. The EL509 is an extreme example due to it's huge
emission current and low g2 gain factor.
> Pentode gain is about equal to RL x Gm, so when two are in class
> AB , and one cuts off, sharply, as pentodes tend to,
> there is then a kink in the transfer curve going from A to AB,
> as the one tube which is conducting is using only half the OPT winding, and its load has
> suddenly changed from half RLa-a, to a quarter.
> Triodes are much less distortion productive during this process in class AB,
> as their cut off is much more gradual, so less odd order upper harmonics
> are produced compared to pentodes.
Absolutely wrong! The pentodes cut off much softer, and also hafe a
softer clipping. Moreover, the PP pair of triodes can exibit the
distortion of special king, due to the presence of a mutual leakage
inductance between the halves of the OPT primary. This is also the old
- known phenomenon.
> And since Triode Ra is about a tenth of a pentode, in say an EL34,
> then the OPT is driven by a lower source impedance,
> and iron distortion is lower.
>
The iron distortion is the same if the output resistance of a pentode
stage is made equal to that of a triode one, by means of the NFB.
> Class A pentodes are livable withable, but you need plenty feedback,
> and that means good OPTs.
>
The only way to get an amplifier with lowest NFB for given distortion
is to use the pentodes.
> It is rare to ever see distortion in a driver amp for pentodes exceed the
> output stage THD, unless the output stage feedback requires a high drive voltage
The full power THD of a good CFB pentode stage can be well below 1%,
and the required drive voltage will vary within 50 - 80 Vpeak, i.e.,
comparable with that for a 300B stage. The optimal CFB percentage is
within 12 - 15 %. The more is simply unnecessary.
The McIntoch stage seems to be far from optimal.
This is a description of a good amplifier, which partially proves my
viewpoint.
> Some folks still prefer an SET, and in some cicumstances,
> they are all that is needed.
>
The tastes are different.
Denis Afanassyev.
Patrick Turner
Denis wrote:
> Patrick Turner <in...@turneraudio.com.au> wrote in message news:<3C5C1491...@turneraudio.com.au>...
>
> >
> > In every single pentode amp that I have ever built or repaired,the load tolerance of pentodes
> > is so peaked around just one load,
> > that their useful power ability is somewhat overstated.
>
> This is not the whole truth. If a pentode without any kind of external
> NFB is driven by the same AC voltage at control grid, the THD will
> have a minimum, and power - a maximum at some definite load. Indeed,
> the properly feedbacked pentode stage is far more load - tolerable
> than any triode, especially when driving the reactive loads.
One cannot escape the open loop character of any device usedas an amplifier. Miracles are only occasionally
available in audio,
and the pentode power ability is more peaked and limited than
the same tube wired up as a triode.
Feedback will not substantially change the curve one can
plot for maximum power output vs 3% distortion.
However, pentodes do have twice the power abilty
than triodes, and if we run the pentode at the same
power output as the triode, and in class A,
then not to many folks will hear any difference,
even though the triode will still have less distortion, open loop.
If we have enough NFB in the pentode circuit to make
the Zout equal to the triodes open loop character,
I would still bet the triode produces less THD.
With any tube, the plate resistance with NFB, Ra' is
Ra divided by [ ( Mu + 1 ) x B ].
We can appoximate this for pentodes,
Ra' = Ra divided by [ Gm x B ]
So if we have 12.5% of the total primary winding as a cathode
winding, and we are using an EL34, in pentode,
then Ra' = 12 kOhms divided by [ (132 + 1 ) x 0.125 ]
and this is 720 ohms, or nearly half that of the triode without
any NFB loops.
If the load was about 2.5 kOhms, then the gain
with no NFB will be 21.3 times, and with the CFB,
the gain would be 7.6, so the gain reduction would be
2.78 times, or 9dB, ie, we have applied 9 dB of NFB.
Probably only 6 dB of NFB is required to
lower most pentode Ra to that of the same tube in triode.
But low mu pentodes would react differently.
The higher the mu, and Gm, the better the results.
EL84 react very favourably to CFB application.
> > All sorts of different combinations of harmonic products are produced
> > in a pentode at different loads.
>
> The pentodes can be put to such a mode of operation, when their
> distortion spectrum is much softer than that of any triode. They are
> most flexible amplification devices ever made.
"Softer"?, what do you mean, in engineering terms?
>
>
> Sorry, but these unique features of the pentodes are seldom properly
> exploited.
Unique?Gee, pentodes have been used now routinely for about at least
65 years. Some of the exploitation was proper, some was improper,
and many old radios are a poor example of pentode power,
as sometimes not a stitch of NFB was used.
But that improved frequency response, and
saved the use of an extra tube,
and nodody worried much about distortion.
>
>
> > Speakers vary widely in their impedance, and most triodes,
> > without any connected loops of feedback, cope much better,
> > and all frequencies, with less distortion.
>
> Triodes already contain feedback, beleive or not.
But not feedback which has be contrived by humans.God put whatever feedback is in triodes there.
Perhaps you can't grasp the idea of God, but there must be one,
as the universe just didn't put itself there, with triodes,
and laws of electron flow in vacuums, all by itself.
Pardon me if I tell you I hav't the fogiest idea of who or what
God is, but I sure can see where He has been, and what
He has been up to for awhile.
Therefore the only
> correct comparison could be done fo a triode without external feedback
> and a pentode with it.
Depends how much NFB is used.
> >
> > It is not unusual for a pentode amp to have a more complex set of
> > distortion components than a triode.
>
> This is simply an example of wrongly designed pentode amplifier (in
> respect of lowering the distortion). Too much pentode amps are
> designed for maximal power, not the sound quality. The triodes also
> can be uase in such a way, and they will sound in this case equally
> dirty.
Yes there is ann optimal operational point for both devices,and it sure isn't at the maximum power point,
as there is no allowance for power to increase without
too much THD if the load value goes below nominal,
which is very usual with most modern speakers.
>
>
> In contrary, I have an amplifier with EL509s in output stage, which
> generates only 2nd hadmonic for most of it's dynamic range, and
> behaves much the same from open output to 2 Ohms load. I'm sure this
> is also possible with other power pentodes, especially 6550.
No, it isn't possible, and I don't see how it could be with theEL 509, unless it is in SE mode, when the 2H
THD will
sometimes be greater than the 3H, depending on load.
With all PP, nearly all the 2H and even numbered harmonics dissapear, and 3H is
usually the main THD component.
The only 2H that can exist in a PP amp is that produced by the
driver stage, but generally it can be kept very low with a
careful design.
Class AB pentodes produce 3H and almost no 2H,
in the output stage. They also produce a lot of other higher odd order
THD components.
>
>
> > And SE pentodes have appalling distortion at their recomended loads and powers,
> > many are 12%, very high number of components, not much to recommend them.
> > UL over comes a lot of these shortcomings.
>
> Proper operating point optimisation and application of cathode
> feedback makes the pentode an ideal tube for a SE stage too.
> Everything I said about the harmonics spectrum is applicable.
> Moreover, the meschanism of 3rd harmonic suppression was described as
> early as in 1950s.
Basic feedback theory.
> >
> > I havn't used EL509, but I bet it is not entirely unique in that
> > there is less distortion than if you connected it as a triode.
>
> Any pentode can work much better in a CFB circuit with optimal CFB
> percentage and screen grid voltage against the same pentode being
> triode connected. The EL509 is an extreme example due to it's huge
> emission current and low g2 gain factor.
One could use paralled tubes of other types, and get good results.I doubt I will ever use the EL509
commercially, as
the tube sockets are novals, and the same as a 7868, and so
all other tubes which are plain octal based cannot be used in an
amp with EL509s, unless you replace the sockets.
I try and configure my amps to use a large range of octal based
sockets, so that tubes from 6V6 to KT88 could be used,
with adjustments to plate and bias voltages.
I don't know what the tube supply situation will be in 20 years.
> > Pentode gain is about equal to RL x Gm, so when two are in class
> > AB , and one cuts off, sharply, as pentodes tend to,
> > there is then a kink in the transfer curve going from A to AB,
> > as the one tube which is conducting is using only half the OPT winding, and its load has
> > suddenly changed from half RLa-a, to a quarter.
> > Triodes are much less distortion productive during this process in class AB,
> > as their cut off is much more gradual, so less odd order upper harmonics
> > are produced compared to pentodes.
>
> Absolutely wrong! The pentodes cut off much softer, and also hafe a
> softer clipping.
OK, you cannot be convinced, try a good read up in the RDH4.
> Moreover, the PP pair of triodes can exibit the
> distortion of special king, due to the presence of a mutual leakage
> inductance between the halves of the OPT primary. This is also the old
> - known phenomenon.
No more so than pentode usage.
>
>
> > And since Triode Ra is about a tenth of a pentode, in say an EL34,
> > then the OPT is driven by a lower source impedance,
> > and iron distortion is lower.
> >
>
> The iron distortion is the same if the output resistance of a pentode
> stage is made equal to that of a triode one, by means of the NFB.
But triodes do it without NFB, they are ahead of pentodes,without any feedback.
But I do agree, NFB, even global loop feedback makes
it possible to control a pentode's wild behaviour.
>
>
> > Class A pentodes are livable withable, but you need plenty feedback,
> > and that means good OPTs.
> >
>
> The only way to get an amplifier with lowest NFB for given distortion
> is to use the pentodes.
Again, go study your books.
> > It is rare to ever see distortion in a driver amp for pentodes exceed the
> > output stage THD, unless the output stage feedback requires a high drive voltage
>
> The full power THD of a good CFB pentode stage can be well below 1%,
> and the required drive voltage will vary within 50 - 80 Vpeak, i.e.,
> comparable with that for a 300B stage. The optimal CFB percentage is
> within 12 - 15 %. The more is simply unnecessary.
I agree that more than 12.5% CFB is a waste of time, butmost pentode class AB stages produce 5% of THD in
the real world, and
if we used only 6 dB of NFB, it would drop to 3%, with
a complex odd order mix still there.
Triodes are better, with no NFB, and the same amount of AB proportion.
I doubt it matters much.
Most of what we listen to done with the first 10% of the power,
regardless of what mode the tube is in, and it is class A anyway.
> The McIntoch stage seems to be far from optimal.
Their driver stage distortion must be similar to the output stage,but it is all subject to global loop NFB,
and the outcome is welcomed
by many listeners.
The transformer design allows for lots of NFB to be used.
It is the fundemental saviour of the McIntosh way of building an amp.
But I am using the tubes as Ultralinear, with CFB.The screens are not being run at the same signal voltage
as the
cathodes, which would put them in pentode mode,
and I tried this, but odd order distortion products
were worse.
I find plain UL with loop feedback give the same excellent results,
and the only thing to keep in mind is the OPT quality,
and driver amp speed. So I prefer to use medium mu triodes
for the driver stage, but sometimes I use 12AT7 for the input tube.
The use of CFB in output stages, does not guarentee
any magical benefits over well done UL, I find.
The fact is that there is some leakage inductance between the
cathode and plate windings, and this tends to limit what one might achieve.
The positioning of the CFB windings relative to the plate windings is crucial
to achieving an outcome which is only ever going to be marginal above
a triode connected stage, or a UL stage, or a pentode stage.
Should you try a plain shunt resistor NFB network from the output plates
to the driver tube plates, the results can be identical to the CFB winding
approach, except that the driver tubes have to work into a lower load,
so some care must be taken with making sure the
feedback network does not load down the driver stage.
One undeniably and theoretically beneficial thing about
local voltage NFB in the output stage that we have discussed
is that the leakage inductance to the secondary is not included
in the local loop of NFB.
And less global NFB needs to be used, therefore
greater overall stability can be attained.
In the amp I am working on at present, described above,
I was able to apply a total of 38 dB of NFB before tha amp
became incurably unstable into a resistive load.
So with 17 dB of total NFB, the margin of stability is 21 dB.
10 dB is usually enough.
However, if one has used about 2,000 turns of primary on a big
core, and has four secondary sections, five primary sections,
or vice versa, the leakage L
will be so low, that the delays caused by the this will be negligible
when using global loop NFB, and a flat response even into
ELS type speakers is routine, up to 30 kHz.
> > Some folks still prefer an SET, and in some cicumstances,
> > they are all that is needed.
> >
>
> The tastes are different.
Indeed.I find most folks don't mind an SE amp's THD which is about10 times that of an equivalent powered PP
circuit, at all levels.
But as long as say we have only 0.5% of 2H at 20 watts,
all is well.
Patrick Turner. http://www.turneraudio.com.au
Henry Pasternack
> Triodes already contain feedback, beleive or not.
Bollocks, as they say in Australia.
Prove it.
-Henry
JOHN L STEWART
covering feedback in triodes. Better have a look. You might change your mind.
And perhaps you won't. Not to worry. I won't!!
By the way, I just simulated a 6L6 using two triodes, the 6AC5 & the 6P5.
It's a 4 terminal thing, two of them inputs. Not a cascode, as you & others might think.
One of the triode plates passes for the screen. It works in a UL connexion to apply FB.
It works, just as you would predict. I will post it one of these days so others can object,
as they are want to do!!!
Cheers John L Stewart
Henry Pasternack
news:3C5F3B81...@sympatico.ca...
> Hello Henry- Just posted a learned paper on alt.binaries.radio.pictures
> covering feedback in triodes. Better have a look.
I've read the paper. It's an operational analysis of an external circuit
model based on the insertion of a fictitious grid into the triode, turning
it into a pentode with internal negative feedback. There's nothing in the
paper whatsoever that shows that the triode inherently has negative feedback
built in. It's just an analysis trick, of limited value as far as I'm
concerned.
Hey, I could take a zero feedback SET amplifier and model it as a pentode
amplifier with negative feedback. I would follow approximately the same
strategy as used in this paper, only apply the analysis to the amplifier
as a whole. Would that then justify my saying that a ZFB SET amplifier
is inherently a negative feedback system? Bollocks.
I posted a pretty clever (if I may say so myself) argument that a triode
does *not* have inherent negative feedback and so far there hasn't been
a single direct rebuttal to any of the points in my article. I have a
feeling nobody even bothered to understand what I was saying. I'm not
going to make a federal case out of it, but heck, from my point of view
why even bother with the discussion? As you say below:
> You might change your mind. And perhaps you won't. Not to worry. I
> won't!!
You've already made up your mind, right? So there you go.
Anyway, the paper is interesting, even if it states outright what I've
said all along, that the premise of inherent negative feedback in a triode
is a fiction that has no basis in the actual operation of the device. I
do wish more people would simply realize this.
-Henry
Steve Bench
>does *not* have inherent negative feedback and so far there hasn't been
>a single direct rebuttal to any of the points in my article. I have a
>feeling nobody even bothered to understand what I was saying.
Hi Henry,
I thought it was a pretty dang good argument.
There was a parallel argument (as triodes are "best" modeled
as inherent feedback devices) by post doc physics "students"
claiming EVERYTHING is moving at the speed of light: used
as a convenience to aid in their research studies. Reality check?
If it floats ones boat to so model things, it can be a convenience.
But for a more complicated model (feedback process) to be
worth while, it's gotta predict something BETTER than the
conventional model.
Lets do a gedanken experiment...
Voltage gain of a triode is lower due to the feedback. (so said)
What else changes with the application of the same kind
of feedback? Hmmm, effective input capacitance. (Miller feedback).
Lower gain of triode should be therefore be mirrored by the
same lowering of input capacitance. Does that seem to be the
case? Don't think so, but more "accurate" comparison may
be needed to make a fair comparison.
Given that, should I conclude that the more complex model produces
worse predictions? If THAT is so, I know which model I'd use.
Perhaps there are some areas that the feedback model produces
better results. If so, it's a valid one to consider. However, I've
yet to see more accurate predictions with the feedback model.
In my "work" field (telecom), the standards bodies use all
sorts of wierd models of various aspects of the world. Some of
them have absolutely no relationship to the underlying physics,
yet produce useable and repeatable results. Should they be
abandoned? Probably not, but you still end up grimacing at
the "model".
Best Regards,
Steve
Check my web page .. <A
HREF="http://members.aol.com/sbench101/">http://members.aol.com/sbench101</A>
Remove the .gov to EMail me
Patrick Turner
Henry Pasternack wrote:
I am sure it is, or was a british word, that there bollocky
word,and far too strong for a tuesday.
> Prove it.
I am all ears and eyes for the model of a triodewhich
incorporates a feedback network in addition to
the basic voltage generator plus resistance model we are
all so
familiar with.
If the FB is real, then a model can be drawn up,
and it can be used for all designs.
I might add, that there is the alternative model to the
gene + R, and that
is the one using a current source, providing current due to
Eg x Gm,
with infinite output resistance, but with the plate
resistance strapped from
its output to ground.
I prefer the voltage gene plus series resistor method of
describing
all tubes, triodes and pentodes.
It does not throw understanding of dissipation in the tube,
of the various DC and AC power, for that, it is easiest to
go to the RDH4, and work through the formulas,
or more simply measure what you have built,
after following empiracle / graphical / simulated design
methods for loads.
Patrick Turner.
Patrick Turner
JOHN L STEWART wrote:
> Hello Henry- Just posted a learned paper on
> alt.binaries.radio.pictures
> covering feedback in triodes. Better have a look.
> You might change your mind.
> And perhaps you won't. Not to worry. I won't!!
>
> By the way, I just simulated a 6L6 using two
> triodes, the 6AC5 & the 6P5.
> It's a 4 terminal thing, two of them inputs. Not a
> cascode, as you & others might think.
> One of the triode plates passes for the screen. It
> works in a UL connexion to apply FB.
> It works, just as you would predict. I will post
> it one of these days so others can object,
> as they are want to do!!!
>
> Cheers John L Stewart
We are waiting anxiously,as we are the only species
with awareness of future being......
Patrick Turner.
Patrick Turner
Henry Pasternack wrote:
> "JOHN L STEWART" <jh.st...@sympatico.ca> wrote in message
> news:3C5F3B81...@sympatico.ca...
> > Hello Henry- Just posted a learned paper on alt.binaries.radio.pictures
> > covering feedback in triodes. Better have a look.
>
> I've read the paper. It's an operational analysis of an external circuit
> model based on the insertion of a fictitious grid into the triode, turning
> it into a pentode with internal negative feedback. There's nothing in the
> paper whatsoever that shows that the triode inherently has negative feedback
> built in. It's just an analysis trick, of limited value as far as I'm
> concerned.
>
> Hey, I could take a zero feedback SET amplifier and model it as a pentode
> amplifier with negative feedback. I would follow approximately the same
> strategy as used in this paper, only apply the analysis to the amplifier
> as a whole. Would that then justify my saying that a ZFB SET amplifier
> is inherently a negative feedback system? Bollocks.
>
> I posted a pretty clever (if I may say so myself) argument that a triode
> does *not* have inherent negative feedback and so far there hasn't been
> a single direct rebuttal to any of the points in my article. I have a
> feeling nobody even bothered to understand what I was saying.
I understood what you wee saying, about triode being like diodes,and so on.
No need to be discouraged.
Should anyone be certain they are correct think us plain dumb,
or incorrect, then they just need to
try a little more to convince us,
which takes time and patience.
There was a guy in 1600 who was burned at the stake by the
Church for suggesting the world went around the sun,
rather than ageeintg that everything went around the earth.
We won't burn heretics anymore, and alternative views
are welcome,
even though the occasional "bollocks" might be uttered.
> I'm not
> going to make a federal case out of it, but heck, from my point of view
> why even bother with the discussion? As you say below:
>
> > You might change your mind. And perhaps you won't. Not to worry. I
> > won't!!
>
> You've already made up your mind, right? So there you go.
>
> Anyway, the paper is interesting, even if it states outright what I've
> said all along, that the premise of inherent negative feedback in a triode
> is a fiction that has no basis in the actual operation of the device. I
> do wish more people would simply realize this.
Well, lets here it from the triode feedback believers, as to whythey are
convinced about FB in triodes.
Patrick Turner.
>
>
> -Henry
Patrick Turner
Steve Bench wrote:
Take the idea of the tube being a combination of a low impedancevoltage generator
producing mu x eg volts output
in series with a mythical resistance equal to the plate resistance.
Suppose we have a pentoad tube, say an EL34.
the mu of EL34 is 132, and Ra is 12,000 ohms, so
if we have 10 vrms applied to the grid, it follows that 1,320 vrms is
produced at the "gene" output, before going to the load,
through the plate resistance of 12,000 ohms.
But we can probe all day with a voltmeter, and never find any sign
of 1,320 volts being generated anywhere.
All we find is the load voltage, a couple of hundred perhaps,
which is produced at the plate.
But the tube acts just as if there was a low Z gene plus 12 kOhm R inside the
tube.
from this model, we can work out simply what
gain and output voltage we'll get with various loads.
The same EL34 can be connected as a triode, and its gene mu is only
10, and its R is 1,250 ohms.
The model can be used to describe signal flows around circuits.
So what's changed?
Well, if we looked at the partial triode, or UL connection,
we'd say the screen accepts feedback fro the plate transformer coil,
and the plate current is varied by partial application of itself
to a grid in the tube.
When the screen is connected to the plate, isn't the circuit like
a pentode with 100% of screen feedback????
But with a 300B, or 2A3, that there screen just ain't there.
So what gives, eh?
Patrick Turner.
Fred Nachbaur
Patrick Turner wrote:
>
>>[...]
>
> Well, lets here it from the triode feedback believers, as to whythey are
> convinced about FB in triodes.
>
> Patrick Turner.
Well, ok. For me it's not a matter of belief or conviction, merely a
convenience which - to my brain - helps make sense of the way a triode
behaves as compared to a tetrode or pentode. As Steve Bench pointed out
in his excellent post, a model does not necessarily have to parallel
actual physical phenomena to be useful.
As another example, I prefer to think of electricity as flowing from
positive to negative, since it makes the maths easier in loop equations
and such. "NO NO," I can almost hear Henry's protestations, "electricity
flows from negative to positive!" Certainly true. But if it easier for
me to conceptualise, what's wrong with assuming a positive-to-negative
flow direction? Or viewing a triode as an inherent NFB device? Or, even
imagining that the world is flat - at least flat enough to use ordinary
two-dimensional coordinates to get from here to the other end of town?
Cheers to all,
Patrick Turner
Fred Nachbaur wrote:
> Patrick Turner wrote:
>
> >
> >>[...]
>
> >
> > Well, lets here it from the triode feedback believers, as to whythey are
> > convinced about FB in triodes.
> >
> > Patrick Turner.
>
> Well, ok. For me it's not a matter of belief or conviction, merely a
> convenience which - to my brain - helps make sense of the way a triode
> behaves as compared to a tetrode or pentode. As Steve Bench pointed out
> in his excellent post, a model does not necessarily have to parallel
> actual physical phenomena to be useful.
That's what I have maintained all along with variousmodels of a tube
incorporating a voltage gene plus R for plate resistance.....
>
>
> As another example, I prefer to think of electricity as flowing from
> positive to negative, since it makes the maths easier in loop equations
> and such. "NO NO," I can almost hear Henry's protestations, "electricity
> flows from negative to positive!" Certainly true. But if it easier for
> me to conceptualise, what's wrong with assuming a positive-to-negative
> flow direction?
I am use to the convention of positive flow, but reallythe electrons are going
from negative to positive.
Don't worry Fred, I am able to have several crazy ideas
in the one brain, for review, and possible updating, at the one time.
> Or viewing a triode as an inherent NFB device? Or, even
> imagining that the world is flat - at least flat enough to use ordinary
> two-dimensional coordinates to get from here to the other end of town?
My piece of Earth around me isn't so flat, there are hills and rivers and
things,and water is flooding down the rivers at present, and so the darn earth
just isn't darn flat.
And when I rode a bicycle, I sure found out it wasn't flat....
Patrick Turner.
Fred Nachbaur
Patrick Turner wrote:
>
>>[...]
>
> My piece of Earth around me isn't so flat, there are hills and rivers and
> things,and water is flooding down the rivers at present, and so the darn earth
>
> just isn't darn flat.
> And when I rode a bicycle, I sure found out it wasn't flat....
You noticed that too? Well, there we have the Scientific Method at work.
I made the exact same observation, proving conclusively that bicycles
cause the world to be unflat. ;-)
Patrick Turner
Fred Nachbaur wrote:
> Patrick Turner wrote:
>
> >
> >>[...]
>
> >
> > My piece of Earth around me isn't so flat, there are hills and rivers and
> > things,and water is flooding down the rivers at present, and so the darn earth
> >
> > just isn't darn flat.
> > And when I rode a bicycle, I sure found out it wasn't flat....
>
> You noticed that too? Well, there we have the Scientific Method at work.
> I made the exact same observation, proving conclusively that bicycles
> cause the world to be unflat. ;-)
And it is only level, flat ground which allows water to flow.Visits to Niagra falls
cause only deluded observations,
caused by the up and down travelling
to get to the falls.
And when travelling the one percent of time during bike rides which are downhill,
and finding the brakes don't work, we have the Religious Method at work.
Lance Armstrong is really quite a mystic, as he is able to proove
the few hills and dales around France are actually flat.
Meanwhile, the God of Triodes hasn't quite explained fully
to us dummies about NFB in His tubes, and really,
he oughta send a Redeemer to put us right.
Patrick Turner.
GlassAmps
answered. The original question is self-limiting. GIVEN: the amp in question
is UL. DESIRE: Modify it to triode. ANSWER: Not a lot of choices here. Find
the screens of the o/p tubes. Lift the wire (that means unsolder) from
connection to the screen. The other end of that wire goes to the screen tap in
the primary of the o/p transformer. INSULATE THE END OF THE WIRE CAREFULLY.
It carries higher voltage B+ than the anodes do.
Solder a 200R or 220R resistor between the plate and screen terminal on the
tube socket.
If available and necessary, rebias back to the same total cathode current that
was drawn when it was a UL layout and you are 95% close. The other 5%, for the
average owner should be disregarded. It involves tweaking the feedback (you
see, changing from UL to triode DOES change the effective p-p impedance) for
which you typically need a 'scope and a sine wave and square wave generator. A
designer would be going for good sine wave, with linearity and minimum
overshoot and minimum droop on the square wave. BUT!!! if you're doing the UL
to triode conversion at home, just do the screen change, and enjoy the music.
Patrick Turner
GlassAmps wrote:
Yes, all of what you say is what you have to do at least, to convert from ULto
triode.
Couple of things to consider though.
The UL load for 2 x EL34 is often 6,600 ohms, which is a bit
low when triodes are used, if our aim is for best fidelity rather than
any great concern about power.
So if there is a 4 ohm tapping point for 4 ohm speaker connection,
assuming there is an 8 ohm tap already, then using the 4 ohm outlet
with 8 ohm speaker will result in the tube load going to a higher,
and more favourable load of 13,200 ohms.
Plate voltage swing will be greater, and the power output may still be
about 12 watts, and distortion more than halved, since the
class A content of what is produced is much higher.
The triode gain with 13,200 ohms will be higher, and no
alteration to the feedback network should have to be made.
The output impedance will also be reduced by half.
One could check all this out with a CRO, but it isn't absolutely necessary,
and the above can be tried anyway.
The triode amp should sound sweet and detailed, with excellent
and precise HF, never sibilant, and bass should be tight and clean,
with neutral mids.
But most well done UL amps provide all this too,
and sometimes switching to triode is not so easy to hear.
Patrick Turner. http://www.turneraudio.com.au
Choky
--
Choky
Prodanovic Aleksandar
choky*remove*@eunet.yu
YU
remove *remove* to reply!!
===================
"Fred Nachbaur" <fr...@netidea.com> wrote in message news:3C6002CE...@netidea.com...
Fred Nachbaur
look, the lumpier it gets. :)
Denis
> Henry Pasternack wrote:
> I am all ears and eyes for the model of a triodewhich
> incorporates a feedback network in addition to
> the basic voltage generator plus resistance model we are
> all so
> familiar with.
> If the FB is real, then a model can be drawn up,
> and it can be used for all designs.
First. There is NO generic voltage generator inside the triode. The
control grid controls just the plate current. If a triode sees, for
example, an infinite load the variation of plate current will always
zero,and it will be nothing that internal feedback that will cause
such variation in plate voltage that corresponds to the Vg variation
causing the same DIp, i.e DVp=-mu*DVg.
Second. The expression for the gain of a triode stage with any plate
load looks exactly like that of a feedback amp.
Third. The feedback model gives absolutely accurate predictions for
all cases of triode operation.
If we examine the model of v gene + Ri, we'll find it wrong for any
intermediate plate load between zero and infinity.
>
> I might add, that there is the alternative model to the
> gene + R, and that
> is the one using a current source, providing current due to
> Eg x Gm,
> with infinite output resistance, but with the plate
> resistance strapped from
> its output to ground.
This model, indeed, can give correct prediction of gain for any load,
but it leaves a lot of questions considering the static (DC) operation
of a triode.
Feedback model describes everything.
> I prefer the voltage gene plus series resistor method of
> describing
> all tubes, triodes and pentodes.
As I've already told, this model has no physical background behind.
Denis Afanassyev.
Denis
>
> Take the idea of the tube being a combination of a low impedancevoltage generator
> producing mu x eg volts output
> in series with a mythical resistance equal to the plate resistance.
>
> Suppose we have a pentoad tube, say an EL34.
> the mu of EL34 is 132, and Ra is 12,000 ohms, so
> if we have 10 vrms applied to the grid, it follows that 1,320 vrms is
> produced at the "gene" output, before going to the load,
> through the plate resistance of 12,000 ohms.
> But we can probe all day with a voltmeter, and never find any sign
> of 1,320 volts being generated anywhere.
> All we find is the load voltage, a couple of hundred perhaps,
> which is produced at the plate.
This proves again that no low-Z generic voltage generator could be
found in any tube.
The pentode or tetrode also has some internal feedback, but much lower
than that of triode. Otherwise the mu of any pentode would be
infinity.
Indeed, as the pentodes usually loaded with impedances much lower than
their Ri, this feedback can be omitted for the first approximation.
> The same EL34 can be connected as a triode, and its gene mu is only
> 10, and its R is 1,250 ohms.
> The model can be used to describe signal flows around circuits.
>
> So what's changed?
>
> Well, if we looked at the partial triode, or UL connection,
> we'd say the screen accepts feedback fro the plate transformer coil,
> and the plate current is varied by partial application of itself
> to a grid in the tube.
> When the screen is connected to the plate, isn't the circuit like
> a pentode with 100% of screen feedback????
>
Certainly yes, and this 'pentode' is a triode.
Denis Afanassyev.
Denis
> With any tube, the plate resistance with NFB, Ra' is
> Ra divided by [ ( Mu + 1 ) x B ].
> We can appoximate this for pentodes,
> Ra' = Ra divided by [ Gm x B ]
>
> So if we have 12.5% of the total primary winding as a cathode
> winding, and we are using an EL34, in pentode,
> then Ra' = 12 kOhms divided by [ (132 + 1 ) x 0.125 ]
> and this is 720 ohms, or nearly half that of the triode without
> any NFB loops.
> If the load was about 2.5 kOhms, then the gain
> with no NFB will be 21.3 times, and with the CFB,
> the gain would be 7.6, so the gain reduction would be
> 2.78 times, or 9dB, ie, we have applied 9 dB of NFB.
This seems close to the correct figure for EL34
Indeed, the amount of total NFB(local+global) around the power stage
is always
A=1+Rl/Ri, and is equal to that damned damping factor. This total NFB
varies with load impedance, what is important for understanding the
operation of the amplifier with different loads.
> Probably only 6 dB of NFB is required to
> lower most pentode Ra to that of the same tube in triode.
>
This amount of local NFB is equal to the damping factor of this tube
in triode connection as referred to the same load. Indeed, the CFB
connection allows one to get the same output power with higher Rl, so
the damping factor and the local NFB will be substantionally higher
than in triode.
> But low mu pentodes would react differently.
> The higher the mu, and Gm, the better the results.
Not always. I'm agree with Gm, but not mu. The CFB connection means
also some amount of UL NFB, because the screen-to-cathode voltage is
modulated by means of the cathode winding. The lower the mu2(the
triode mu of a pentode), the higher the contribution from this UL NFB.
And it is desirable to keep UL percentage below 20%.
> EL84 react very favourably to CFB application.
Just as any pentode.
>
> > > All sorts of different combinations of harmonic products are produced
> > > in a pentode at different loads.
> >
> > The pentodes can be put to such a mode of operation, when their
> > distortion spectrum is much softer than that of any triode. They are
> > most flexible amplification devices ever made.
>
> "Softer"?, what do you mean, in engineering terms?
>
It is possible to cancel almost totally the whole set of odd
harmonics, especially the third one. The physical mechanism of such
cancellation is the redistribution of the cathode current between the
plate and the screen grid.
> >
> > > Speakers vary widely in their impedance, and most triodes,
> > > without any connected loops of feedback, cope much better,
> > > and all frequencies, with less distortion.
> >
Sometimes the triodes actually ADD distortions instead of lowering
them.
The pentodes with CFB are much linear that those triode connected.
> > Triodes already contain feedback, beleive or not.
>
> But not feedback which has be contrived by humans.God put whatever feedback is in triodes there.
> Perhaps you can't grasp the idea of God, but there must be one,
> as the universe just didn't put itself there, with triodes,
> and laws of electron flow in vacuums, all by itself.
> Pardon me if I tell you I hav't the fogiest idea of who or what
> God is, but I sure can see where He has been, and what
> He has been up to for awhile.
>
The God or Mother nature gives us the only possibility to obtain a
low-Z amplification device - with NFB.
> Therefore the only
>
> > correct comparison could be done fo a triode without external feedback
> > and a pentode with it.
>
> Depends how much NFB is used.
>
The same in both cases, i.e., the external feedback applied to the
pentode is to be equal to the internal one of a triode with the same
load.
> >
> > In contrary, I have an amplifier with EL509s in output stage, which
> > generates only 2nd hadmonic for most of it's dynamic range, and
> > behaves much the same from open output to 2 Ohms load. I'm sure this
> > is also possible with other power pentodes, especially 6550.
>
> No, it isn't possible, and I don't see how it could be with theEL 509, unless it is in SE mode, when the 2H
> THD will
> sometimes be greater than the 3H, depending on load.
If I'll show the selected distortion signal of my amplifier, will you
beleive?
> With all PP, nearly all the 2H and even numbered harmonics dissapear, and 3H is
> usually the main THD component.
One can make it zero, at last well below the noise.
> The only 2H that can exist in a PP amp is that produced by the
> driver stage, but generally it can be kept very low with a
> careful design.
Yes, this 2nd harmonic goes from the driver stage, and equally from
the residual imbalance of the power tubes.
> Class AB pentodes produce 3H and almost no 2H,
> in the output stage. They also produce a lot of other higher odd order
> THD components.
>
The SET amps also can make the same, and even worse, approaching
clipping point.
> >
> >
> > > And SE pentodes have appalling distortion at their recomended loads and powers,
> > > many are 12%, very high number of components, not much to recommend them.
> > > UL over comes a lot of these shortcomings.
> >
> > Proper operating point optimisation and application of cathode
> > feedback makes the pentode an ideal tube for a SE stage too.
> > Everything I said about the harmonics spectrum is applicable.
> > Moreover, the meschanism of 3rd harmonic suppression was described as
> > early as in 1950s.
>
> Basic feedback theory.
This mechanism has nothing common with feedback.
>
> > >
> > > I havn't used EL509, but I bet it is not entirely unique in that
> > > there is less distortion than if you connected it as a triode.
Quite opposite. Moreover, the low mu2 of EL509 makes it especially
suitable for CFB application.
> >
> > The tastes are different.
>
> Indeed.I find most folks don't mind an SE amp's THD which is about10 times that of an equivalent powered PP
> circuit, at all levels.
> But as long as say we have only 0.5% of 2H at 20 watts,
> all is well.
I have rather 0.05% of H2, and that is all distortion at this level.
Denis Afanassyev. www.cortmi.com.ua/omak
Patrick Turner
Denis wrote:
> Patrick Turner <in...@turneraudio.com.au> wrote in message news:<3C5F89D5...@turneraudio.com.au>...
> > Henry Pasternack wrote:
>
>
> > I am all ears and eyes for the model of a triodewhich
> > incorporates a feedback network in addition to
> > the basic voltage generator plus resistance model we are
> > all so
> > familiar with.
> > If the FB is real, then a model can be drawn up,
> > and it can be used for all designs.
>
> First. There is NO generic voltage generator inside the triode. The
> control grid controls just the plate current.
Nobody, including my humble self, ever said there WAS a REAL voltage geneinside any triode.
But the triode acts as if there was an IMAGINERY generator inside....
The gene part is just a mental concept, and very useful.......
> If a triode sees, for
> example, an infinite load the variation of plate current will always
> zero,and it will be nothing that internal feedback that will cause
> such variation in plate voltage that corresponds to the Vg variation
> causing the same DIp, i.e DVp=-mu*DVg.
When a true CCS load is connected to the triode, there is, I agree,no current change, and triodes are
remarkable voltage devices,
and all that changes is the plate voltage, when the grid voltage is changed.
And it is under these conditions that the triode is at its most natural linearity.
So what is meant by "and it will be nothing that internal feedback that will cause
such variation in plate voltage that corresponds to the Vg variation
causing the same DIp, i.e DVp=-mu*DVg."
Do you mean "there is zero feedback"
Or do you mean there is infinite feedback, and hence
infinite linearity, which doesn't occur....
>
>
> Second. The expression for the gain of a triode stage with any plate
> load looks exactly like that of a feedback amp.
Gain, A, = Mu x RL divided by ( Ra plus RL )
Where is any mention of the fraction fed back, beta,
or the common feedback factor ( A + 1 ) ???
>
>
> Third. The feedback model gives absolutely accurate predictions for
> all cases of triode operation.
OK, what's the model?
> If we examine the model of v gene + Ri, we'll find it wrong for any
> intermediate plate load between zero and infinity.
Nah, it works with ALL loads which are between zero, and infinity.Imagine a gene with Mu x Eg, with an EL84
in triode.
say we have 1 volt input.
as Mu = 20, 20 volts come out the gene, and is assumed
to be a perfect imaginery voltage source.
So we have a resistor in series with any load you like, and this R
is 2 kOhms, equal to our plate resistance,
and we can figure out whatever load voltage exists,
as a simple divider is there, and ohms law is easy.
If the load was 2 kOhms, same as the plate resistance R, then
10 volts output would be across this load, and
the circuit has a gain of 10.
l
Same with EL84 in pentode, which might have the imaginery R equal to
38 kOhms, and the gene gain is still Mu x Eg, which is 400 times 1 volt
in the case of the EL84.
The gene output will be 400 volts, which doesn't actually exist,
and it cannot be measured.
But if you built a very low output resistance voltage amp with a gain of 400,
and added a resistor of 38 kOhms, in series with the load,
the two together would behave exactly like
an EL84 pentode.
Well not quite 100% exactly, as the voltage distortion products would be
present with a real pentoad, and a fake, or imaginery pentoad
could be made free of such artifacts.
Sometimes we have to ignore some aspects of operation to
understand other aspects, in this case the main gain and nature and value of
plate resistance.
> > I might add, that there is the alternative model to the
> > gene + R, and that
> > is the one using a current source, providing current due to
> > Eg x Gm,
> > with infinite output resistance, but with the plate
> > resistance strapped from
> > its output to ground.
>
> This model, indeed, can give correct prediction of gain for any load,
> but it leaves a lot of questions considering the static (DC) operation
> of a triode.
Neither model explains plate power dissipation well.
>
>
> Feedback model describes everything.
If only we had your model drawn up in front of us.
>
>
> > I prefer the voltage gene plus series resistor method of
> > describing
> > all tubes, triodes and pentodes.
>
> As I've already told, this model has no physical background behind.
Sure, but models are just tools of trade, for the fella who wants towork out what operational voltages and
gains he's gonna get with any
tube.
Something unreal is invented and used to explain something real.
Imagine if you had a metal box, and inside this box was
some mystery device.
But on the front of the box are a pair of input terminals,
with very high input impedance, and on the back of the box
there were two output terminals,
of unknown output impedance, and the changes you
made to the input voltage affected the output voltage.
You could easily apply a fixed voltage to the input terminals,
and then connect various load values to the output terminals,
and measure the gains with each.
It is very simple to work out what the output impedance of the device
in the box, by dividing the voltage change between two different loads,
by the current change between the two loads.
I use this method to measure output impedances of amplifiers.
Now if we found that our box displayed Ro of 2 kOhms,
and gain just like an EL84 connected as a triode,
then I bet when we opened the box, voila, a bloomin EL84
will stare us in the face.
But will the presence of any feedback stare at us?Patrick Turner.
Patrick Turner
Denis wrote:
Not necessarily below 20%.Manley amps use 60% UL.
I use 60% for 13E1 SEUL.
All sorts of values can be used, to maximise power, and linearity,
and freedom from multiple high order THD products.
>
>
> > EL84 react very favourably to CFB application.
>
> Just as any pentode.
>
> >
> > > > All sorts of different combinations of harmonic products are produced
> > > > in a pentode at different loads.
> > >
> > > The pentodes can be put to such a mode of operation, when their
> > > distortion spectrum is much softer than that of any triode. They are
> > > most flexible amplification devices ever made.
> >
> > "Softer"?, what do you mean, in engineering terms?
> >
>
> It is possible to cancel almost totally the whole set of odd
> harmonics, especially the third one. The physical mechanism of such
> cancellation is the redistribution of the cathode current between the
> plate and the screen grid.
Only at one load value, under certain conditions.Most pentoads have tranfer curves resembling a dog's hind leg,
and so I rarely bother to build any amp in pure pentoad.
> > >
> > > > Speakers vary widely in their impedance, and most triodes,
> > > > without any connected loops of feedback, cope much better,
> > > > and all frequencies, with less distortion.
> > >
> Sometimes the triodes actually ADD distortions instead of lowering
> them.
All devices add distortions, pentoads are worse than triodes,when optimally set up, at full power.
>
>
> The pentodes with CFB are much linear that those triode connected.
Not always.It depends on how much CFB, class of operation.
>
>
> > > Triodes already contain feedback, beleive or not.
> >
> > But not feedback which has be contrived by humans.God put whatever feedback is in triodes there.
> > Perhaps you can't grasp the idea of God, but there must be one,
> > as the universe just didn't put itself there, with triodes,
> > and laws of electron flow in vacuums, all by itself.
> > Pardon me if I tell you I hav't the fogiest idea of who or what
> > God is, but I sure can see where He has been, and what
> > He has been up to for awhile.
> >
>
> The God or Mother nature gives us the only possibility to obtain a
> low-Z amplification device - with NFB.
Father Nature allows us to thumb our nose at Mother Nature,and use Zero Feedback triode amps, and enter Sonic
Heaven,
if we want.
>
>
> > Therefore the only
> >
> > > correct comparison could be done fo a triode without external feedback
> > > and a pentode with it.
> >
> > Depends how much NFB is used.
> >
>
> The same in both cases, i.e., the external feedback applied to the
> pentode is to be equal to the internal one of a triode with the same
> load.
Ah, but we havn't agreed there actally is NFB in triodes,have we?
> >
> > > In contrary, I have an amplifier with EL509s in output stage, which
> > > generates only 2nd hadmonic for most of it's dynamic range, and
> > > behaves much the same from open output to 2 Ohms load. I'm sure this
> > > is also possible with other power pentodes, especially 6550.
> >
> > No, it isn't possible, and I don't see how it could be with theEL 509, unless it is in SE mode, when the 2H
> > THD will
> > sometimes be greater than the 3H, depending on load.
>
> If I'll show the selected distortion signal of my amplifier, will you
> beleive?
Not until I get around to using the EL509 I have in my stock,and then I will have to confirm if you are correct.
This all follows the laudable scientific tradition of
following someone's postulation, and then repeating the experiment
independantly, and comparing results.
You may recall Fleishman and Pons some years back said they
had achieved atomic fusion power at room temperature in a test tube.
The world was agog with hope, and many laboratories
carried out the same experiment, and no fusion power was
ever discovered.
All these years later, we still have no useful fusion power.
Fleishman and Pons were plain wrong.
So I believe Zero until I can proove it to myself,
but I will consider any idea.
> > With all PP, nearly all the 2H and even numbered harmonics dissapear, and 3H is
> > usually the main THD component.
>
> One can make it zero, at last well below the noise.
>
> > The only 2H that can exist in a PP amp is that produced by the
> > driver stage, but generally it can be kept very low with a
> > careful design.
>
> Yes, this 2nd harmonic goes from the driver stage, and equally from
> the residual imbalance of the power tubes.
>
> > Class AB pentodes produce 3H and almost no 2H,
> > in the output stage. They also produce a lot of other higher odd order
> > THD components.
> >
>
> The SET amps also can make the same, and even worse, approaching
> clipping point.
But not at a couple of watts, when the THD is nearly all 2H, unlessit is a pentoad, and depending what sort it
is, and the load,
various amounts of thd are still present.
> > >
> > >
> > > > And SE pentodes have appalling distortion at their recomended loads and powers,
> > > > many are 12%, very high number of components, not much to recommend them.
> > > > UL over comes a lot of these shortcomings.
> > >
> > > Proper operating point optimisation and application of cathode
> > > feedback makes the pentode an ideal tube for a SE stage too.
> > > Everything I said about the harmonics spectrum is applicable.
> > > Moreover, the meschanism of 3rd harmonic suppression was described as
> > > early as in 1950s.
> >
> > Basic feedback theory.
>
> This mechanism has nothing common with feedback.
You will have to write a book about it.
> > > > I havn't used EL509, but I bet it is not entirely unique in that
> > > > there is less distortion than if you connected it as a triode.
>
> Quite opposite. Moreover, the low mu2 of EL509 makes it especially
> suitable for CFB application.
Maybe, who else knows?
> > > The tastes are different.
> >
> > Indeed.I find most folks don't mind an SE amp's THD which is about10 times that of an equivalent powered PP
> > circuit, at all levels.
> > But as long as say we have only 0.5% of 2H at 20 watts,
> > all is well.
>
> I have rather 0.05% of H2, and that is all distortion at this level.
Fine.
Patrick Turner.
Denis
>
> Well, ok. For me it's not a matter of belief or conviction, merely a
> convenience which - to my brain - helps make sense of the way a triode
> behaves as compared to a tetrode or pentode. As Steve Bench pointed out
> in his excellent post, a model does not necessarily have to parallel
> actual physical phenomena to be useful.
>
And it is the model involving the internal negative feedback that is
based on the real physical phenomenon.
Not only the FB model allows one to make correct predictions of a
particular circuit performance, it reveals certain important facts
regarding the nonlinear effects in triodes.
For example, if one assumes the generic action of the grid is to
control the cathode current, it becomes clear that this action always
will be highly nonlinear. As the current is proportional to Vg^3/2,
the simple calculation will confirm that the THD at full class A
dynamic range( when the amplitude of AC voltage at the grid is
strictly equal to the bias voltage) will be as much as 14.5%.
Actally, such triode as 300B gives just half of this figure being
fully modulated. The FB model predicts 1/4th of this THD value as the
intenal NFB at this particular load is 12 dB(assuming the internal
feedback is linear< or the mu is constant). The difference in this
figures is attributed to the nonlinearity of internal NFB, which is
clearly evident as the mu varies depending both on Vg and Va. And this
is one of very best triodes.
From other hand the higher THD figures for SE pentode stages without
NFB are actually lower than those 14.5%. This is because the current
redistribution between the plate and the control grid can alter the
content of both even and odd harmonics, and actually it is possible to
get out of the 3H for a significant part of a dynamic range. Finally,
the addition of those 12 dB of NFB will give far better performance
than that of 300B.
Denis Afanassyev.
Henry Pasternack
news:<3C5F89D5...@turneraudio.com.au>...
> > Henry Pasternack wrote:
>
>
> > I am all ears and eyes for the model of a triodewhich
> > incorporates a feedback network in addition to
> > the basic voltage generator plus resistance model we are
> > all so
> > familiar with.
> > If the FB is real, then a model can be drawn up,
> > and it can be used for all designs.
I didn't write this.
-Henry
Henry Pasternack
Patrick Turner
Denis wrote:
> Fred Nachbaur <fr...@netidea.com> wrote in message news:<3C6002CE...@netidea.com>...
>
> >
> > Well, ok. For me it's not a matter of belief or conviction, merely a
> > convenience which - to my brain - helps make sense of the way a triode
> > behaves as compared to a tetrode or pentode. As Steve Bench pointed out
> > in his excellent post, a model does not necessarily have to parallel
> > actual physical phenomena to be useful.
> >
>
> And it is the model involving the internal negative feedback that is
> based on the real physical phenomenon.
What is the model????
>
>
> Not only the FB model allows one to make correct predictions of a
> particular circuit performance, it reveals certain important facts
> regarding the nonlinear effects in triodes.
>
> For example, if one assumes the generic action of the grid is to
> control the cathode current, it becomes clear that this action always
> will be highly nonlinear. As the current is proportional to Vg^3/2,
> the simple calculation will confirm that the THD at full class A
> dynamic range( when the amplitude of AC voltage at the grid is
> strictly equal to the bias voltage) will be as much as 14.5%.
Readers may not understand what Denis is saying.But basically, cathode current changes
proportionately
to ( square root of voltage between gid and cathode ) all cubed.
One can draw a graph of this line and it resembles the Vg = 0 volts line,
on the Ep / Ip data graphs.
Indeed, 14.5 % of THD is possible, with the placement
of the loadline at the worst position, where nobdy in their right mind would put it.
All the theory books rightly suggest locating the load line so
THD is limited to 5%, at maximum power at clip.
The RDH4 spells it all out clearly.
> Actally, such triode as 300B gives just half of this figure being
> fully modulated. The FB model predicts 1/4th of this THD value as the
> intenal NFB at this particular load is 12 dB(assuming the internal
> feedback is linear< or the mu is constant).
That's right, keep your feedback model a secret, and we'llall finish up like the mushroom
boys,
fed on bulldust, and kept in the dark.
> The difference in this
> figures is attributed to the nonlinearity of internal NFB, which is
> clearly evident as the mu varies depending both on Vg and Va. And this
> is one of very best triodes.
> From other hand the higher THD figures for SE pentode stages without
> NFB are actually lower than those 14.5%. This is because the current
> redistribution between the plate and the control grid can alter the
> content of both even and odd harmonics, and actually it is possible to
> get out of the 3H for a significant part of a dynamic range. Finally,
> the addition of those 12 dB of NFB will give far better performance
> than that of 300B.
Not necessarily so.Readers should see the comparison of harmonics produced by
a 2A3 at full power, and a 6F6 at full power, all well tabulated in the RDH4.
The pentoad is miles worse.
12 dB of NFB with the 300B will work wonders.
The same 12 dB of NFB in most pentodes may work wonders too, but there will be
a more complex mix of THD present.
In fact, the 300B is quite able to get by with zero NFB,
and also the 2A3, and I have two friends who have such systems.
Patrick Turner.
>
>
> Denis Afanassyev.
Denis
>
> Nobody, including my humble self, ever said there WAS a REAL voltage geneinside any triode.
> But the triode acts as if there was an IMAGINERY generator inside....
> The gene part is just a mental concept, and very useful.......
>
Indeed, the feedback model involves only those physical mechanisms,
which are clearly present in triode, therefore this model is not only
consistent with the observed results, but it is physically correct.
> > If a triode sees, for
> > example, an infinite load the variation of plate current will always
> > zero,and it will be nothing that internal feedback that will cause
> > such variation in plate voltage that corresponds to the Vg variation
> > causing the same DIp, i.e DVp=-mu*DVg.
>
> When a true CCS load is connected to the triode, there is, I agree,no current change, and triodes are
> remarkable voltage devices,
> and all that changes is the plate voltage, when the grid voltage is changed.
> And it is under these conditions that the triode is at its most natural linearity.
>
This loading condition reveals nothing but the natural nonlinearity
of the internal feedback(the variation of the mu over the loadline).
> So what is meant by "and it will be nothing that internal feedback that will cause
> such variation in plate voltage that corresponds to the Vg variation
> causing the same DIp, i.e DVp=-mu*DVg."
> Do you mean "there is zero feedback"
> Or do you mean there is infinite feedback, and hence
> infinite linearity, which doesn't occur....
>
The internal feedback in this ultimate case is INFINITE.
One should expect the zero distortions, but instead of this we'll see
the distortions due to the mu nonlinearity. There is no real triode
with perfectly constant mu.
> >
> >
> > Second. The expression for the gain of a triode stage with any plate
> > load looks exactly like that of a feedback amp.
>
> Gain, A, = Mu x RL divided by ( Ra plus RL )
>
> Where is any mention of the fraction fed back, beta,
> or the common feedback factor ( A + 1 ) ???
>
> >
> > Third. The feedback model gives absolutely accurate predictions for
> > all cases of triode operation.
>
> OK, what's the model?
>
The FB model is following:
The triode is considered a voltage-controlled current generator
enclosed into a negative voltage feedback loop, and the feedback
signal is taken from the plate.
Let us consider first the small-signal operation of the triode, i.e.,
the case of Vac<<Vg0.
In this condition the mentioned current generator is described by the
definite value of transconductance, gm, and the inverse gain factor,
1/mu is considered the return ratio(the beta) of the feedback loop.
These two parameters, gm and my are the generic ones fully describing
the amplification properties of a triode.
Let us assume the triode sees an active load Rl.
The plate current increment DIp corresponding to some increment in
grid voltage is
DIp= gm*(DVg+(1/mu)*DVp) this is a standard equation for a
triode
and DVp=Rl*DIp
omitting the necessary transformations, one obtains:
the open loop gain (without internal NFB),
G0=gm*Rl
the depth of internal nfb,
A=1+beta*G0=1+(1/mu)*gm*Rl=1+Rl/Ri
and the gain of loaded triode,
G=G0/(1+(1/mu)*gm*Rl)=(gm*Rl)/(1+Rl/Ri)=gm*Rl*Ri/(Rl+Ri)
Let one remember, that the small-signal gain is G=DVout/DVin,
in our case, G=DVp/DVg,
and the internal resistance is
Ri=DVp/DIp
Of course,
Ri=mu/gm,
and it is the secondary parameter of a triode.
There is a simple explaination for an appearance of some finite output
resistance in any current amplifier after addition of the voltage NFB.
The FB model, of course, allows one to fully describe a triode in
large signal mode of operation.
In this case the mu and gm become the functions of Vp and Vg, and the
math will go out of a possibility to post it there.
> Something unreal is invented and used to explain something real.
>
I've just shown how real it is.
Denis Afanassyev.
Patrick Turner
Denis wrote:
> Patrick Turner <in...@turneraudio.com.au> wrote in message news:<3C627D48...@turneraudio.com.au>...
>
> >
> > Nobody, including my humble self, ever said there WAS a REAL voltage geneinside any triode.
> > But the triode acts as if there was an IMAGINERY generator inside....
> > The gene part is just a mental concept, and very useful.......
> >
>
> Indeed, the feedback model involves only those physical mechanisms,
> which are clearly present in triode, therefore this model is not only
> consistent with the observed results, but it is physically correct.
But Denis, you are getting on everyon'es nerves by not statingexactly what the feedback model is.
I don't think there's anyone who knows what you are talking about.
>
>
> > > If a triode sees, for
> > > example, an infinite load the variation of plate current will always
> > > zero,and it will be nothing that internal feedback that will cause
> > > such variation in plate voltage that corresponds to the Vg variation
> > > causing the same DIp, i.e DVp=-mu*DVg.
> >
> > When a true CCS load is connected to the triode, there is, I agree,no current change, and triodes are
> > remarkable voltage devices,
> > and all that changes is the plate voltage, when the grid voltage is changed.
> > And it is under these conditions that the triode is at its most natural linearity.
> >
>
> This loading condition reveals nothing but the natural nonlinearity
> of the internal feedback(the variation of the mu over the loadline).
>
> > So what is meant by "and it will be nothing that internal feedback that will cause
> > such variation in plate voltage that corresponds to the Vg variation
> > causing the same DIp, i.e DVp=-mu*DVg."
> > Do you mean "there is zero feedback"
> > Or do you mean there is infinite feedback, and hence
> > infinite linearity, which doesn't occur....
> >
>
> The internal feedback in this ultimate case is INFINITE.
>
> One should expect the zero distortions, but instead of this we'll see
> the distortions due to the mu nonlinearity. There is no real triode
> with perfectly constant mu.
Perfection only exits in the mind of a dreamer.
> > > Second. The expression for the gain of a triode stage with any plate
> > > load looks exactly like that of a feedback amp.
> >
> > Gain, A, = Mu x RL divided by ( Ra plus RL )
> >
> > Where is any mention of the fraction fed back, beta,
> > or the common feedback factor ( A + 1 ) ???
> >
> > >
> > > Third. The feedback model gives absolutely accurate predictions for
> > > all cases of triode operation.
> >
> > OK, what's the model?
> >
>
> The FB model is following:
>
> The triode is considered a voltage-controlled current generator
> enclosed into a negative voltage feedback loop, and the feedback
> signal is taken from the plate.
Series or shunt feedback?With or without a resistor network?
When its drawn up, what does it look like?
Why not draw it up, and post it so everyone can see it?
Why do you leave us to fill in your unexplained gaps?
to proove your point,
you will have to be much clearer in what you are suggesting.
> Let us consider first the small-signal operation of the triode, i.e.,
> the case of Vac<<Vg0.
Not sure I understand. please be clearer.
>
>
> In this condition the mentioned current generator is described by the
> definite value of transconductance, gm, and the inverse gain factor,
> 1/mu is considered the return ratio(the beta) of the feedback loop.
I dunno anyone who understands this.
>
>
> These two parameters, gm and my are the generic ones fully describing
> the amplification properties of a triode.
You mean Gm, and Mu....Why use the word "generic"??
>
>
> Let us assume the triode sees an active load Rl.
Why active?Would not the load be a passive resistor?
>
>
> The plate current increment DIp corresponding to some increment in
> grid voltage is
> DIp= gm*(DVg+(1/mu)*DVp) this is a standard equation for a
> triode
"D" means, I think, delta, and readers, dear Denis means, I suspect,that delta Ip, etc, is a small quantity
of Ip, etc.
> and DVp=Rl*DIp
>
> omitting the necessary transformations, one obtains:
>
> the open loop gain (without internal NFB),
>
> G0=gm*Rl
>
> the depth of internal nfb,
>
> A=1+beta*G0=1+(1/mu)*gm*Rl=1+Rl/Ri
>
> and the gain of loaded triode,
>
> G=G0/(1+(1/mu)*gm*Rl)=(gm*Rl)/(1+Rl/Ri)=gm*Rl*Ri/(Rl+Ri)
>
> Let one remember, that the small-signal gain is G=DVout/DVin,
> in our case, G=DVp/DVg,
>
> and the internal resistance is
>
> Ri=DVp/DIp
>
> Of course,
>
> Ri=mu/gm,
>
> and it is the secondary parameter of a triode.
>
> There is a simple explaination for an appearance of some finite output
> resistance in any current amplifier after addition of the voltage NFB.
>
> The FB model, of course, allows one to fully describe a triode in
> large signal mode of operation.
>
> In this case the mu and gm become the functions of Vp and Vg, and the
> math will go out of a possibility to post it there.
>
>
> > Something unreal is invented and used to explain something real.
> >
>
> I've just shown how real it is.
>
Well you may think you are part of the way towards proving somereality, but you have an awful
long way to go to explain what is a simple thing simply.
But it would be worth it, and probably take a carefully written page,
with an appropriate diagram, which anyone can understand.
Most, if not all the readers of the newsgroup are terrible
at understanding simple maths, let alone any differential calculus equations
which just confuse everyone who hasn't spent 10 years at university.
Why cannot the simple concept of internal triode feedback be explained
without any maths, like standard external feedback paths?
Where are your book references for your idea
about negative feedback in triodes?
Where is the idea expressed elsewhere, so we can see it more
sucinctly written, easily understood, and proved?
Patrick Turner.
> Denis Afanassyev.
Henry Pasternack
> which are clearly present in triode, therefore this model is not only
> consistent with the observed results, but it is physically correct.
Your derivation of the triode gain equation using the negative feedback
model does not mean that there is literally negative feedback inside the
tube. The feedback equation requires that we divide the triode into two
parts, one representing the forward path gain and the other representing
the reverse path gain. But it's physically impossible to separate these
two things apart in a real triode. Your model of the forward path means
there is cathode current in the absence of the influence of plate voltage.
Clearly this is meaningless; the triode cannot conduct with no voltage on
the plate.
Your model is correct as an abstraction, but it has no more physical
significance than the more common voltage source or current source models.
To prove the point, I will now prove an ordinary voltage divider is also
an example of negative feedback.
Consider a single resistor, Rc. Connect one end to ground and apply a
voltage, vc, across it. This creates a current generator:
ic = vc * (1 / Rc)
For the sake of argument, define:
gm = 1/Rc
Now, insert a load resistor, Rl, between the negative end of Rc and ground.
Let's compute the forward "gain" of this circuit:
A = gm * Rl = Rl / Rc
Clearly there is 100% negative feedback. Any voltage appearing across Rl
subtracts fully from the input voltage, vc. So we can say:
B = -1
Now, plug A and B into the feedback equation:
Acl = A / (1 - AB) = (Rl / Rc) / (1 - (-1) * Rl / Rc)
Acl = (Rl / Rc) / (1 + Rl / Rc)
Multiplying both sides by Rc gives:
Acl = Rl / (Rc + Rl)
This, of course, is the standard voltage divider equation.
I've proven that there is 100% negative feedback in a passive voltage
divider. The model is correct, but is it physically significant? My
feeling is that the result is logically valid but conceptually absurd.
The same can be said of your triode analysis.
The triode is harder to understand than the resistor, and so the holes
in your feedback argument are not as obvious as they are in my resistor
example. I don't feel like making a big stink out of this, but I do
feel you've put yourself out on a limb by insisting your feedback model
is physically correct in the absence of a physical model to justify it,
especially given the many counter-arguments.
Have a day.
-Henry