Why IC Regulators are Bad
dang...@earthlink.net
Greetings Rat,
Well, I'll jump in w/ both feet this time -- flammers to the alert.
IC regulation w/ small Cout caps is bad for audio. I'll argue why.
This is in response the current threads on small caps after
regulators, and in general, to threads of this nature that appear in
RAT recurrently.
Consider: the tube stage as a reductionist model 3 port device,
input, power supply, output.
The PS port feeds through to output, attenuated by the stage PSRR.
The sound you hear at the output of the stage, is the composite of the
signal and the PS port regulator.
IC regulators are solid state monolithic negative feedback devices.
They work by comparing output to reference, and adjusting output to
minimize difference between output and reference. ( A classic OA type
NFB design. )
For most non-audio applications they are great. They can potentially
maintain good regulation, good enough for digital logic, industrial
controls, servo systems ... you name it.
Audio is different because some people can hear what comes out of the
IC regulator.
To work, IC regulators require a large bandwidth to respond to the
transient load changes on the output, and maintain constant supply
voltage. ( Straight from NFB theory ).
High bandwidth solid state NFB devices which are attached directly to
the PS port, inject high frequency solid state noise/harmonics
directly into the audio signal out.
( I'm not going to argue tubes vs. transistors here. For the sake of
argument I'll just presume that solid state NFB sound is to be
avoided. )
When you put a Large C on the output of the IC regulator, or any other
type of regulator, like tube or FET, you reduce the dv/dt on the out
cap. The reduction in dv/dt for any given current load, is directly
proportional to the increase in C out.
Reducing the dv/dt reduces the maximum response frequency of the out
node, and proportionally reduces the high frequency components, which
mix with the input signal to form the audio output of the stage.
This is the reason experimenters always learn that more is better w/
output capacitors. As C increases, dv/dt decreases, high frequency PS
noise decreases.
This is a continuous function of C out, and a proportional process.
Simply put, when you add enough C such that YOU AND YOUR HEARING can't
sense the noise/distortion anymore -- you're done. In theory, as C
out approaches infinity, dv/dt approaches zero or DC At some point
the dv/dt on the output cap approaches a low frequency/amplitude limit
inaudible to the listener, dependent upon hearing accuity.
I realize that according to classic NFB theory as properly quoted here
by many talented and dedicated EE's, and highly motivated
experimenters, this argument is sometimes not considered. For most
applications like servos and so on as quoted above, it really would
make no difference, and for an industrial designer to pursue
"purification" of the harmonic structure of the PS port, could amount
to professional suicide, due to the poor cost/performance tradeoff, as
measured against standard population distributions for hearing accuity
and home entertainment needs.
This argument is inseperable from psychoacoustic reality. Some people
find high frequency noise and distortion so objectionable, it is like
fingernails on the blackboard, and can ruin the listening experience.
Others think a high school band sounds like great music, played back
through a boom box. ( Especially when their kids are in the band !)
I think the difference in individual hearing acutity muddies this
arguement, because plenty of smart, well educated, highly trained
EE's, don't have a clue as to what the fuss is about, when audio
"fanatics" complain about poor sound. This has nothing to do w/ pro
engineer vs. amateur. It's just that everyone has different hearing
accuity.
The real technical challenge for the engineers in RAT, is to develop
an appropriate model for the relationship between SS/IC regulation,
high frequency harmonics and noise, and the preception of such by
population distribution. (I'm not volunteering)
Anyway, that's why IC regulators are bad, and big C on the output of
the regulator keeps improving things, in proportion to how much C your
hearing needs, and how much C you can afford to add.
Regards,
Dangerdave
Henry Pasternack
dang...@earthlink.net wrote:
: Anyway, that's why IC regulators are bad, and big C on the output of
: the regulator keeps improving things, in proportion to how much C your
: hearing needs, and how much C you can afford to add.
Nah, three-terminal IC regulators are bad because they are noisy and
have very low bandwidths. There's no reason a well-designed regulator
should sound bad. It's not clear, though, that the added complexity
is justified. As well, some circuits may sound better with unregulated
supplies for reasons that aren't objectively obvious.
Power supply and regulator design, and negative feedback design in
general, are complicated subjects. It's no wonder random hacking leads
to mixed sonic results.
-Henry
--
ATTENTION! Reply to h...@nortel.ca (hen...@nortel.ca won't work).
Thor...@tnt-audio.com
Yo all,
> IC regulation w/ small Cout caps is bad for audio. I'll argue why.
IC regulation is bad. Full stop. It's pretty useless in Digital too. Sure,
you must keep the Voltage stable, but the "regulation", the Voltage imprinted
on the Rail, is lousy unless you sorth these suckers out.....
> IC regulators are solid state monolithic negative feedback devices.
> They work by comparing output to reference, and adjusting output to
> minimize difference between output and reference. ( A classic OA type
> NFB design. )
And most Reg's have enough bandwidth to get them into trouble and not enough
to sound good.....
That's why I'm using the old LM340 Series so much. Low Bandwidth. Sloooow
Regulator. Good at DC, not good at AC.....
> For most non-audio applications they are great. They can potentially
> maintain good regulation, good enough for digital logic, industrial
> controls, servo systems ... you name it.
Actually no. Not good enough. Hence the extensive decoupling....
> To work, IC regulators require a large bandwidth to respond to the
> transient load changes on the output, and maintain constant supply
> voltage. ( Straight from NFB theory ).
I'd say at least > 200kHz open loop for the -3db point.....
> Anyway, that's why IC regulators are bad, and big C on the output of
> the regulator keeps improving things, in proportion to how much C your
> hearing needs, and how much C you can afford to add.
And now just imagine for a moment, what could happen if you where to to use a
regulation scheme that will keep the DC constant (stabilisation) but will
keep ANY AC out of the regulator and allows the Capacitor (which must be
large in Value) to dominate the AC Impedance (Regulation) of the PSU.
Implement that propperly and you get about 10Hz - a few MHz for a PSU rail
impedance of about 1-2 Ohm which in addition has only small reactive
components (do it propperly and there are only VERY SMALL reactive
Components....).
And no distortion from the SS Regulator get's anywhere....
Guess what? That is how I do thing and it sure works....
Later Thorsten
-----== Posted via Deja News, The Leader in Internet Discussion ==-----
http://www.dejanews.com/ Now offering spam-free web-based newsreading
dang...@earthlink.net
On 17 Jun 1998 19:04:16 GMT, hen...@nortel.ca (Henry Pasternack)
wrote:
>dang...@earthlink.net wrote:
>: Anyway, that's why IC regulators are bad, and big C on the output of
>: the regulator keeps improving things, in proportion to how much C your
>: hearing needs, and how much C you can afford to add.
>
> Nah, three-terminal IC regulators are bad because they are noisy and
>have very low bandwidths. There's no reason a well-designed regulator
>should sound bad. It's not clear, though, that the added complexity
>is justified. As well, some circuits may sound better with unregulated
>supplies for reasons that aren't objectively obvious.
>
> Power supply and regulator design, and negative feedback design in
>general, are complicated subjects. It's no wonder random hacking leads
>to mixed sonic results.
>
>-Henry
>
>--
>ATTENTION! Reply to h...@nortel.ca (hen...@nortel.ca won't work).
Thanks for the reply, Henry.
In principle, using ideal devices, you can surely design a "perfect"
NFB regulator (or audio amplifier), whose high frequency noise and
"sidebanded" distortion products are so far above the audio band as
to be imperceptible to any listener, golden ear'd or not.
However, in practice, the most expensive regulators I've seen use
discrete combinations of FET and BJT, in diff amp configs, w/ high
bandwidth, but still the same problem of feeding high frequency noise
into the PS port, unless swamped out by large Cout.
This seems analogous to the recurrent argument about SS vs Tube,
insofar as ideally a SS NFB amplifier can be designed w/ is "perfect"
for audio, but in practice, low NFB tubes sound better to a lot of
musicians, studio guys, audiophile hobbyists, and other more casual
listeners, when compared to even the best SS NFB designs.
One example that comes to mind -- Levinson megahertz BW SS diff
preamps. The best tube designs still sound better to many listeners.
Like the low level tube mic pre's and compressors going for megabucks
in the pro circles.
To other listeners (by one measure the lucky ones -- no need to
spend big bucks on audio stuff) a string and a can are hi fi, long as
you use one for each ear.
Regards,
Dangerdave
dang...@earthlink.net
On Wed, 17 Jun 1998 21:22:40 GMT, Thor...@tnt-audio.com wrote:
>
>> Anyway, that's why IC regulators are bad, and big C on the output of
>> the regulator keeps improving things, in proportion to how much C your
>> hearing needs, and how much C you can afford to add.
>
>And now just imagine for a moment, what could happen if you where to to use a
>regulation scheme that will keep the DC constant (stabilisation) but will
>keep ANY AC out of the regulator and allows the Capacitor (which must be
>large in Value) to dominate the AC Impedance (Regulation) of the PSU.
>
>Implement that propperly and you get about 10Hz - a few MHz for a PSU rail
>impedance of about 1-2 Ohm which in addition has only small reactive
>components (do it propperly and there are only VERY SMALL reactive
>Components....).
>
>And no distortion from the SS Regulator get's anywhere....
>
>Guess what? That is how I do thing and it sure works....
>
>Later Thorsten
Thanks for the reply, Thorsten. This is turning into a good thread.
I think I may see what you mean, but I'm not entirely clear on how you
achieve this.
Are you saying that the pre-regulator filter is very stiff and limits
AC ripple at regulator in, w/ a rail impedance at this node of 1-2
ohms?
Good to hear from you again.
Regards,
Dangerdave
Kurt Strain
Thor...@tnt-audio.com wrote:
: Yo all,
: > IC regulation w/ small Cout caps is bad for audio. I'll argue why.
: IC regulation is bad. Full stop. It's pretty useless in Digital too. Sure,
: you must keep the Voltage stable, but the "regulation", the Voltage imprinted
: on the Rail, is lousy unless you sorth these suckers out.....
This reminds me of the time I worked on a very old developmental
computer (not HP) (back in 1981 - am I dating myself?). Anyway, they had
big computers then as you ought to know, and a medium mainframe like
this had no 5V regulators in it. What they did was build an UNregulated
5V 200A or so supply. The power supply was essentially three-phase (I think)
to rectified 5V direct, smoothed by about 2 Farads of capacitance or so,
and to get it to the right voltage they put in big load resistors until
it came down to that voltage. They of course had big power busses and
careful layout and they relied on the fact that all that switching will
statistically yield a fairly constant load. I'm sure they had many
local high frequency bypasses on the chips.
Regulators? I don't need no stinkin' regulators. My analog chain has none
throughout - except for phono tube filaments (high technology for lowest
tech application, I like that). I could probably benefit from light
pre-regulation, attempting to prevent brown-outs from sagging voltage but
doesn't try to correct output voltage with feedback, isolating it from
the final B+ with LC filtering.
Kurt
LIEUTENDGN
Integrated circuit regulators using " Bandgap Reference " is a lot better
sounding than the ones using " Zener diodes".
End user has to call the application engineering to find out which one they
are using in the producton floor.
The equivalent circiut does not tell you which type they use in production.
dang...@earthlink.net
True, they're distinguishable by the reference technology, but the SS
NFB configuration is common to both types, and so the effect of SS NFB
on sound quality.
Regards,
Dangerdave.
dang...@earthlink.net
On Wed, 17 Jun 1998 21:22:40 GMT, Thor...@tnt-audio.com wrote:
>Yo all,
>
>> IC regulation w/ small Cout caps is bad for audio. I'll argue why.
>
>IC regulation is bad. Full stop. It's pretty useless in Digital too. Sure,
>you must keep the Voltage stable, but the "regulation", the Voltage imprinted
>on the Rail, is lousy unless you sorth these suckers out.....
>
>> IC regulators are solid state monolithic negative feedback devices.
>> They work by comparing output to reference, and adjusting output to
>> minimize difference between output and reference. ( A classic OA type
>> NFB design. )
>
>And most Reg's have enough bandwidth to get them into trouble and not enough
>to sound good.....
>
>That's why I'm using the old LM340 Series so much. Low Bandwidth. Sloooow
>Regulator. Good at DC, not good at AC.....
>
>> For most non-audio applications they are great. They can potentially
>> maintain good regulation, good enough for digital logic, industrial
>> controls, servo systems ... you name it.
>
>Actually no. Not good enough. Hence the extensive decoupling....
>
I stand corrected. They're not just bad for Audio...
Regards,
Dangerdave
LIEUTENDGN
NFB introduces phase shift which in turns introduces phase distortion which no
consumer manufacturer is willing to specified in the advertisements.
I have the feeling that the regular consumers are being taken for a long ride.
Kind of sad !!!
LIEUTENDGN
It also heavily depends on whether the FAB is a " class 10" or "class 100" for
the production.
Class 10 usually yields lower noise.
SONY R&D Lab in Japan has the lowest nosie FET on their back burner.
I am surprised they never border to use what they have from their R&D lab for
the consurmer products.
May be they are saving it for the last minute shoot out ???
Their comptetitors are trying to do a reverse engineering on the FET with no
luck.
My suggestion : hold your breath.
Russ Sadd
Looks like everyone's agreeing that SS regulation isn't that great for valve
amps. So, what about using valves and stabilizer neons? There were a good
few things designed for valve circuits - neon stabilizers and 6AS7s
controlled by a feedback circuit were de rigeur in studio and military
applications.
Russ
---------------
E-Mail: sad...@aston.ac.uk
Web Pages: http://homepages.enterprise.net/icedragon/new/index.htm
Thor...@tnt-audio.com
Hi all,
> Thanks for the reply, Thorsten. This is turning into a good thread.
>
> I think I may see what you mean, but I'm not entirely clear on how you
> achieve this.
>
> Are you saying that the pre-regulator filter is very stiff and limits
> AC ripple at regulator in, w/ a rail impedance at this node of 1-2
> ohms?
I'll let my little secrout out of the bag. Here is how my PSU for the +B on
my MC-Phono/Line Preamp Works. The Pre is based on a variation of the Arthur
Loesch Circuit with ECC88's used throughout, Cascode Frontend and runs on
(nominal) 195V +B.
I use a 240VA Toroidal Mains-Isolation Transformer which here in the UK gives
me a convenient 240V AC Output. The (HEXFRED) Rectified Voltage goes into a
Choke Input Filter using a 10H Choke and two 470uF 450V Nichcon Computergrade
Cap's....
That leaves me with about 235V under load across the Cap's....
This Voltage is applied to two IRF830 based "Zener Followers" (one per
channel - I split the PSU frm there). These use a few "trick" bit's to
improve noise rejection and such and to limit the surge current (why will
become apparent soon)....
I follow these "Zenner Followers" with a 10 H 30 Ohm Choke (cute little thing
from Stein-Music in Germany which is custom-made on the largest standard
Potcore and can deal with about 50-100 mA each) followed by a Computergrade
(in my Sprague) 10,000uF 250V Capacitor per channel.....
All this sit's in a seperate Chassis and is connected to teh Preamp by a
Umbilical that has each stages PSU Line individually connected to the + of
the Cap....
Each stage also has a local Polypropylen PSU Cap of 4.7uF (or 10uF in the
output stage).....
I will try one of these Day's Allen Wrights Super-Reg to see if it is better
(as he asserts) but at the moment I'm extremley stretched at work....
Certainly the Difference between using the LC in the PSU and not using it
(taking the feed directly of the FET's Source) is HUGE.....
The cost for these Parts was not that high, as the Big Cap's I use all are
surplus Parts.....
The Heaters BTW get their own seprerate 120VA Toroidal X-Former and use
68,000uF 16V Cap's per channel and the 5A TO3 Adjustable Reg's from Nat Semi
(forgot the #) with plenty of Capacitance in the Preamp Chassis itself (about
10,000uF of Low-Z Cap's per channel).
Later Thorsten
Thor...@tnt-audio.com
Hi all,
> Looks like everyone's agreeing that SS regulation isn't that great for valve
> amps. So, what about using valves and stabilizer neons?
The OC Series of Stabilisers (Oc1 to 3 are known to me) ain't so bad, but
current is limited and the Impednace of the Rail remains fairly high....
So you still need the Large Cap....
>There were a good few things designed for valve circuits - neon stabilizers
>and 6AS7s controlled by a feedback circuit were de rigeur in studio and
>military applications.
I have two Gipes with most (Series) Regulators....
All of these do not really have a low source-Impedance. They use NFB (even
the Cathode Follower) to fake a low Impedance that in reality does not exist.
And the transient Response and Distortion of any Loop-Feedback using
Regulater I have ever Analysed or Measured is phatetic to say the lest...
So in either Case, there is no substitution for the real thing.... You can
work around it by using electronics to trick the Circuits into thinking that
there is a low Impedance, but for that You'd need Shunt-Regulators with about
as good a Circuit Design and partsquality at the Amplifier fed, if not
better....
And I still have problems making NFB free Circuits to have enough Bandwidth
and low enough Distortion.... Before you can close the loop "safely" (with
respect to sonics) you need something with impecable performance
(Distortion monotonic and simple, -3db open loop Bandwith >> 100kHz and so
on...).
And if you get something good enough, there is probably not enough spare Gain
left to NFB much.....
Of course I might be wrong, but so far non of the Regulation Circuits I have
seen proposed are in my opinion good enough.... Even my favourite Design
(Allen Wrights "Super-reg" which is a currentsource fed shuntreg) is way too
slow for my taste.....
I'm currently considering having very High value Chokes custom-made (> 50H)
and also massive Polyprop Capacitors, so that I can keep the L*C Value near
0.02 (10H*2000uF - which is the Minimum Combination I have found to work well
- so 50H 100 Ohm Chokes and 400uF should work well....) in order to loose the
'lytic cap's I'm using now.....
Sheldon D. Stokes
In article <6manjb$g58$1...@news.enterprise.net>, "Russ Sadd"
<sad...@aston.ac.uk> wrote:
> Looks like everyone's agreeing that SS regulation isn't that great for valve
> amps. So, what about using valves and stabilizer neons? There were a good
> few things designed for valve circuits - neon stabilizers and 6AS7s
> controlled by a feedback circuit were de rigeur in studio and military
> applications.
I wouldn't say that regulation is a bad thing for tube circuits (maybe
it's misguided for an amplifier), you just can't drop an LM317 at the end
of your b+ supply and hope for an improvement. I use Zener Referenced
pass transistors (either NPN or N channel FETs) as a way to drop the
ripple at the begining of the b+ filter chain. I then use big caps and
small resistors after the regulator to further smooth the b+ and get rid
of transistor noise. With the transistor regulator, the resistors between
caps can be much smaller (due to the much much lower ripple), and the
supply can have a lower impedance.
this is the best sounding supply topology that I have tried, of course
best sounding is pretty subjective.
Sheldon
--
"...from Genghis Kahn to the Fuller brush man, they're just a bunch
of loosers like me" -Dave VanRonk
Remove SPAM_BE_GONE from my address to reply to me.
Henry Pasternack
Thor...@tnt-audio.com wrote:
: All of these do not really have a low source-Impedance. They use NFB (even
: the Cathode Follower) to fake a low Impedance that in reality does not exist.
Nonsense. Impedance is impedance.
John Diamantis
My experience follows yours, although, at least for low level circuits,
I've found the mosfets to provide superior sound to the bipolars. If the
mosfets in use don't have gate to source protection zeners on chip, they
need to be installed to prevent destroying the device on power up/down.
If you've got the case real estate to use a tube pass element, the sound
improves even further.
One trick I've found to really lower the zener noise, is *not* to hang a
cap directly across the zener, but to use a series resistance of about
10Kohms, shunted by the cap to ground, with the base/gate/grid connected
at this junction. The reason for this, is that the dynamic impedance of
the zener is low, and by building this out with the series resistance,
the filter fo is lower, hence, any noise will be lower as well. This
setup doesn't seem to effect regulation in any way.
John Diamantis
John Diamantis
Several years ago, I played with some shunt regulator circuits, with a
(heaven forbid) solid state phono preamp and line stage. While they
sounded good, they were tricky to set up right, and to be honest, the
old tried and true shunt zener with pass regulator sounded better.
Actually, cascading two stages of this sounded even better in the phono
stage.
I bring this up to see if anyone else has played with shunt reg's, and
what they experienced with them
John D
Sheldon D. Stokes
> One trick I've found to really lower the zener noise, is *not* to hang a
> cap directly across the zener, but to use a series resistance of about
> 10Kohms, shunted by the cap to ground, with the base/gate/grid connected
> at this junction. The reason for this, is that the dynamic impedance of
> the zener is low, and by building this out with the series resistance,
> the filter fo is lower, hence, any noise will be lower as well. This
> setup doesn't seem to effect regulation in any way.
That's a good idea. I put a film cap across the string to stabilize it
and to try to absorb much of that diode noise. Adding a series resitor in
theory could degrade the regulation, but if you are using a fet or tube,
the gate/grid power draw is very nearly constant, so a low resistance like
10K shouldn't effect the regulation much at all.
The problem I have conceptually with the tube pass element is that I
regulate with a pass transistor to get rid of ripple but also to create a
low Z supply. Using a tube pass element will have a lot higher Z than a
transistor. If you use a beefy tube, it's still pretty low though.
I will say after trying LM317's and zener referenced op-amp regulator
circuits on the b+, that a simple transistor pass element is BY FAR less
noisey and to my ears better sounding.
Thor...@tnt-audio.com
Hi there,
> : All of these do not really have a low source-Impedance. They use NFB (even
> : the Cathode Follower) to fake a low Impedance that in reality does not exist.
>
> Nonsense. Impedance is impedance.
Is it? A 1 Ohm Resistor (or for AC a 1 Ohm impedance at the specified
Frequency) can sink current.... You will get 1 V per Ampere injected....
How much current does your regulator sink?
So is then the Impedance of your Reg what you claim it is?
As long as you look at things with steady-state signals they are fine....
Push them with Impulse Signals (say Inject White Noise into the Rail with a
number of notched out frequency areas and observe what comes back....) and
they show what they are made of.... <unprintable>
That of course is only half the story, as we also have that rather nasty
Feedbackloop to deal with.... I say way too much trouble for much too little
gain....
Henry Pasternack
Thor...@tnt-audio.com wrote:
: > Nonsense. Impedance is impedance.
: Is it? A 1 Ohm Resistor (or for AC a 1 Ohm impedance at the specified
: Frequency) can sink current.... You will get 1 V per Ampere injected....
: How much current does your regulator sink?
As much as it's designed to sink.
: So is then the Impedance of your Reg what you claim it is?
Its impedance is whatever its impedance is.
: As long as you look at things with steady-state signals they are fine....
: Push them with Impulse Signals (say Inject White Noise into the Rail with a
: number of notched out frequency areas and observe what comes back....) and
: they show what they are made of.... <unprintable>
Design a regulator wrong and it will perform badly.
: That of course is only half the story, as we also have that rather nasty
: Feedbackloop to deal with.... I say way too much trouble for much too little
: gain....
That darned feedback. If only it had never been invented.
I hate lame discussions.
André Jute
Tube regulation for any voltage or current required by any tube one
wants to use is real easy once one stops thinking it is complicated and
baffling oneself by seeking a correspondingly complicated solution.
You simply use another tube exactly the same (or bigger) as a shunt
regulator. The circuit is simple (it is like PSE tubes but with the
output transformer primary between the two plates). I have done it with
845 for 845, and with triode coupled EL-34 (because they were handy) for
6SN7. Works a treat, a hell of a lot better than a whole chandelier of
pissweak little regulators all sitting there grudgingly giving up their
few miserable milliamps each and making an intolerable noise claiming
they're doing you a favour. (And, for those who insist on complication,
they can fiddle to their hearts' content with +ve and -ve supplies for
the fils, or compensating voltages on the grid, tailoring the bandwidth
and noise figures ever more tightly.)
For those really obsessed with the sound of a tube type or brand, it is
also possible to use any tube as a rectifier: every audio tube is just a
diode with some extra wires chucked in, so once you have taken care of
cathode stripping by dropping some AC on the fils and attended to
whatever other gradients between those undesirable wires might affect
operation, yo, bingo, rectifiers!-- you need two for fullwave--with
exactly the same sonic signature as the one in your signal chain that
you love so much, of course shunt regulated by yet another precisely the
same. Emotional overload!
Just think, you could have four WE 300B on each of your monoblocs and
still just produce 6W per channel. Hell, there is no reason to have a
different driver. Add two more WE 300B per channel as an input and a
driver for 2W sensitivity out of CD, or as many as you need to multiply
up a phono signal, and have yourself a WE 300B chandelier. Hmm, not such
a wild idea at all...
The same applies to small-signal tubes.
Andre "The Lateral Thinker" Jute
Nil carborandum NIHbimbo!
--
Andre Jute an...@indigo.ie COMMUNICATION JUTE
--see our pages for music lovers, writers and audiophiles at
http://indigo.ie/~andre/ComJuteF1.html
dang...@earthlink.net
On Thu, 18 Jun 1998 18:40:38 GMT, Thor...@tnt-audio.com wrote:
>Hi there,
>
>> : All of these do not really have a low source-Impedance. They use NFB (even
>> : the Cathode Follower) to fake a low Impedance that in reality does not exist.
>>
>> Nonsense. Impedance is impedance.
>
>Is it? A 1 Ohm Resistor (or for AC a 1 Ohm impedance at the specified
>Frequency) can sink current.... You will get 1 V per Ampere injected....
>
>How much current does your regulator sink?
>
>So is then the Impedance of your Reg what you claim it is?
>
>As long as you look at things with steady-state signals they are fine....
>Push them with Impulse Signals (say Inject White Noise into the Rail with a
>number of notched out frequency areas and observe what comes back....) and
>they show what they are made of.... <unprintable>
>
>That of course is only half the story, as we also have that rather nasty
>Feedbackloop to deal with.... I say way too much trouble for much too little
>gain....
>
>Later Thorsten
>
>-----== Posted via Deja News, The Leader in Internet Discussion ==-----
>http://www.dejanews.com/ Now offering spam-free web-based newsreading
Greetings Thorsten,
Thanks for spilling about your PS design. Sounds great about the
custom poly caps and chokes.
Any thoughts about using function generator waveforms and impulse
signals to drive a resistor loaded PS regulator node, and using
spectrum analysis of the PS regulator node to tune the regulation
system?
Sounds like you have already tried this or something similar.
Regards,
Dangerdave
Kirk R. Patton
John Diamantis wrote:
>
> Sheldon D. Stokes wrote:
> >
> > In article <6manjb$g58$1...@news.enterprise.net>, "Russ Sadd"
> > <sad...@aston.ac.uk> wrote:
> >
> > > Looks like everyone's agreeing that SS regulation isn't that great for valve
> > > amps. So, what about using valves and stabilizer neons? There were a good
> > > few things designed for valve circuits - neon stabilizers and 6AS7s
> > > controlled by a feedback circuit were de rigeur in studio and military
> > > applications.
> >
> > I wouldn't say that regulation is a bad thing for tube circuits (maybe
> > it's misguided for an amplifier), you just can't drop an LM317 at the end
> > of your b+ supply and hope for an improvement. I use Zener Referenced
> > pass transistors (either NPN or N channel FETs) as a way to drop the
> > ripple at the begining of the b+ filter chain. I then use big caps and
> > small resistors after the regulator to further smooth the b+ and get rid
> > of transistor noise. With the transistor regulator, the resistors between
> > caps can be much smaller (due to the much much lower ripple), and the
> > supply can have a lower impedance.
> >
> > this is the best sounding supply topology that I have tried, of course
> > best sounding is pretty subjective.
> >
> >
> > --
> > "...from Genghis Kahn to the Fuller brush man, they're just a bunch
> > of loosers like me" -Dave VanRonk
> >
> > Remove SPAM_BE_GONE from my address to reply to me.
>
> I've found the mosfets to provide superior sound to the bipolars. If the
> mosfets in use don't have gate to source protection zeners on chip, they
> need to be installed to prevent destroying the device on power up/down.
> If you've got the case real estate to use a tube pass element, the sound
> improves even further.
>
> cap directly across the zener, but to use a series resistance of about
> 10Kohms, shunted by the cap to ground, with the base/gate/grid connected
> at this junction. The reason for this, is that the dynamic impedance of
> the zener is low, and by building this out with the series resistance,
> the filter fo is lower, hence, any noise will be lower as well. This
> setup doesn't seem to effect regulation in any way.
This makes some sense, just make sure that the input impedance of the
reference amp (or gain of the pass-transistor) is much higher than
resistor used . . . 10K probably works well for FET's, but seems a
little high to me for a bipolar stage. This will also affect temperature
stability slightly, but absolute voltage isn't a big deal for tube
circuits anyway.
Regards,
Kirk Patton
Kirk R. Patton
Henry Pasternack wrote:
> Nah, three-terminal IC regulators are bad because they are noisy and
> have very low bandwidths. There's no reason a well-designed regulator
> should sound bad. It's not clear, though, that the added complexity
> is justified. As well, some circuits may sound better with unregulated
> supplies for reasons that aren't objectively obvious.
>
Uh huh. The main jist of the matter is that an active voltage regulator
is an AC amplifier, and must be optimized as one . . . it requires the
same level of frequency and transient analysis that the actual
amplifying stage does. And really, the extra effort of true regulation
isn't worth it in the vast majority of tube stages . . .
A true voltage regulator is a circuit which delivers a linear output
impedance and constant voltage regardless of input voltage and output
current. In my book, this means "active" regulation, yes, with
feedback. "Pseudo-regulators" such as the common
zener-on-the-base-of-a-high-gain-transistor thing, and simple zener and
single gas-tube shunt circuits are voltage-stabilizing circuits, not
regulators. And the passive networks on their outputs are decoupling
networks, not load regulation.
Now this wasn't intended as a value judgement . . . but different
approaches are used to design decoupling networks than are used to
design stable active regulators. And this thread has had an appalling
mix of the logic of decoupling networks applied to the design of active
regulators.
Put simply, when simple voltage-stabilizing techniques are used (no
feedback) then the decoupling network may be designed very much like an
unregulated filter network, the main differences being the noise spectra
that must be filtered out, and the low-frequency behaviour. And with
these topologies, the "more is better" approach to capacitance . . . uh
. . . kinda works . . . . huge oversimplification, but not along
incorrect lines.
But applying this thinking to feedback circuits, including
three-terminal regulators, is asking for trouble, and really defeats any
benefits of active regulation in the first place. And if you want real,
specific information on what values to use with which IC's, and what
stability problems might be encountered, then jeeeez, guys, read the
damn applications information published by the manufacturers!!!! They
literally spoon-feed you all of this stuff . . . . specifically, no one
mentioned the application of capacitance on the "adjust" pin and how it
greatly improves ripple rejection . . . but it's widely published
information . . .
Series-versus-shunt regulation is an arguement for another thread . . .
but someone inquired about limitations . . . . and with shunt regulation
it's the efficiency that goes out the window . . . . it is possible to
build hybrid circuits with controlled pass- and shunt- devices to get
the best of both worlds . . . .
Regards,
Kirk Patton
Thor...@tnt-audio.com
Hi all,
> My experience follows yours, although, at least for low level circuits,
> I've found the mosfets to provide superior sound to the bipolars. If the
> mosfets in use don't have gate to source protection zeners on chip, they
> need to be installed to prevent destroying the device on power up/down.
Yup.... It also might be a good Idea to put a suitable current-limiting
resistance into the Drain Linem if You'r gonna use notable Levels of
Capacitance behind it....
> If you've got the case real estate to use a tube pass element, the sound
> improves even further.
Never tried it.
> One trick I've found to really lower the zener noise, is *not* to hang a
> cap directly across the zener, but to use a series resistance of about
> 10Kohms, shunted by the cap to ground, with the base/gate/grid connected
> at this junction.
This is exactly my solution. It does help. Still, I found using the
additional LC Decoupling I describe in another Post (10H into 10,000uF) to
provide VERY NOTABLE improvements in sound....
Thor...@tnt-audio.com
Hi all,
> Thanks for spilling about your PS design. Sounds great about the
> custom poly caps and chokes.
If you know sources where you can afford the "custom stuff" yes....
> Any thoughts about using function generator waveforms and impulse
> signals to drive a resistor loaded PS regulator node, and using
> spectrum analysis of the PS regulator node to tune the regulation
> system?
>
> Sounds like you have already tried this or something similar.
I tried that in PSpice only.... I really have not got good enought a
laboratry at my hands to do the practical research. That said, I discussed my
PSpice results with a few people who do have the facilities and to at least
one of the Guy's the whole story was very old news.... He also seems to use a
Reg of high internal Impedance and plenty of custom Polyprops for the preamp
he is developing (a solid state Design BTW)....
If you have a nice Generator that can produce a few different Wave-Forms and
Frequencies simultaniously (like the AP1 does I believe) and inject that
Mix....
Be prepared for some interesting results.... All Designs with large amounts
of NFB and using an OPA604 (also tried LF351, 741 and LM324 - all way worse)
as simulated Op-Amp (including a lot of the Circuits parasitics as well) did
look rather horrible....
Switching to a simulated LM6181 (a high Speed Video Op-Amp) resulted in
instability in all cases as the Pass/Shunt-Elements where too slow.....
Again, that is only PSpice derived and the reality may look slightly
different, but I feel that the sonic differences observed corelate somewhat
with the results of the Simulations....
BTW, the simple Zener-Follower was the best under dynamic conditions, but
still woefully short of the Zenner-Follower with LC Decoupling....
dang...@earthlink.net
>Now this wasn't intended as a value judgement . . . but different
>approaches are used to design decoupling networks than are used to
>design stable active regulators. And this thread has had an appalling
>mix of the logic of decoupling networks applied to the design of active
>regulators.
>
>Put simply, when simple voltage-stabilizing techniques are used (no
>feedback) then the decoupling network may be designed very much like an
>unregulated filter network, the main differences being the noise spectra
>that must be filtered out, and the low-frequency behaviour. And with
>these topologies, the "more is better" approach to capacitance . . . uh
>. . . kinda works . . . . huge oversimplification, but not along
>incorrect lines.
>
>But applying this thinking to feedback circuits, including
>three-terminal regulators, is asking for trouble, and really defeats any
>benefits of active regulation in the first place. And if you want real,
>specific information on what values to use with which IC's, and what
>stability problems might be encountered, then jeeeez, guys, read the
>damn applications information published by the manufacturers!!!! They
>literally spoon-feed you all of this stuff . . . . specifically, no one
>mentioned the application of capacitance on the "adjust" pin and how it
>greatly improves ripple rejection . . . but it's widely published
>information . . .
When I was contributing to those same app notes for chip houses,
primary motivation was lowest cost implementation to form a complete
system solution. It goes something like this: IC chip + cheap ext.
components = a complete function that works. If total pkg. cheaper
than competition, I win.
The idea of recommending external components, like giant caps, that
cost 10x or 100x more than the chip itself, is not one that I would be
caught dead representing to chip corporations, due to competitive
pressure to reduce total pkg. cost of chip-based designs. Chip
makers are in business to portray their products as "complete"
solutions to circuit designs problems, not highlight chip
deficiencies by emphasizing use of very expensive external components.
Re-read the same IC regulator app. notes you refer to, particularly
the paragraph about "increasing the Cout will lower the output
impedance and reduce the ripple". Of course. The part they don't
emphasize is just how much Cout it takes for these chips to work well
for high end audio.
There are good reason for experimenters to use an "appaling mix" of
NFB regulation and swamp-out filter caps.
1) They may already own one or more vintage pieces, that use some form
of regulation that can't be easily changed. (hardwired onto a PCB for
example) But it is relatively easy to add-on big caps to improve
sound quality **. Look at all the creative ways experimenters have
found to stuff more caps into vintage design. (You really got to
admire the ingenuity involved.
** (Uncle Ned, I hope I've learned something from you,
because right here I'll state -- don't try this at home if
you have a tube audio component that uses a vacuum
tube rectifier. You may blow the tube rectifier)
2) For scratch designs, it's easy to use IC regulator chips to set a
DC level, and use big Cout to improve the audio noise and distortion.
This way, you can get a pretty good sounding supply, w/o the complete
anlysis required to design a feedback circuit. Resources to do
complete NFB designs , like SPICE, spec analyzers, network analyzers,
etc., aren't available to many RATs.
3) And even for advanced designers with good resources and corporate
level support, who go deep into NFB and controls theory, a "more Cout
is better" design paradigm is still revealed after they do
sophisticated analyses that include the spurious spectral components
generated by NFB and their deleterious effect upon sound quality.
Look at the lengths Krell, Levinson, and other SS NFB design houses go
to in their power supplies, including the giant Couts they use.
An interesting example is the BAT PAK. Faced w/ the design delima of
"it takes so much Cout to make good sound that my packaging now looks
like shit", BAT offers an add on box, the "BAT PAK" which is a
separate enclosure completly stuffed full of giant PS capacitors.
I think the good opportunity here, is to think the problem through,
and ask the question; why would high end companies like Krell,
Levinson, BAT, put all those big Cout caps in their products?
I bet they read the IC regulator app notes too :-)
Regards,
Dangerdave
dang...@earthlink.net
On Fri, 19 Jun 1998 14:10:04 GMT, dang...@earthlink.net wrote:
>On Fri, 19 Jun 1998 00:03:03 -0500, "Kirk R. Patton"
><kirkp...@sprintmail.com> wrote:
>
>>Now this wasn't intended as a value judgement . . .
BTW Kirk,
I should have said in the previous reply, thanks very much for your
response to my post.
I really appreciate the interest shown, and I think the contributions
from you and all the others, have helped to examine the issues
involved w/ good PS design.
I have really had fun responding to the reply posts, and I think it's
been a good opp for all us audio freaks to sharpen our debating
skills, and push the design limits a little farther [ and maybe
improve some RAT stereos by hanging on some of those big nasty Cout
Caps :-) ]
RATs to the fore -- Keep Experimenting!
Regards,
Dangerdave
fdeu...@blackrock.com
In article <6mbssl$4...@bmtlh10.bnr.ca>,
h...@nortel.ca wrote:
>
> Thor...@tnt-audio.com wrote:
> : As long as you look at things with steady-state signals they are fine....
> : Push them with Impulse Signals (say Inject White Noise into the Rail with a
> : number of notched out frequency areas and observe what comes back....) and
> : they show what they are made of.... <unprintable>
>
This is the crux of this whole thread: a regulator needs to be properly
designed to source and/or sink current at a rate appropriate to the circuit
at hand.
While Thorsten points out an important problem with many (poorly applied, and
therefore poorly designed for the problem at-hand) regulators, that of
inability to sink current at the same rate as the regulator can source
current, it must be recognized that this is a design problem, and not a fault
of voltage regulators in general: you must design your regulator for the load
you wish to drive; if your load is reactive, a regulator designed to drive
purely resistive loads will make you quite unhappy.
It is a well-known fact that followers (cathode or emitter) can have great
difficulty driving reactive loads; this is because a simple follower can not
sink current as well as it can source current. To understand why this is so,
we look at the follower circuit: it is a pass element controlled by the
voltage difference between its base/grid and its emitter/cathode. While this
pass device is capable of conducting very well in the forward direction, the
best we can do in the reverse direction is to cut off the pass device. In
other words, the pass device is incapable of sinking current on its own; when
the voltage at the output node is above the reference voltage of our simple
follower, it must be bled through some path other than the pass element.
With a purely resistive load, this situation never arises: vary the load
resistance all you like, at any frequency, but the voltage at the output of
the regulator will never go above the reference voltage. However, reactive
loads are an entirely different story: with a reactive load, the voltage at
the output node of the regulator may well go above the reference voltage, and
thus "lock up" a simple follower pass element. When the output node voltage
goes above the reference voltage, there are only two paths in a simple
follower regulator for it to dissipate via: either through the resistive part
of the load impedance, or through the feedback network of the regulator.
Depending on the relative magnitudes and angles of the impedances involved,
the results of this situation will vary from acceptable to downright disaster
(burt-up components in the feedback network, pass elements destroyed from
excessive reverse bias, etc).
(A side note on driving power tubes with a cathode follower to lower Zout of
the driver: the grid of a power tube is readily seen to be nothing more than
a highly capacitive load....)
However, what must be recognized here is that the fault lies not within the
concept of a regulated power supply, but rather with the faulty application
of the concept: using a simple follower to drive or voltage regulate a highly
reactive load is a Bad Idea. Similarly, blaming the concept of feedback
because your feedback network went up in smoke, or your supply/amp
oscillated, is folly; as Henry pointed out, proper design solves all these
problems.
As with any technology, there are benefits and drawbacks to voltage regulation
and feedback; however, the things presented thus far in this thread aren't
representative of those drawbacks.
-frank
Sheldon D. Stokes
> But applying this thinking to feedback circuits, including
> three-terminal regulators, is asking for trouble, and really defeats any
> benefits of active regulation in the first place. And if you want real,
> specific information on what values to use with which IC's, and what
> stability problems might be encountered, then jeeeez, guys, read the
> damn applications information published by the manufacturers!!!! They
> literally spoon-feed you all of this stuff . . . . specifically, no one
> mentioned the application of capacitance on the "adjust" pin and how it
> greatly improves ripple rejection . . . but it's widely published
> information . . .
The data sheet is in the back of "Art of Electronics", it's quite useful.
The sheet says : Adding a cap from the adjust pin to ground increases the
ripple rejection from about 60 dB to about 85 dB, but that's dependant on
input/output voltage differential. The data sheet also says that if you
are using an output cap of more than about 25 uF then you should put a
diode from the output pin to the adjust pin to prevent the adjust cap from
discharging back through the regulator.
Looking at the performance graphs, the LM317 output impedance isn't flat
at all, it has a minimun of aabout 500 Hz, and steadily rises up to about
a MHz. It's about 0.005 ohms at 500 Hz, and about 1 ohm at 1 mHz (this is
with a 10 uF adjust cap). With the adjust cap the impedance goes up again
below 500 Hz. Without this cap, the impedance is higher than with this
cap, but stays relitively flat below 500 Hz. It's always higher than with
the cap. I'd say that even though it's output impedance changes, it's
quite low compared to other circuit elements.
Here's what the data sheet say about output caps:
Although the LM117 is stable with no output capacitors, like
any feedback circuit, certain values of external capacitance
can cause excessive ringing. This occurs with values be-tween
500 pF and 5000 pF. A 1 湩 solid tantalum (or 25 湩 aluminum
electrolytic) on the output swamps this effect and
insures stability. Any increase of the load capacitance larger
than 10 湩 will merely improve the loop stability and output
impedance.
From what I read above, it sounds likemore caps are better. This agrees
with my listening experiences.
The datasheet is available at: http://www.national.com/ds/LM/LM117.pdf
Sheldon D. Stokes
In article <1dau5bl.pqt...@ts01-07.cork.indigo.ie>,
an...@indigo.ie (=?ISO-8859-1?Q?Andr=E9_Jute?=) wrote:
> Tube regulation for any voltage or current required by any tube one
> wants to use is real easy once one stops thinking it is complicated and
> baffling oneself by seeking a correspondingly complicated solution.
Audio circuits are really simple in the spectrum of all possible
circuits. But not exploring all the aspects of a circuit is not really a
complete design, but rather kludging something together. The best audio
products are engineered using electronics and math, it's the tools of the
trade, and valuable ones they are.
> You simply use another tube exactly the same (or bigger) as a shunt
> regulator. The circuit is simple (it is like PSE tubes but with the
> output transformer primary between the two plates). I have done it with
> 845 for 845, and with triode coupled EL-34 (because they were handy) for
> 6SN7. Works a treat, a hell of a lot better than a whole chandelier of
> pissweak little regulators all sitting there grudgingly giving up their
> few miserable milliamps each and making an intolerable noise claiming
> they're doing you a favour. (And, for those who insist on complication,
> they can fiddle to their hearts' content with +ve and -ve supplies for
> the fils, or compensating voltages on the grid, tailoring the bandwidth
> and noise figures ever more tightly.)
So your saying that you can get a continuous 1.5 amps out of a EL34 for
years like you can with an LM317? I'm not arguing that SS regulators
should be used everywhere, but I think you are missing the value of these
things. I use them all the time for filaments.
> For those really obsessed with the sound of a tube type or brand, it is
> also possible to use any tube as a rectifier: every audio tube is just a
> diode with some extra wires chucked in, so once you have taken care of
> cathode stripping by dropping some AC on the fils and attended to
> whatever other gradients between those undesirable wires might affect
> operation, yo, bingo, rectifiers!-- you need two for fullwave--with
> exactly the same sonic signature as the one in your signal chain that
> you love so much, of course shunt regulated by yet another precisely the
> same. Emotional overload!
But real tube regulators have HUGE cathodes to supply the needed
instantanious current. Tubes designed for other uses really make
sub-optimum rectifiers. And I'd argue that rectifiers are the thing that
tubes do the worst.
> The same applies to small-signal tubes.
It would take a boatload of 12AX7's to to what a 5AR4 does.
Pascal Sternis
Hello,
in the other thread (about capacitor value after the reg) I was
evocating the possibility of a PSpice simulation depending on the
availability of a LM317 model.
The problem with Pspice simulation for IC regulators is that there are
no good models readily available or affordable (more about this later)
to my knowledge.
I have lurked the net for LM317 (or similar) models for example, nor at
T.I, Motorola, Harris, Temic, Analog Devices, N.S, Linear technology you
will find models for series IC regulators. (quite frankly, I realized
later that today who would care for such a model ?)
The only models I have found are free samples at
http://commerce1.ba.best.com/~symetry1/ or commercial ones from this
same company.
1) The problem is these are Spice, not Pspice models and I do not have
skill or time to adapt it to Pspice.
2) The other problem is that they have commercial models, but they are
each at $199 and I can't quitefrankly afford this :-(
3) the third problem with IC manufacturer and spice models is that it is
a part of the manufacturer's know-how on this sensitive topic (LDO
voltage reg.) and they don't want to divulgate their internal IC
topology this way. They sometimes supply downgraded (I mean built with
only a few basic functions) spice models very far from the reality.
4) The complexity of a series voltage reg. (or a macrocell) is about 40
transistors and it takes about 2 days on a PC computer to achieve
relevant simlations.
So we could turn in simulating by assembling an OP amp and a power trans
:
1) simulating a series regulator with a 741; 324 op amp seems unfair
cause these op amps are really crappy and you cannot fulfill the
requirements for a large open-loop bandwidth , low serial ouput
impedance (on the AOP point of view), neither with a bipolar pass
transistor nor with a MOSfet (because of the Cgs/Cgd).
2) When having a look at the diagrams given in regulators data-sheet,
there are region of instability depending on the output current, and
minimum and maximum value of the series resistance of the output cap.
(this depends also on the operating temperature and on the frequency :
higher the frequency ,lower the series resistance)
3) I don't know whether you used MOS or bipolar transistor in your
simulations, but the disadvantage of MOS is that the gain is higher than
a bipolar when the phase shift is 180 deg. so it's more difficult to
compensate.
4) I once tried to design a simple high voltage regulator on my own, and
performed Pspice simulations. I was so discouraged by the bad results I
got and by the potential growing complexity that I gave up. But I think
this is due to my insufficient background on this topic rather than the
regulator principle itself. As said Mr Pasternack, it's not that simple
;-)
5) It is true that the application notes say also that compensation has
to be designed for a special frequency range. they don't emphasize on
audio ;-)
Please notice that I am not especially advocating IC regulators vs big
capacitors or so, I just wanted to point out that spice simulations with
inadequate models may lead to strange results.
However, I should say that designs including both regulators AND big
output capacitors surspise me, and I wonder if the sound would be so
different if the regulators were removed. have you heard sonic
differences with and without the MOS transistors in your PS design ? It
seems so well furnished with chokes and capacitors that I wonder if the
silicon components are trully mandatory ;-)
Best Regards
Pascal
--------
please, remove ".NOSPAM" from adress before replying
Henry Pasternack
=?ISO-8859-1?Q?Andr=E9_Jute?= (an...@indigo.ie) wrote:
: Tube regulation for any voltage or current required by any tube one
: wants to use is real easy once one stops thinking it is complicated and
: baffling oneself by seeking a correspondingly complicated solution.
: You simply use another tube exactly the same (or bigger) as a shunt
: regulator. The circuit is simple (it is like PSE tubes but with the
: output transformer primary between the two plates). I have done it with
: 845 for 845, and with triode coupled EL-34 (because they were handy) for
: 6SN7. Works a treat, a hell of a lot better than a whole chandelier of
: pissweak little regulators all sitting there grudgingly giving up their
: few miserable milliamps each and making an intolerable noise claiming
: they're doing you a favour.
Proper wideband shunt regulator design is as tricky as series
regulator design. The big disadvantage of shunt regulators is
that they dissipate more power than the circuit they feed. They
are inefficient, then, and this is a serious issue in any circuit
that draws a lot of power.
As for the notion of using regular audio tubes as rectifiers,
this may be lateral thinking, but it's also a bad idea. A rectifier
is optimized for rectifier service. An audio tube is not. It'll
work in a pinch, but I don't really see the point.
John Diamantis
Kirk R. Patton wrote:
>
> John Diamantis wrote:
> >
> > Sheldon D. Stokes wrote:
> > >
> > > In article <6manjb$g58$1...@news.enterprise.net>, "Russ Sadd"
> > > <sad...@aston.ac.uk> wrote:
> > >
> > > > Looks like everyone's agreeing that SS regulation isn't that great for valve
> > > > amps. So, what about using valves and stabilizer neons? There were a good
> > > > few things designed for valve circuits - neon stabilizers and 6AS7s
> > > > controlled by a feedback circuit were de rigeur in studio and military
> > > > applications.
> > >
> > > I wouldn't say that regulation is a bad thing for tube circuits (maybe
> > > it's misguided for an amplifier), you just can't drop an LM317 at the end
> > > of your b+ supply and hope for an improvement. I use Zener Referenced
> > > pass transistors (either NPN or N channel FETs) as a way to drop the
> > > ripple at the begining of the b+ filter chain. I then use big caps and
> > > small resistors after the regulator to further smooth the b+ and get rid
> > > of transistor noise. With the transistor regulator, the resistors between
> > > caps can be much smaller (due to the much much lower ripple), and the
> > > supply can have a lower impedance.
> > >
> > > this is the best sounding supply topology that I have tried, of course
> > > best sounding is pretty subjective.
> > >
> > >
> > > --
> > > "...from Genghis Kahn to the Fuller brush man, they're just a bunch
> > > of loosers like me" -Dave VanRonk
> > >
> > > Remove SPAM_BE_GONE from my address to reply to me.
> >
> > I've found the mosfets to provide superior sound to the bipolars. If the
> > mosfets in use don't have gate to source protection zeners on chip, they
> > need to be installed to prevent destroying the device on power up/down.
> > improves even further.
> >
> > cap directly across the zener, but to use a series resistance of about
> > 10Kohms, shunted by the cap to ground, with the base/gate/grid connected
> > the zener is low, and by building this out with the series resistance,
> > the filter fo is lower, hence, any noise will be lower as well. This
> > setup doesn't seem to effect regulation in any way.
>
> This makes some sense, just make sure that the input impedance of the
> reference amp (or gain of the pass-transistor) is much higher than
> resistor used . . . 10K probably works well for FET's, but seems a
> little high to me for a bipolar stage. This will also affect temperature
> stability slightly, but absolute voltage isn't a big deal for tube
> circuits anyway.
>
> Regards,
>
> Kirk Patton
The 10K value is not a problem with FET's or Tubes. For bipolar, you can
either lower the value some, or better still, use a darlington topology.
I've tried this all of these schemes in both tube and solid state
equipment with success. While high speed opamp circuits require the
absolute lowest of PS impedances to work properly, most other audio type
circuits involving discrete solid state, or tube circuitry work fine
with a simple zener/pass element regulator (ne: stabilizer).
John Diamantis
Russ Sadd
As Henry has already pointed out, a good valve regulated power supply will
require a good few Watts to run it. For example, you're going to have extra
current being dissipated in the circuit to run the feedback components
(amplifier valve and neon reference), and there's always going to be a
voltage drop. In other words, you're going to need a larger PSU than you'd
normally need.
However, these do provide a very stable voltage supply, unachievable in any
other way save by having huge (and this means bloody expensive) capacitors.
I'm sure that a 10F cap would make a hell of a filter, but now we're in the
territory of at least as much (if not more) money being spent on the filter
than you would on the extra components and allowing for extra Watts in the
PSU end. Surely, if we're talking Class A and single-ended circuits, we're
not exactly being efficient with power?
In small signal cases, I think I'll stick to a stabilizer neon - they're
simple, and very effective. Why worry about the impedance of the circuit
when the entire point is that the neon will have a varying impedance to
provide a stable voltage?
When it comes down to it, there are surely only a few instances where you'll
be after a very stable power supply. After all, in the case of:
- single-ended amplifiers, so you'll probably want to filter things heavily
- 'professional' use, so you'll have a rack mounted regulated PSU in any
case
- Class A push-pull, so you don't need to worry too much
- Class AB and B, where you'll REALLY have to worry about the PSU. Leaks
sound great on a regulated PSU.
- screen grids, where a neon will do nicely
- small signals, where again a neon would be ideal
I wonder if this comes down to the big old question of how much power should
you be using in a valve amplifier, and to what class of operation you're
willing to run things in. Traditional British designs favour class A
push-pull, which makes them less sensitive to fluctuations - they also tend
to be lower-powered. Traditional American designs favour class AB, which
makes them rather more sensitive to power supply fluctuations, and they tend
to be higher-powered designs. Seems like regulation is more practical with
the lower-powered British designs, but impractical for the American ones,
where they'd probably benefit more.
marcel
wasn't this a tube-group? ,am i missing out on something??
what is all talk about fet's lm317 (only the thought about them) ,they
sure as hell don't belong in my tube-amps or tube -dac's
Tube rectifiers ,chokes ,paper/oil caps do the work for me ,it seems
outdated ,but every single time i try another approach it doesn't deliver
the goods sonically
--
<tubes for real emotion>
LIEUTENDGN <lieut...@aol.com> schreef in artikel
<199806180605...@ladder01.news.aol.com>...
> It also heavily depends on whether the FAB is a " class 10" or "class
100" for
> the production.
>
> Class 10 usually yields lower noise.
>
> SONY R&D Lab in Japan has the lowest nosie FET on their back burner.
>
> I am surprised they never border to use what they have from their R&D lab
for
> the consurmer products.
>
> May be they are saving it for the last minute shoot out ???
>
> Their comptetitors are trying to do a reverse engineering on the FET with
no
> luck.
>
> My suggestion : hold your breath.
>
>
>
dang...@earthlink.net
On Fri, 19 Jun 1998 11:25:51 -0400, stokes@SPAM_BE_GONEexis.net
(Sheldon D. Stokes) wrote:
>In article <1dau5bl.pqt...@ts01-07.cork.indigo.ie>,
>an...@indigo.ie (=?ISO-8859-1?Q?Andr=E9_Jute?=) wrote:
-- SNIP --
>But real tube regulators have HUGE cathodes to supply the needed
>instantanious current. Tubes designed for other uses really make
>sub-optimum rectifiers. And I'd argue that rectifiers are the thing that
>tubes do the worst.
>
I think it's a good arguement, Sheldon.
I'll jump in for (3) comments.
1) The tube can't supply any more current than the emissivity of the
cathode coating will support, without causing damage/failure over
sustained operation. The cathode coating will simply be destroyed and
the tube will become NONOP if you exceed the steady state emissivity
limit for very long.
This is a very good reliability arguement for using SS rectification
over tube rectification (even gas tube rectification), for all audio
applications.
2) The packaging footprint and package volume requirement of current
supply tube(s) can become so large, for medium to high currents apps,
the tubes become physically difficult to package into audio enclosures
that are compatible with audio enclosure standard practice.
A worse package size vs. performance trade-off than big Cout caps,
IMHO.
3) You generate a reliability paradox if you use these things for
medium and high current apps. The more you use, the more current you
can supply, the shorter the MTBF (mean time between failure).
Argueably true for tube amps in general, but a non-favorable
performance vs. reliability trade-off IMHO, when tubes are used to
supply B+ current, for these apps.
Summary: Rectification is one area where you really might want to
think twice about bucking orthodox theory, and sub'ing tubes for SS,
particularly for medium and high current apps.
Thanks for the good info, Sheldon.
"Hey, don't get excited, it's just a physics thing "
somebody, somewhere, sometime
Regards,
Dangerdave
Bob C
It is bad enough we have to read his bullshit, now he has to do this!!!!!
BC
>Andre
>--
>Andre Jute an...@indigo.ie COMMUNICATION JUTE
>--see our pages for music lovers, writers and audiophiles at
>http://indigo.ie/~andre/ComJuteF1.html
>
>Andre Jute an...@indigo.ie COMMUNICATION JUTE
>--see our pages for music lovers, writers and audiophiles at
>http://indigo.ie/~andre/ComJuteF1.html Andre Jute an...@indigo.ie
>COMMUNICATION JUTE
>--see our pages for music lovers, writers and audiophiles at
>http://indigo.ie/~andre/ComJuteF1.html Andre Jute an...@indigo.ie
>COMMUNICATION JUTE
>--see our pages for music lovers, writers and audiophiles at
>http://indigo.ie/~andre/ComJuteF1.html Andre Jute an...@indigo.ie
>COMMUNICATION JUTE
>--see our pages for music lovers, writers and audiophiles at
>http://indigo.ie/~andre/ComJuteF1.html Andre Jute an...@indigo.ie
>COMMUNICATION JUTE
>--see our pages for music lovers, writers and audiophiles at
>http://indigo.ie/~andre/ComJuteF1.html Andre Jute an...@indigo.ie
>COMMUNICATION JUTE
>--see our pages for music lovers, writers and audiophiles at
>http://indigo.ie/~andre/ComJuteF1.html Andre Jute an...@indigo.ie
>COMMUNICATION JUTE
>--see our pages for music lovers, writers and audiophiles at
>http://indigo.ie/~andre/ComJuteF1.html Andre Jute an...@indigo.ie
>COMMUNICATION JUTE
>--see our pages for music lovers, writers and audiophiles at
>http://indigo.ie/~andre/ComJuteF1.html Andre Jute an...@indigo.ie
>COMMUNICATION JUTE
>--see our pages for music lovers, writers and audiophiles at
>http://indigo.ie/~andre/ComJuteF1.html Andre Jute an...@indigo.ie
>COMMUNICATION JUTE
>--see our pages for music lovers, writers and audiophiles at
>http://indigo.ie/~andre/ComJuteF1.html Andre Jute an...@indigo.ie
>COMMUNICATION JUTE
>--see our pages for music lovers, writers and audiophiles at
>http://indigo.ie/~andre/ComJuteF1.html Andre Jute an...@indigo.ie
>COMMUNICATION JUTE
>--see our pages for music lovers, writers and audiophiles at
>http://indigo.ie/~andre/ComJuteF1.html Andre Jute an...@indigo.ie
>COMMUNICATION JUTE
>--see our pages for music lovers, writers and audiophiles at
>http://indigo.ie/~andre/ComJuteF1.html Andre Jute an...@indigo.ie
>COMMUNICATION JUTE
>--see our pages for music lovers, writers and audiophiles at
>http://indigo.ie/~andre/ComJuteF1.html Andre Jute an...@indigo.ie
André Jute
Sheldon D. Stokes <stokes@SPAM_BE_GONEexis.net> wrote:
> In article <1dau5bl.pqt...@ts01-07.cork.indigo.ie>,
> an...@indigo.ie (=?ISO-8859-1?Q?Andr=E9_Jute?=) wrote:
>
> > Tube regulation for any voltage or current required by any tube one
> > wants to use is real easy once one stops thinking it is complicated and
> > baffling oneself by seeking a correspondingly complicated solution.
>
> Audio circuits are really simple in the spectrum of all possible
> circuits. But not exploring all the aspects of a circuit is not really a
> complete design, but rather kludging something together. The best audio
> products are engineered using electronics and math, it's the tools of the
> trade, and valuable ones they are.
Surely it is a matter of perspective whether using a tube instead of
silicon is a "kludge" or a matter of taste. An engineer's "kludge" is a
cultured person's "aesthetic sensibility". The requirement for
"electronics and math" appears in both cases.
> > You simply use another tube exactly the same (or bigger) as a shunt
> > regulator. The circuit is simple (it is like PSE tubes but with the
> > output transformer primary between the two plates). I have done it with
> > 845 for 845, and with triode coupled EL-34 (because they were handy) for
> > 6SN7. Works a treat, a hell of a lot better than a whole chandelier of
> > pissweak little regulators all sitting there grudgingly giving up their
> > few miserable milliamps each and making an intolerable noise claiming
> > they're doing you a favour. (And, for those who insist on complication,
> > they can fiddle to their hearts' content with +ve and -ve supplies for
> > the fils, or compensating voltages on the grid, tailoring the bandwidth
> > and noise figures ever more tightly.)
>
> So your saying that you can get a continuous 1.5 amps out of a EL34 for
> years like you can with an LM317? I'm not arguing that SS regulators
> should be used everywhere, but I think you are missing the value of these
> things. I use them all the time for filaments.
Giving you the benefit of the doubt, Sheldon, you must have read my post
in a hell of a hurry, or you would have noticed that my case above deals
regulating plate voltages. Who drives 6SN7 plates at 1.5A? Anyone you
know? They must have inherited a whole boatload of 6SN7...
> > For those really obsessed with the sound of a tube type or brand, it is
> > also possible to use any tube as a rectifier: every audio tube is just a
> > diode with some extra wires chucked in, so once you have taken care of
> > cathode stripping by dropping some AC on the fils and attended to
> > whatever other gradients between those undesirable wires might affect
> > operation, yo, bingo, rectifiers!-- you need two for fullwave--with
> > exactly the same sonic signature as the one in your signal chain that
> > you love so much, of course shunt regulated by yet another precisely the
> > same. Emotional overload!
>
> But real tube regulators have HUGE cathodes to supply the needed
> instantanious current. Tubes designed for other uses really make
> sub-optimum rectifiers. And I'd argue that rectifiers are the thing that
> tubes do the worst.
I was sending up the obsessives, Sheldon, not suggesting a topology for
anyone. The point is that it is possible to do what I suggest. Whether
it is desirable from either an engineering or a sonic viewpoint others
may decide for themselves. I must say, the net is a very poor medium for
humour.
> > The same applies to small-signal tubes.
>
> It would take a boatload of 12AX7's to to what a 5AR4 does.
Huh? Who's suggesting that one replaces GZ34 with 12AX7? I'm talking
about shunt regulating low-current supplies, like shunt regulating one
ECC83 with another.
> Sheldon
André Jute
Russ Sadd <sad...@aston.ac.uk> wrote:
> As Henry has already pointed out, a good valve regulated power supply will
> require a good few Watts to run it. For example, you're going to have extra
> current being dissipated in the circuit to run the feedback components
> (amplifier valve and neon reference), and there's always going to be a
> voltage drop. In other words, you're going to need a larger PSU than you'd
> normally need.
Actually, Russ, I was writing about possibilities, because someone
(perhaps John Diamantis?) suggested we discuss shunt regulation; I
wasn't advocating anything. Personally I go to DC fils only when I have
to; since I use only CD, that is not always by any means, even on input
and driver tubes, and I regulate fil voltages only as a last resort. As
for regulated HV, I tried it as described but it "refeens" the music too
much for my taste even with tubes. Too often with regulated HV it sounds
like the amp has its pinky in the air and is waiting to finish chewing a
tiny bite of crustless cucumber sandwich before it speaks...
>
> However, these do provide a very stable voltage supply, unachievable in any
> other way save by having huge (and this means bloody expensive) capacitors.
> I'm sure that a 10F cap would make a hell of a filter, but now we're in the
> territory of at least as much (if not more) money being spent on the filter
> than you would on the extra components and allowing for extra Watts in the
> PSU end. Surely, if we're talking Class A and single-ended circuits, we're
> not exactly being efficient with power?
I don't know if this is an engineering or social argument you're making,
but in either case the aficionadoes of class A and SE are at the least
wasteful of resources and in some "green" eyes we might be social
criminals. But I am resigned to going to hell... Where does one get a
10F 630Vdc cap, then? <G>
>
> In small signal cases, I think I'll stick to a stabilizer neon - they're
> simple, and very effective. Why worry about the impedance of the circuit
> when the entire point is that the neon will have a varying impedance to
> provide a stable voltage?
>
> When it comes down to it, there are surely only a few instances where you'll
> be after a very stable power supply. After all, in the case of:
>
> - single-ended amplifiers, so you'll probably want to filter things heavily
> - 'professional' use, so you'll have a rack mounted regulated PSU in any
> case
> - Class A push-pull, so you don't need to worry too much
> - Class AB and B, where you'll REALLY have to worry about the PSU. Leaks
> sound great on a regulated PSU.
> - screen grids, where a neon will do nicely
> - small signals, where again a neon would be ideal
Thanks for the list. I would say that the one case where a regulated
supply seems almost invariably to be beneficial is in phono amps--your
small signal case. Every good phono amp I have ever heard seems to have
had a regulated supply.
> I wonder if this comes down to the big old question of how much power should
> you be using in a valve amplifier, and to what class of operation you're
> willing to run things in. Traditional British designs favour class A
> push-pull, which makes them less sensitive to fluctuations - they also tend
> to be lower-powered. Traditional American designs favour class AB, which
> makes them rather more sensitive to power supply fluctuations, and they tend
> to be higher-powered designs. Seems like regulation is more practical with
> the lower-powered British designs, but impractical for the American ones,
> where they'd probably benefit more.
This is a good point in the historical perspective. Though when I
started in tube amps at the beginning of the decade, I got some old
radar engineers in nursing homes in England as my gurus, guys who spoke
of "young Peter Walker", and several of them gave me as key tutorial
material heavy metal power supply schematics, especially from AR. To
them the essence of amp design was in the power supply and the signal
section of the amp was a sort of afterthought. Not one of them ever
volunteered the thought that it was only the inefficiency of modern
speakers that created the need for what I thought of as monstrous amps;
when I brought it up, they thought the monstrous amp was a small price
to pay for the vastly better frequency response and extension of even
cheap modern drivers. Their outlook had changed over the years. (Mind
you, the Sugden A21 is probably today, as for the past twenty years,
the leading-edge British amp, and, in comparison with the Nelson Pass
commercial designs, it fits your model perfectly.)
> Russ
> ---------------
> E-Mail: sad...@aston.ac.uk
> Web Pages: http://homepages.enterprise.net/icedragon/new/index.htm
Andre
Russ Sadd
Seems like some of the standard misconceptions are back. Dear me, sometimes
this channel is just brimmin' over with modernism!
1. You're not meant to run a valve harder than it's designed for. But apart
from that, it is a very old trick to use a diode-strapped triode as a
rectifier. This was done with surplus power triodes in the lean War and
Post-War years - although the practice predates it. In the 1930s, for
example, when people were considering standardizing radio valve design to
three multi-functional valves, strapping triodes as diodes was intended from
the outset. Small, low-power radios also used diode-strapped triodes as
their rectifiers. It can be done alright, it just isn't quite as efficient
as using a properly designed rectifier - but it is a very elegant idea!
2. Sure, SS might drop less power than a valve, and not require heaters, but
SS doesn't last forever, either. And there is that HF switching noise
problem!
3. The packaging footprint of a rectifier valve isn't exactly great: space
for 1 octal base and a little bit more heater current from the transformer.
Dear me, when you think about the sheer size of a high-end valve amp, one
extra valve won't make that much difference!
Russ Sadd
André's quite right in saying that you don't need that large a valve to work
as your series pass element (or whatever we call them now). After all, you
only need to pass enough power to feed the audio circuit. Why use a 6AS7
when you only need maybe 15mA? In this case, it makes much more sense to use
a triode-strapped 6V6 or the like.
Shunt regulating half a valve with another? This is an old, old trick -
what's so shocking about it?
Russ
--
dang...@earthlink.net
On Sat, 20 Jun 1998 18:35:50 +0100, "Russ Sadd" <sad...@aston.ac.uk>
wrote:
>Seems like some of the standard misconceptions are back. Dear me, sometimes
>this channel is just brimmin' over with modernism!
>
Are you reporting to us from some kind of time machine in the 19th
century? :-)
>1. You're not meant to run a valve harder than it's designed for. But apart
>from that, it is a very old trick to use a diode-strapped triode as a
>rectifier. This was done with surplus power triodes in the lean War and
>Post-War years - although the practice predates it. In the 1930s, for
>example, when people were considering standardizing radio valve design to
>three multi-functional valves, strapping triodes as diodes was intended from
>the outset. Small, low-power radios also used diode-strapped triodes as
>their rectifiers. It can be done alright, it just isn't quite as efficient
>as using a properly designed rectifier - but it is a very elegant idea!
>
Oh come on, Russ. Why is it elegant to misuse a triode as a diode?
When the tube designers wanted proper rectifiers, they designed them
-- diodes.
Using a triode strapped as a diode just means you now have a triode
cathode emission element instead of an emission element that was
designed for rectification service -- as found in the diode.
Designing a tube to do both is just a cheap way to save on tooling
cost.
In any event, the problem is that when you run a hard vacuum rectifer
as it's designed to run, it is ORDERS OF MAGNITUDE less reliable than
solid state and ORDERS OF MAGNITUDE weaker in current output.
Unless you deliberately plan to use the current starvation of a hard
vacuum rectifer to generate a specific distortion quality, like
musical instrument amplifiers do, vacuum rectifiers are weak and
unreliable compared to SS.
>2. Sure, SS might drop less power than a valve, and not require heaters, but
>SS doesn't last forever, either. And there is that HF switching noise
>problem!
SS lasts ORDERS OF MAGNITUDE longer.
For HF noise, put in a few 5nF bypass caps across the diodes, use low
noise HEXFRETS, and use a large Cout filter, which you can do easily
w/ SS and not with Tubes, and gain the unequalled sonic benefits the
large Cout provides
>Dear me, when you think about the sheer size of a high-end valve amp, one
>extra valve won't make that much difference!
For medium to high current apps, it's NOT ONE extra valve it is A
FOREST OF VALVES. For a good PS w/ perhaps 3 or 4 regulated voltage
outputs, times (2) 1/2 wavers or (1) full waver per voltage supplied,
times the paralleled tube multiplier to get the current outputs up to
anywhere near SS levels, you would be talking about a FOREST of vacuum
tubes to even come close to the current capability of tiny diodes that
cost a minute fraction of what tube rectifiers cost.
Sure, you can run an amp off a single GZ-34, GZ-37, 82, 83, etc., but
w/ a few hundred mills of current capacity, you will never get the
kind of clean specs and clean sound you can achieve by a high quality
brick wall supply w/ loads of transient current capability.
For amp restorers, vintage equipment collectors, vintage design retro
enthusiasts, etc., this argument is not really relavent. The older
designs are great just as they are, and have a great sound for what
they are. I have a bunch of older amps and preamps and I run some of
them in original configuration -- it's fun. But for new designs, why
starve the B+ when excellent technical solutions are now available at
a resonable price?
If those same vintage amp designers were alive today, would they be
starving the B+ by using hard vacuum rectifiers? Check out what the
contemporary high end designers are doing at Krell, BAT, Levinson.
The neat thing about tube amp design to me, is that I am free to chose
the very best technolgies to solve each design problem of the amp at
the highest level. For example, despite decades of device
development, BJT, FET, IGBT, GAS, etc., the Triode still remains the
most linear active amplifying element available. So for a good linear
amp, in go the Triodes. Why not be open minded and choose the best
technology available for each circuit design issue in each part of the
amp, regardless of the era in which these technologies were developed?
The difference in Philosophy here is:
Do you want a vintage/retro design?
or
Do you want the best sounding design?
Regards,
Dangerdave
>
>
Russ Sadd
dang...@earthlink.net wrote in message ...
>>Seems like some of the standard misconceptions are back. Dear me,
sometimes
>>this channel is just brimmin' over with modernism!
>>
>Are you reporting to us from some kind of time machine in the 19th
>century? :-)
Modernism is the basis of all modern culture, it is how modern culture
defines itself. One of the most fundamental aspects this is the
over-emphasis of progress. New things are better, simply because they are
new. Today's model is infinitely better than the previous one. The extreme
of this is known as 'hyperdevelopmentalism'. All of us on RAT should be
familiar with this - valves were the dominant technology of the 20th
Century, only fading from popular view in the '70s, and are still used in
high-end audio, broadcasting, Russian military hardware, microwave ovens,
televisions, and UHF applications - yet they're spoken about and thought of
as a 'dead' technology!
Because of this, a good few ideas are thought of as being much more recent
than they are: the 'Japanese production system' concepts were being
implemented in the '30s, as were high-definition colour television and
voice-controlled elevators.
>Oh come on, Russ. Why is it elegant to misuse a triode as a diode?
>When the tube designers wanted proper rectifiers, they designed them
>-- diodes.
My point, was that when you don't need much current, and have cheap valves
available to you, you can easily build something that uses only one or two
types of valve. This kind of standardization is wonderfully functional. For
example, the RK34 RF double-triode was used as both rectifier and push-pull
output to give an 8W amplifier. The benefits here were that (1) cost was
considerably reduced and (2) only two types of valve were used [Camm, F. J.
(1957) 'Amplifiers: Design and Construction', Newnes: London]. Or perhaps a
little mains radio using two 6SL7s?
>In any event, the problem is that when you run a hard vacuum rectifer
>as it's designed to run, it is ORDERS OF MAGNITUDE less reliable than
>solid state and ORDERS OF MAGNITUDE weaker in current output.
How are we defining reliability? And why the "orders of magnitude"?
>For medium to high current apps, it's NOT ONE extra valve it is A
>FOREST OF VALVES. For a good PS w/ perhaps 3 or 4 regulated voltage
>outputs, times (2) 1/2 wavers or (1) full waver per voltage supplied,
>times the paralleled tube multiplier to get the current outputs up to
>anywhere near SS levels, you would be talking about a FOREST of vacuum
>tubes to even come close to the current capability of tiny diodes that
>cost a minute fraction of what tube rectifiers cost.
Let's see - for a valve regulated power supply stage, I'd use a single
rectifier, or parallel two to save money, add a power valve, a small
amplifying valve, and a neon, and use neons for the individual low-power
stages. The solid-state alternatives would take up a good bit of room, too.
If I wanted to, I could use a stabilovolt to get all the taps I needed off a
single unit - a nice, elegant solution stabilovolts were designed to give.
Hardly a forest of valves!
>Sure, you can run an amp off a single GZ-34, GZ-37, 82, 83, etc., but
>w/ a few hundred mills of current capacity, you will never get the
>kind of clean specs and clean sound you can achieve by a high quality
>brick wall supply w/ loads of transient current capability.
These valves were intended for standard small-equipment currents. For
example, the U54 can handle 250mA, 500V, and handle peaks of 1.25 KV and 1.5
A. To use higher voltages, as you might in amateur transmission equipment,
you'd use a xenon half-wave like the GXU50, which can handle 1.75 KV at
250mA, and peaks of 5.2 KV at 1.5 A. For high-current applications, you
could use an ESU872 half-wave mercury to give a mean current of 1.25A, and
voltages in the KV range. If you needed lower voltages, but high current,
then a 68530 full-wave tungar would give you 30V at 6A. Clearly, valves can
deliver transient current and voltage. If you need a rectifier to give more
current than a standard hard vacuum can deliver, you use xenon or mercury
rectifiers. Horses for courses, same as always.
>If those same vintage amp designers were alive today, would they be
>starving the B+ by using hard vacuum rectifiers? Check out what the
>contemporary high end designers are doing at Krell, BAT, Levinson.
I still don't see how the HT is being starved here. You design the power
supply to provide what the amplifier needs, and to provide it in as clean a
way as possible. Contemporary high-end designers might have flair, but
remember that they don't have, say, 40 years of experience designing
laboratory, military or studio-grade equipment using valves or have
postgraduate qualifications in valve electronics under their belts. Funnily
enough, people did know what they were doing. I certainly wonder what D. T.
N. Williamson would make of Mark Levinson - not that his 'Amplifiers and
Superlatives' article springs to mind!
>The neat thing about tube amp design to me, is that I am free to chose
>the very best technolgies to solve each design problem of the amp at
>the highest level. For example, despite decades of device
>development, BJT, FET, IGBT, GAS, etc., the Triode still remains the
>most linear active amplifying element available. So for a good linear
>amp, in go the Triodes. Why not be open minded and choose the best
>technology available for each circuit design issue in each part of the
>amp, regardless of the era in which these technologies were developed?
The point is, that modern solid-state electronics are designed around
high-current, low-voltage devices. Valve-era electronics were designed
around high-voltage, low-current devices. Neon stabilizers, for example, are
rather preferable to zener diodes. Oh, and valves (aren't pentodes and beam
tetrodes more linear than triodes?) have had several more decades of
development to their name than the relatively-recent solid-state devices
like transistors of various ilks. People weren't too happy with transistors
originally - they were simply too unreliable! Besides, solid-state
rectifiers are about the same age as valves, and have had an equal amount of
development behind them. Surely it makes you wonder why people went for
rectifier valves rather than Westinghouse rectifiers?
>The difference in Philosophy here is:
>Do you want a vintage/retro design?
>or do you want the best sounding design?
The 'best sound' is a completely subjective thing - a designer will follow
his own tastes - some people like bass, some people want an airy treble. I'd
go for the old British values of 'good practice' and go for as transparent a
sound as possible. 'Good practice' here means using established circuits
that are known to be reliable, running components well within their design
parameters, and then adding that extra 50% safety margin wherever possible.
Russ
TubeGarden
Just wandered in. The conceptualization and philosophy are interesting, but I
would like some feedback on the homebrew amp I am using. Walking through my B+
power supply:
200V-0-200V 870mA Hammond filament transformer
5V4G or GT or GA (variety on the upside)
and
pair of 6DM4As (dull on the downside)
forming a full wave bridge rectifier
pair of 22uF 280Vac GE oil caps (in series, with 470K resistors in parallel to
balance voltages)
pair of 5H 75 mA chokes in parallel
32uF 570Vac GE oil cap
four each 10H 100mA chokes (one to each stage)
32uF 570Vac oil can cap (one each for each EF86 as triode voltage amplifier and
two each for each SV83 or 8CW5 or XL86 (all as triode) power amplifier)
I tried both discrete and bridge SS rectification, with and without low value
resistors on the AC ends, but the tubes sound better. Also tried other (5U4G,
5AW4, 5Y3...) tubes.
I do use SS rectifier bridge for amp filaments, with 62,000 uF 16V cap as
filter and low (0.5 ohm 10W) resistance on the AC legs to SMEAR (Subdue
Malignant Electrical Artifacts Realtime) the hash.
Rectifier tubes use seperate AC filament supplies.
Power supply is box on floor, amp is box on table. What could I do to improve
the sound?
Al B^}
Al Marcy
TubeG...@aol.com
dang...@earthlink.net
On Sat, 20 Jun 1998 22:54:36 +0100, "Russ Sadd" <sad...@aston.ac.uk>
wrote:
>
>dang...@earthlink.net wrote in message ...
>>>Seems like some of the standard misconceptions are back. Dear me,
>sometimes
>>>this channel is just brimmin' over with modernism!
>>>
>>Are you reporting to us from some kind of time machine in the 19th
>>century? :-)
>
>
>Modernism is the basis of all modern culture, it is how modern culture
>defines itself. One of the most fundamental aspects this is the
>over-emphasis of progress. New things are better, simply because they are
>new.
Sometimes new things are better, simply because they are better :-)
>Today's model is infinitely better than the previous one. The extreme
>of this is known as 'hyperdevelopmentalism'.
Is that the same as hyperventilation? I get that sometimes when I
think about tube rectifiers in my brick wall supplies :-)
> All of us on RAT should be
>familiar with this - valves were the dominant technology of the 20th
>Century, only fading from popular view in the '70s, and are still used in
>high-end audio, broadcasting, Russian military hardware, microwave ovens,
>televisions, and UHF applications - yet they're spoken about and thought of
>as a 'dead' technology!
>
>Because of this, a good few ideas are thought of as being much more recent
>than they are: the 'Japanese production system' concepts were being
>implemented in the '30s, as were high-definition colour television and
>voice-controlled elevators.
Yes, I agree. The way history rewrites and re-accredits technological
achievements can be pretty dissapointing.
>>Oh come on, Russ. Why is it elegant to misuse a triode as a diode?
>>When the tube designers wanted proper rectifiers, they designed them
>>-- diodes.
>
>
>My point, was that when you don't need much current, and have cheap valves
>available to you, you can easily build something that uses only one or two
>types of valve. This kind of standardization is wonderfully functional. For
>example, the RK34 RF double-triode was used as both rectifier and push-pull
>output to give an 8W amplifier. The benefits here were that (1) cost was
>considerably reduced and (2) only two types of valve were used [Camm, F. J.
>(1957) 'Amplifiers: Design and Construction', Newnes: London]. Or perhaps a
>little mains radio using two 6SL7s?
Yes, I agree. It was a cheap solution to use the parts available,
like the surplus triodes used as rectifiers, as you stated. Cheap
isn't necessarily bad. It's just when cheap overpowers sound quality
in audio design that my "hyperdevelopmentalistic" alarm bells start
ringing.
>>In any event, the problem is that when you run a hard vacuum rectifer
>>as it's designed to run, it is ORDERS OF MAGNITUDE less reliable than
>>solid state and ORDERS OF MAGNITUDE weaker in current output.
>
>
>How are we defining reliability? And why the "orders of magnitude"?
Oh boy, Russ. I think we may be coming at tube audio design from
completly different backgrounds.
They way I am defining reliability is "how long can you use it before
it breaks". Industry standard terms define this kind of thing by MTBF
(mean time between failure) or time/temperature reliability curves, or
other more technical descriptors, but still, it's how long you can use
the thing that defines it's "useful lifetime".
The "orders of magnitude" term is often used by people who have to
predict the reliable lifetime of electronic components for a living.
The reason they use it as an emphatic type of adjectival phrase, is
that often, the relative reliability of competing technologies they
measure is a real close call, like just a few 10's of precent. When
something has a reliablitiy lifetime that is 10X or 100X or 1000X
different, it is a no brainer for them to pick a winner, so they don't
have to do an extensive differential analysis, they can just "declare"
a winner by saying "it's orders of magnitude" better.
You may already know this, and if you do, I appologize for being
pendantic, but each time something is ten time greater, that is one
order of magnitude, in our base 10 math system. So for example, if
something is 1000x larger, that is the same as three orders of
magnitude larger. The "orders of magnitude" term is also used by
engineers sometimes as a kind of verbal short-hand to indicate factors
of 10 larger or smaller.
>>For medium to high current apps, it's NOT ONE extra valve it is A
>>FOREST OF VALVES. For a good PS w/ perhaps 3 or 4 regulated voltage
>>outputs, times (2) 1/2 wavers or (1) full waver per voltage supplied,
>>times the paralleled tube multiplier to get the current outputs up to
>>anywhere near SS levels, you would be talking about a FOREST of vacuum
>>tubes to even come close to the current capability of tiny diodes that
>>cost a minute fraction of what tube rectifiers cost.
>
>
>Let's see - for a valve regulated power supply stage, I'd use a single
>rectifier, or parallel two to save money, add a power valve, a small
>amplifying valve, and a neon, and use neons for the individual low-power
>stages. The solid-state alternatives would take up a good bit of room, too.
>If I wanted to, I could use a stabilovolt to get all the taps I needed off a
>single unit - a nice, elegant solution stabilovolts were designed to give.
>Hardly a forest of valves!
Sure, that will work. And I'm sure you have designed some very fine
amplifers using that approach.
It's just that if you want to explore the brick wall types of
supplies, and the kinds of acoustic properties they afford, you need
high capacity rectifiers to charge them without destroying the
rectifiers.
Note to Uncle Ned: I really am trying to emphasize this!
It's not my fault if someone actually listens to me, hangs
10,000uF on their Mark III's, and blows the GZ-34's you
sold them!
>>Sure, you can run an amp off a single GZ-34, GZ-37, 82, 83, etc., but
>>w/ a few hundred mills of current capacity, you will never get the
>>kind of clean specs and clean sound you can achieve by a high quality
>>brick wall supply w/ loads of transient current capability.
>
>
>These valves were intended for standard small-equipment currents. For
>example, the U54 can handle 250mA, 500V, and handle peaks of 1.25 KV and 1.5
>A. To use higher voltages, as you might in amateur transmission equipment,
>you'd use a xenon half-wave like the GXU50, which can handle 1.75 KV at
>250mA, and peaks of 5.2 KV at 1.5 A. For high-current applications, you
>could use an ESU872 half-wave mercury to give a mean current of 1.25A, and
>voltages in the KV range. If you needed lower voltages, but high current,
>then a 68530 full-wave tungar would give you 30V at 6A. Clearly, valves can
>deliver transient current and voltage. If you need a rectifier to give more
>current than a standard hard vacuum can deliver, you use xenon or mercury
>rectifiers. Horses for courses, same as always.
>
I agree. I breadboarded many types of high current gas rectifiers
for PS applications (cost me a bunch, unfortunately). Problem is,
they are really physically large, really expensive, and less reliable,
compared to SS diodes. Although the gas rectifiers do look beautiful
in operation. (See my prior post for some emotional gushing by me on
this topic)
>>If those same vintage amp designers were alive today, would they be
>>starving the B+ by using hard vacuum rectifiers? Check out what the
>>contemporary high end designers are doing at Krell, BAT, Levinson.
>
>
>I still don't see how the HT is being starved here. You design the power
>supply to provide what the amplifier needs, and to provide it in as clean a
>way as possible.
True. But w/ a big capacity filter, the impedance is pretty low, and
if the tubes surge into the low impedance, they can fail. This can
happen of course during power-up charging, and also if you have a high
power output stage event that significantly diminishes the V on the
output C's during hard transients, loudspeaker EMF kickbacks, etc.
EMF Kickbacks. Sounds like something congressional :-)
>Contemporary high-end designers might have flair, but
>remember that they don't have, say, 40 years of experience designing
>laboratory, military or studio-grade equipment using valves or have
>postgraduate qualifications in valve electronics under their belts.
I think in most cases, you are probably right.
>Funnily
>enough, people did know what they were doing. I certainly wonder what D. T.
>N. Williamson would make of Mark Levinson - not that his 'Amplifiers and
>Superlatives' article springs to mind!
Well, if Mark came out in his speedo, with his Oakley shades, hopped
onto his mountain bike, to grab a few DVDs for the kids, and biked
through McDonalds for a fast food dinner, I think N. Willamson would
be totally floored. :-)
>>The neat thing about tube amp design to me, is that I am free to chose
>>the very best technolgies to solve each design problem of the amp at
>>the highest level. For example, despite decades of device
>>development, BJT, FET, IGBT, GAS, etc., the Triode still remains the
>>most linear active amplifying element available. So for a good linear
>>amp, in go the Triodes. Why not be open minded and choose the best
>>technology available for each circuit design issue in each part of the
>>amp, regardless of the era in which these technologies were developed?
>
>
>The point is, that modern solid-state electronics are designed around
>high-current, low-voltage devices. Valve-era electronics were designed
>around high-voltage, low-current devices.
I agree that is basically true. They are a few exceptions, but not
really that many in this primarily digital era.
>Neon stabilizers, for example, are
>rather preferable to zener diodes. Oh, and valves
>(aren't pentodes and beam tetrodes more linear than triodes?)
Well, this is an important one to emphasize, Russ. They are definetly
not. The wonderful attribute of Triodes for linear amplifier designs,
is they are more linear than the other tube types you mentioned, and
they are more linear than all the more modern active devices invented
in the many decades since the first triode design. In fact, in what I
consider to be a supreme kind of irony (I used to design solid state
devices for a living and it's all based upon the most "modern" physics
theories) due to the simple "cylindrical" elements used in the very
first most primitive triode designs, they are in some cases more
linear the the triodes designed in the 40's up until today!
This is one case where older is better.
>have had several more decades of
>development to their name than the relatively-recent solid-state devices
>like transistors of various ilks. People weren't too happy with transistors
>originally - they were simply too unreliable! Besides, solid-state
>rectifiers are about the same age as valves, and have had an equal amount of
>development behind them. Surely it makes you wonder why people went for
>rectifier valves rather than Westinghouse rectifiers?
>
Well, there are some older metallic rectifiers that are pretty
vintage. But for new designs, you might want to try some of the
HEXFRETS which use techniques to minimize charge stored in the diode
juction. They don't make too much noise..
>The difference in Philosophy here is:
>>Do you want a vintage/retro design?
>>or do you want the best sounding design?
>
>The 'best sound' is a completely subjective thing - a designer will follow
>his own tastes - some people like bass, some people want an airy treble. I'd
>go for the old British values of 'good practice' and go for as transparent a
>sound as possible. '
Good for you!
>Good practice' here means using established circuits
>that are known to be reliable, running components well within their design
>parameters, and then adding that extra 50% safety margin wherever possible.
>
You can't argue with that.
Thanks for your comments, Russ. I enjoyed responding to this thread.
Regards,
Dangerdave
Dave Cigna
<dang...@earthlink.net> wrote:
>Oh come on, Russ. Why is it elegant to misuse a triode as a diode?
>When the tube designers wanted proper rectifiers, they designed them
>-- diodes.
I can think of a few cases where they might be considered elegant.
The diode line of a 6336B looks as good as most common full wave
rectifiers (high perveance, i.e. low voltage drop.) Obviously, you're
not going to replace a 5R4 in a high voltage supply, but, since they're
indirectly heated and have separate cathodes, a pair of them could be
used as a bridge rectifier in a choke input supply for 250V @ 500mA
without exceeding the tubes ratings. I might call that elegant, though
I suppose I'd never call it cheap.
How about using a high perveance miniature twin triode like a 5687 for
full wave rectification in a bias supply? You'd have to be careful to
limit peak currents, but I'd call it elegant, maybe even cute. If you
want real elegance, how about using one side of the tube as a half wave
rectifier and the other side as a regulator.
>In any event, the problem is that when you run a hard vacuum rectifer
>as it's designed to run, it is ORDERS OF MAGNITUDE less reliable than
>solid state and ORDERS OF MAGNITUDE weaker in current output.
Perhaps, but the same could be said when comparing SS to tubes when
employed as amplifiers. [Truth is, I've never had a 5U4 fail. I have
fried quite a few silicon diodes by hanging too much capacitance off of
them, but I suppose that's my own fault. :)]
>The difference in Philosophy here is:
>
>Do you want a vintage/retro design?
>or
>Do you want the best sounding design?
I don't think the two choices are mutually exclusive.
In the end, I think there are probably many good applications for SS
diodes in audio. The real problem here -- as in any other debate of
this kind -- is in clinging too zealously to one side of the fence
or the other.
-- Dave
Sheldon D. Stokes
In article <Euvpo...@boss.cs.ohiou.edu>, ci...@helios.phy.OhioU.Edu
(Dave Cigna) wrote:
> I can think of a few cases where they might be considered elegant.
> The diode line of a 6336B looks as good as most common full wave
> rectifiers (high perveance, i.e. low voltage drop.) Obviously, you're
> not going to replace a 5R4 in a high voltage supply, but, since they're
> indirectly heated and have separate cathodes, a pair of them could be
> used as a bridge rectifier in a choke input supply for 250V @ 500mA
> without exceeding the tubes ratings. I might call that elegant, though
> I suppose I'd never call it cheap.
>
> How about using a high perveance miniature twin triode like a 5687 for
> full wave rectification in a bias supply? You'd have to be careful to
> limit peak currents, but I'd call it elegant, maybe even cute. If you
> want real elegance, how about using one side of the tube as a half wave
> rectifier and the other side as a regulator.
The peak currents in a tube rectifier are the limiting factors. So you
really have to look at the peak current available from a cathode of any
tube used as a rectifier. I don't know the specs for the peak emmision os
something like a 6SN7, but I'd bet it's a lot lot lower than a diode
designed for rectifier service. You can get away with using it as a
rectifier, but I'm betting you'll want to derate it a lot.
As an earlier poster said, triodes were often used as diodes during and
after the war. This is clearly a very clever and elegant solution to the
scarecity of parts. But when rectifiers were availble again, I'd bet they
went right back to them. In my opinion, using the best part for the job
is a nessesary part of an elegant design. I've seen a lot of designs that
I respect and that I think are really clever, but aren't what I'd call
elegant.
>
> >In any event, the problem is that when you run a hard vacuum rectifer
> >as it's designed to run, it is ORDERS OF MAGNITUDE less reliable than
> >solid state and ORDERS OF MAGNITUDE weaker in current output.
>
> Perhaps, but the same could be said when comparing SS to tubes when
> employed as amplifiers. [Truth is, I've never had a 5U4 fail. I have
> fried quite a few silicon diodes by hanging too much capacitance off of
> them, but I suppose that's my own fault. :)]
If you look objectively at tube rectifiers, they really don't perform as
well as other tube devices when you compare them to their solid state
counterparts. If you look at a triode gain stage compared to a single
transistor gain stage, the tube one is really close if not better (this is
just a looking at the scope discussion). But when you look at a tube
rectifier vs. solid state, and the power supply qualities that each one
has, I'd say the solid state rectifiers are nicer. I'm not saying that
you can't build nice supplies with tube rectifiers, I'm just saying that
tube rectifiers are the weakest tube application.
Sheldon
--
"...from Genghis Kahn to the Fuller brush man, they're just a bunch
of losers like me" -Dave VanRonk
Remove SPAM_BE_GONE from my e-mail to reply directly to me
André Jute
<dang...@earthlink.net> wrote:
> You may already know this, and if you do, I appologize for being
> pendantic, but each time something is ten time greater, that is one
> order of magnitude, in our base 10 math system. So for example, if
> something is 1000x larger, that is the same as three orders of
> magnitude larger. The "orders of magnitude" term is also used by
> engineers sometimes as a kind of verbal short-hand to indicate factors
> of 10 larger or smaller.
No apologies are necessary. We often say things on the assumption that
everyone shares our axiomatic precepts and vocabulary. On a newsgroup
where all kinds can wander in it is not a realistic assumption.
Here's a simple example. When talking about farads, how many zeroes
after the decimal in uF, in microfarad, in mF and in MF?* (Note the mix
of lower case and caps and never mind pf and nf.) Think back to before
you "just knew" that capacitance is commonly not expressed in integers
of Farads and you'll understand how confusing our common usage is to
someone who doesn't know. I can see that a gratuitously offered
explanation might piss off some of the kneejerk fulminators, but I would
always think more of the man who generously thought to offer an
explanation that many here who might find it useful are probably too
embarrassed to ask about.
If the RAT FAQ someone (Matt Warren?) suggested comes about, I'll
contribute a table of capacitance measurements...
Andre
*These ranges are copied out of the spec sheet for Wavetek's CR50:
200pF, 2nF, 20nF, 200nF, 2uF, 20uF, 200uF, 2mF, 20mF
Compare with the ranges from Wavetek's LCR55 spec sheet on the same
page:
200pF, 2nF, 20nF, 200nF, 2uF, 20uF, 200uF, 2000uF
Dave Cigna
Sheldon D. Stokes <stokes@SPAM_BE_GONEexis.net> wrote:
>(Dave Cigna) wrote:
>
>> I can think of a few cases where they might be considered elegant.
>> The diode line of a 6336B looks as good as most common full wave
>> rectifiers (high perveance, i.e. low voltage drop.) Obviously, you're
>> not going to replace a 5R4 in a high voltage supply, but, since they're
>> indirectly heated and have separate cathodes, a pair of them could be
>> used as a bridge rectifier in a choke input supply for 250V @ 500mA
>> without exceeding the tubes ratings. I might call that elegant, though
>> I suppose I'd never call it cheap.
>>
>> How about using a high perveance miniature twin triode like a 5687 for
>> full wave rectification in a bias supply? You'd have to be careful to
>> limit peak currents, but I'd call it elegant, maybe even cute. If you
>> want real elegance, how about using one side of the tube as a half wave
>> rectifier and the other side as a regulator.
>
>The peak currents in a tube rectifier are the limiting factors. So you
>really have to look at the peak current available from a cathode of any
>tube used as a rectifier. I don't know the specs for the peak emmision os
>something like a 6SN7, but I'd bet it's a lot lot lower than a diode
>designed for rectifier service. You can get away with using it as a
>rectifier, but I'm betting you'll want to derate it a lot.
Right, which is why I spec'ed a choke-input supply for the PS. With a
choke-input the peak current is only a little higher than the filtered DC
current. The current requirements of a bias supply is very low (unless
you go into grid current), so I think a healthy amount of series resistance
before the first cap would limit peak current enough that a 5687 could
live a long happy life. Remember that the voltage drop across a diode is
very low, so plate dissipation is not an issue.
Incidentally, I don't really expect anyone to build a power supply using
6336B's as rectifiers. If nothing else the cost and heater supply
requirements would be ridiculous. A better idea is to use a quad of damper
diodes. On the other hand, I still think a 5687 as half-wave rectifier
and regulator in a bias supply might be a workable solution for anyone
that wants to stick to an all tube design.
>If you look objectively at tube rectifiers, they really don't perform as
>well as other tube devices when you compare them to their solid state
>counterparts. If you look at a triode gain stage compared to a single
>transistor gain stage, the tube one is really close if not better (this is
>just a looking at the scope discussion). But when you look at a tube
>rectifier vs. solid state, and the power supply qualities that each one
>has, I'd say the solid state rectifiers are nicer. I'm not saying that
>you can't build nice supplies with tube rectifiers, I'm just saying that
>tube rectifiers are the weakest tube application.
Yes, a triode is intrinsically more linear than either bipolar or fet
transistors, but is this really the reason that tube audio has survived
all these years? I don't think so, at least not according to theory. The
theory seems to indicate that any device can be made arbitrarily linear
if enough NFB is applied -- at least over some frequency range and as long
as the phase shift requirements are met. Whatever the technical reason,
tube audio has survived because listeners think it sounds better. In
exactly the same way, some people find that tube rectifiers sound better.
Again let me state that I am not necessarily taking a position that SS
rectifiers are bad or that tube rectifiers are the only way to go, or
any permutation of the above. The position that I'm taking is that one
shouldn't decide what sounds good based on a technical argument. The
listening experience is a weird, mysterious and wonderful thing.
-- Dave
Russ Sadd
dang...@earthlink.net wrote in message
<8725D99DC36BDE42.A65FAEFD...@library-proxy.airnews.ne
t>...
>Yes, I agree. The way history rewrites and re-accredits technological
>achievements can be pretty dissapointing.
That's modernity for you. Which is why popular perceptions of 'new' things
are often at odds with reality.
>Oh boy, Russ. I think we may be coming at tube audio design from
>completly different backgrounds.
>
>They way I am defining reliability is "how long can you use it before
>it breaks". Industry standard terms define this kind of thing by MTBF
>(mean time between failure) or time/temperature reliability curves, or
>other more technical descriptors, but still, it's how long you can use
>the thing that defines it's "useful lifetime".
Um... If we're talking MTBF, then how come I can fire up a valve 'scope from
the '50s, and have it working perfectly? Valves last more than long enough,
and are easy to replace. Think of modern 'scopes - after a few years, the
PSU burns out and you're left with something that's just so much expensive
junk. You can't just replace a completely modular component once every few
years as you can with valves. If valves really didn't last that long, then
you wouldn't find Quad IIs 30 years on with their same valve complement.
Just think how much electronic equipment you have that hasn't survived past
the eighties! MTBF isn't exactly an exact denotation of reliability. ALPS
have a distinct tendency towards supplying pots that only last a month - and
then claim it's just a 'bad batch'. Damn good MTBF figures, though.
Remember, we do have mil. spec. valves available to us - JAN, CV and modern
Russian. If they're good enough for the Cold War, then they'll surely last a
few weeks in simple audio equipment!
> You may already know this, and if you do, I appologize for being
>pendantic, but each time something is ten time greater, that is one
>order of magnitude, in our base 10 math system. So for example, if
>something is 1000x larger, that is the same as three orders of
>magnitude larger. The "orders of magnitude" term is also used by
>engineers sometimes as a kind of verbal short-hand to indicate factors
>of 10 larger or smaller.
Ah yes, coming at it from a statistics background, I'd tend to be more
exact. How many hours operation can we expect of a standard bridge rectifier
and valve levels? And for that matter, how easy is it to replace one
compared with plugging in a valve?
>It's just that if you want to explore the brick wall types of
>supplies, and the kinds of acoustic properties they afford, you need
>high capacity rectifiers to charge them without destroying the
>rectifiers.
What do you mean by a 'brick wall' supply? You can't plug huge reservoir
caps in with valves, but then again, EU regulations don't allow it.
Something to do with placing too much strain on the transformer. None of
this stops you having an adequate filtering system, anyway. If you mean that
valves can't handle peak voltages, then I'm rather confused. Why else would
their transient capabilities be in their technical specifications? Even a
little 6SN7 can handle rather high transients!
>True. But w/ a big capacity filter, the impedance is pretty low, and
>if the tubes surge into the low impedance, they can fail. This can
>happen of course during power-up charging, and also if you have a high
>power output stage event that significantly diminishes the V on the
>output C's during hard transients, loudspeaker EMF kickbacks, etc.
Then why would vintage gear, like the Quad IIs, have such great transients?
Especially when paired with vicious speakers like the ESLs?
>Well, if Mark came out in his speedo, with his Oakley shades, hopped
>onto his mountain bike, to grab a few DVDs for the kids, and biked
>through McDonalds for a fast food dinner, I think N. Willamson would
>be totally floored. :-)
More likely he'd be charmingly sarcastic. After all, unless American
cultural norms are very different from here, our Mr. Levinson would be
acting very working class by this. :-)
>Well, this is an important one to emphasize, Russ. They are definetly
>not. The wonderful attribute of Triodes for linear amplifier designs,
>is they are more linear than the other tube types you mentioned, and
>they are more linear than all the more modern active devices invented
>in the many decades since the first triode design. In fact, in what I
>consider to be a supreme kind of irony (I used to design solid state
>devices for a living and it's all based upon the most "modern" physics
>theories) due to the simple "cylindrical" elements used in the very
>first most primitive triode designs, they are in some cases more
>linear the the triodes designed in the 40's up until today!
Well, by the '30s, people were getting much more interested in the geometry
of the electrode systems. They reckoned that valves would develop along
'cathode ray' lines, making better use of things like the space charge
effect. They weren't far off - beam tetrodes, virtual oscillator cathodes,
and magnetrons were there by the end of the decade.
Russ
Sportster Dave
dang...@earthlink.net wrote:
> In any event, the problem is that when you run a hard vacuum rectifer
> as it's designed to run, it is ORDERS OF MAGNITUDE less reliable than
> solid state and ORDERS OF MAGNITUDE weaker in current output.
True, BUT you are neglecting the fact that the power supply exists as a
part of the whole. The tubed power supply is more APPROPRIATE for a
high-end tube amp.
>
> Unless you deliberately plan to use the current starvation of a hard
> vacuum rectifer to generate a specific distortion quality, like
> musical instrument amplifiers do, vacuum rectifiers are weak and
> unreliable compared to SS.
The term 'weak' is not technically correct. It IS correct to say that
the vacuum diodes have a higher internal resistance than the SS
rectifiers. As far as unreliable is concerned; properly designed tubed
power supplies generally last thousands of hours. Generally, the
capacitors go before the tubes.
SS power supplies suffer from a number of system-related problems when
attempting to mate them with tubed equipment. First is the surge; an IH
diode warms up slowly, preventing surges. A complex SS circuit is
required to achieve the same effect; the large number of devices
inevitably drags the MTBF down lower than that of the tubed power
supply, making IT UNRELIABLE COMPARING TO THE TUBES!!!
>
> SS lasts ORDERS OF MAGNITUDE longer.
See above for my answer to this claim.
>
> For HF noise, put in a few 5nF bypass caps across the diodes, use low
> noise HEXFRETS, and use a large Cout filter, which you can do easily
> w/ SS and not with Tubes, and gain the unequalled sonic benefits the
> large Cout provides
...in a solid state amp; however, a tube amp with a choke-input filter
and IH diodes will sound better!!
>
> For medium to high current apps, it's NOT ONE extra valve it is A
> FOREST OF VALVES. For a good PS w/ perhaps 3 or 4 regulated voltage
> outputs, times (2) 1/2 wavers or (1) full waver per voltage supplied,
> times the paralleled tube multiplier to get the current outputs up to
> anywhere near SS levels, you would be talking about a FOREST of vacuum
> tubes to even come close to the current capability of tiny diodes that
> cost a minute fraction of what tube rectifiers cost.
So who's worried about cost in a hobby amp??? In a pentode amp, only
the screens need to be regulated, anyway, and this is handled with a
couple of cheap TV tubes. On lower-powered amps, a single gas tube and
a couple of passive parts will do the trick, with a fraction of the
effort required to sully the amp with SS devices. Sometimes I use a gas
diode to regulate the control grid bias supply, as well.
>
> Sure, you can run an amp off a single GZ-34, GZ-37, 82, 83, etc., but
> w/ a few hundred mills of current capacity, you will never get the
> kind of clean specs and clean sound you can achieve by a high quality
> brick wall supply w/ loads of transient current capability.
Again, the brick wall sounds good on a SS amp, but a good tube amp needs
lots of iron. Eric Barbour sez that a good amp should weigh at least
five pounds per watt!!! Use choke input main power supplies.
>
> For amp restorers, vintage equipment collectors, vintage design retro
> enthusiasts, etc., this argument is not really relavent. The older
> designs are great just as they are, and have a great sound for what
> they are. I have a bunch of older amps and preamps and I run some of
> them in original configuration -- it's fun. But for new designs, why
> starve the B+ when excellent technical solutions are now available at
> a resonable price?
OK, your SS 'stuff' against my MI-200s...any day, any time!!!
>
> If those same vintage amp designers were alive today, would they be
> starving the B+ by using hard vacuum rectifiers? Check out what the
> contemporary high end designers are doing at Krell, BAT, Levinson.
They do this for COST purposes, primarily. COST is the main thing that
these designers were interested in!! These 'vintage' designers went
pell-mell to CHEAP SS designs in the '60's!!!
>
> The neat thing about tube amp design to me, is that I am free to chose
> the very best technolgies to solve each design problem of the amp at
> the highest level. For example, despite decades of device
> development, BJT, FET, IGBT, GAS, etc., the Triode still remains the
> most linear active amplifying element available. So for a good linear
> amp, in go the Triodes. Why not be open minded and choose the best
> technology available for each circuit design issue in each part of the
> amp, regardless of the era in which these technologies were developed?
>
(oh,well, different strokes...)
Mind you, I like SS rectifiers. I do want the BEST sound, and when I
was a young DangerDave, I thought that the brick wall approach was the
way to go (a very '70's concept) Later, my designs began to reflect a
more global approach, taking a more dynamic approach. These amps and
preamps have won much popular acclaim, even though their circuits behave
VERY differently than that old '70s concept of a 'perfect circuit' with
lots of feedback and DC-to light bandwidth.
Thor...@tnt-audio.com
Hi all,
> Tube regulation for any voltage or current required by any tube one
> wants to use is real easy once one stops thinking it is complicated and
> baffling oneself by seeking a correspondingly complicated solution.
Good statement. But why use Valve Regulation.... Any Type of regualtion is as
much a pain to get to work well as is the actual Amplifier, usually more
so....
> You simply use another tube exactly the same (or bigger) as a shunt
> regulator.
Now that is a good Idea.... But You MUST feed the Shuntreg from a Current
Source and of couse, Andre' you would not use FET or BJT for that, would you
now?
BTW, there is an intersting reprint of a Sound Practices Article on the web
on exactly this....
http://www.chimeralabs.com/~CL/KyrieArticles/KyrieArticles.html
> For those really obsessed with the sound of a tube type or brand, it is
> also possible to use any tube as a rectifier:
For Examples look at the Marantz T-1 (a 845 P-P Amp which uses two more 845's
as Rectifiers....) or Ciro Marzio's Euperte (?) 845 SE Amp (he uses two 211's
as Rectifier)....
Nothing new under the sun....
dang...@earthlink.net
On 18 Jun 1998 06:05:38 GMT, lieut...@aol.com (LIEUTENDGN) wrote:
>It also heavily depends on whether the FAB is a " class 10" or "class 100" for
>the production.
>
>Class 10 usually yields lower noise.
>
>SONY R&D Lab in Japan has the lowest nosie FET on their back burner.
>
>I am surprised they never border to use what they have from their R&D lab for
>the consurmer products.
>
>May be they are saving it for the last minute shoot out ???
>
>Their comptetitors are trying to do a reverse engineering on the FET with no
>luck.
>
>My suggestion : hold your breath.
>
Greetings Lieutendgn,
Scanning through your replies to this thread I would have to agree,
and would offer this thought:
Consider: the absolute junk the "offshore" electronics companies
pedal to non-discriminating US and Europaen consumers. The worst SS
NFB crap w/ the most "imaginative" specs. They dish out all the bad
equipment we are "conditioned" to buy.
Consider: the extraordinary stuff many of the same eastern
"corporate" designers, and of course their honorable friends, use for
their own home audio equipment. Highly expensiver hand made
transformers, difficult to source and expensive Triode active
amplifying elements, low noise passive components, low NFB circuit
topology -- all lovingly assembled, because they would never think to
assult their own hearing w/ the crap they peddle for foreign
(primarily US) consumption.
Interesting too, that most of the eastern high end audio cells
communicate in obscure languages that are inaccessible to a majority
of western readers. However, when it's time to "market" the crap,
their communications skills improve tremendously. My, but are those
SON* advertisements cute and funny. Takes a lot of insight to
understand "foreign" humor, and generate warm fuzzy feelings that
stroke "barbarian" minds.
I think it shows disrespect to westerners to sell us shi* and hide the
good stuff behind walls of communication barriers. Why not apply the
same communications ingenuity they employ when selling junk, to
getting out the message that there are much better sounding designs
available, that if reproduced in mass volumes, could be reduced in
price sufficently to be affordable by the average "barbarian"
consumer?
The reason is: listening to good sound is their role, dishing out
USD's is our role.
Just something to consider the next time you read an oh so appealing
advert for cheap SS NFB crap. Remember too, that these are the same
eastern corporations that busted and bought out US high end audio
companies, a practice that continues today.
Anyone want to buy a McIntosh "Clarion" car stereo?
Regards,
Dangerdave
Steve Jones
A small addendum to a really great exchange:
dang...@earthlink.net wrote:
>
> On Sat, 20 Jun 1998 22:54:36 +0100, "Russ Sadd" <sad...@aston.ac.uk>
> wrote:
> >
> >Because of this, a good few ideas are thought of as being much more recent
> >than they are: the 'Japanese production system' concepts were being
> >implemented in the '30s, <snip>
>
> Yes, I agree. The way history rewrites and re-accredits technological
> achievements can be pretty dissapointing.
The so called 'Japanese production system' concepts were indeed being
implemented in the '30s.......the 1830s (in the US). It was unique and
labeled as The American System of Manufacture (as early as London's The
Crystal Palace Exposition in 1851).
Use of French armaments (and the Uniformity System) during the
Revolutionary and 1812 Wars, practically demonstrated the advantages of
the concept of interchangeable parts. Add the inheirent 'tinkering'
nature of the Americans, combined with the lack of a large skilled labor
pool (or Class as in Britian), and Congressional mandate in "The Uniform
Ordinance Act of 1815", and you've got the genesis of all this 'modern'
technology that seemingly gets (re)-invented every 60 years or so.
-Steve Jones
Henry Pasternack
=?ISO-8859-1?Q?Andr=E9_Jute?= (an...@indigo.ie) wrote:
: I must say, the net is a very poor medium for humour.
I'll say. Maybe you should try rec.audio.humor.
Andy Moss
dang...@earthlink.net wrote:
> SS lasts ORDERS OF MAGNITUDE longer.
Nonsense.
> For HF noise, put in a few 5nF bypass caps across the diodes, use low
> noise HEXFRETS, and use a large Cout filter, which you can do easily
> w/ SS and not with Tubes, and gain the unequalled sonic benefits the
> large Cout provides
Um, I think you mean Hexfreds, Dave. They are high speed diodes - and
they are noiser than your typical 1N4007 - and GZ34. Also, they are
designed for high-speed switching power conversion. Not line voltage
rectification.
> Sure, you can run an amp off a single GZ-34, GZ-37, 82, 83, etc., but
> w/ a few hundred mills of current capacity, you will never get the
> kind of clean specs and clean sound you can achieve by a high quality
> brick wall supply w/ loads of transient current capability.
Well, to extrapolate, Dave. I think you'd be happier with a fully Solid
State amplifier.
> The neat thing about tube amp design to me, is that I am free to chose
> the very best technolgies to solve each design problem of the amp at
> the highest level. For example, despite decades of device
> development, BJT, FET, IGBT, GAS, etc., the Triode still remains the
> most linear active amplifying element available. So for a good linear
> amp, in go the Triodes. Why not be open minded and choose the best
> technology available for each circuit design issue in each part of the
> amp, regardless of the era in which these technologies were developed?
Um, you can easily create a more linear circuit with solid state than
with a lone triode.
Cheers
Andy
Russ Sadd
Steve Jones wrote in message <358E89...@vcd.hp.com>...
>A small addendum to a really great exchange:
>implemented in the '30s.......the 1830s (in the US). It was unique and
>labeled as The American System of Manufacture (as early as London's The
>Crystal Palace Exposition in 1851).
>
>Use of French armaments (and the Uniformity System) during the
>Revolutionary and 1812 Wars, practically demonstrated the advantages of
>the concept of interchangeable parts. Add the inheirent 'tinkering'
>nature of the Americans, combined with the lack of a large skilled labor
>pool (or Class as in Britian), and Congressional mandate in "The Uniform
>Ordinance Act of 1815", and you've got the genesis of all this 'modern'
>technology that seemingly gets (re)-invented every 60 years or so.
I was actually thinking of statistical process control here, the basis of
TQM and the Japanese (and now Western) definitions of 'quality'. It was
certainly implemented by both Britain and America during WW2, after much
theoretical discussion in the '30s. However, in the post-War years, Britain
concentrated on cost-cutting instead (the War was devasting to the economy
in general and industry in particular) and America was able to export purely
on the ability to supply, which later turned into competition on grounds of
cost.
Standardization also came of age in the '30s and '40s - the American
standard valves, for example, and the standardization of electronic
components derives from this era. There were British plans during the War to
concentrate on 3 multi-purpose valve models, and the Germans only used one
type of valve in their aircraft.
ASM is more what we'd term 'mass manufacturing', and yes, that's definitely
industrial revolution stuff. Babbage actually set out the principles of MRP
I and II, not to mention a good deal of modern ideas concerning CIMS, in his
1832 book, 'The Economy of Manufacture'. It could easily be a modern text.
As an aside, Quad actually stand out amongst British companies of the '40s
and '50s in their customer service, and what we'd now call a 'Japanese'
style product - the build of the Quad IIs is pretty unique for the time.
Russ
Steve Jones
Russ Sadd wrote:
>
> ASM is more what we'd term 'mass manufacturing', and yes, that's definitely
> industrial revolution stuff. Babbage actually set out the principles of MRP
> I and II, not to mention a good deal of modern ideas concerning CIMS, in his
> 1832 book, 'The Economy of Manufacture'. It could easily be a modern text.
Excellent!!
BTW, Babbage was expounding the differences between the 'making' system
(of British in origin) and the 'manufacturing' system (of American in
origin) after his visits to America to view the native New England
textile and small arms manufacturers in the 1820s (think of him as a
British contemporary to America with de Toqueville, with his eye on
technology vs, Alexis deT's observations on America's democracy).
Even more interesting, Babbage in the 1850s intended to solve some of
the problems with manually calculated astrological charts. At that time,
three mathematicians worked independantly. If there were errors, it was
assumed that the other two working blind to the error, would referee the
"correct" answer. Unfortuneately, there were a number of instances of
all three mathematicians comming up with figures that did not correspond
with each other. Babbage sought to use the punch card, used in textile
factories to easily change cloth designs, with the automated machinery
developed for mass production of small arms. His intent was to produce
"a weaver of numbers". He invented die casting to aid in the unformity
of parts required (wheels, gears) for his analog machine.
Today we know it as the programable computer.
-Steve Jones
TubeGarden
Quality has been whored into meaning statistical consistancy. Any moron with an
HP can determine good or bad with no knowledge or understanding of anything but
statistics and the HP keyboard.
I build my own amplifiers for the same reason anybody else who understands the
implications of mass production does, the game is simply not the same.
I want good sound.
They want something that sells.
I want to spread my understanding of Music.
They want to spread their margins.
Modern?
"If you want something done right, do it yourself."
- Adam, after eating the peach he sent Eve to pick
André Jute
TubeGarden <tubeg...@aol.com> wrote:
> What could I do to improve
> the sound?
You have broken the KISS principle. If you listen with a high-res
stethoscope (Stereophile-recommended MedicPose, of course), you will
hear that your amp sounds far too liquid. This is because the power
supply is not a thing on its own but part of the transfer function of
the amp. What you are hearing are the deleterious effects of all those
chokes acting like slurry on the oil in the caps, lopping off the tops
of the waves in the oil and so in your music. If you think of a heavy
oil spill on the sea dampening the waves, making them roll real slooow,
you will know exactly what you are hearing. You simply have too many
time constants adding cholesterol to the arteries of your amp. One time
constant is protein, two time constants are fat. (I am indebted to Henry
Pasternack for this fine explication of a very difficult subject, which
he confided to me one valve-lit night as we danced cheek to cheek in LA
Place Domain, the *in* restaurant of the City of Angels with a lot of
Poles that he said were invading his head.)
You need a truly radical rethink to speed up your amp. Start your
slimming course by ripping out every second choke and oil cap and send
them to me, postage prepaid, for socially-responsible disposal. In
return I shall send you the correct KKKustom KKKaps to use. These
electrolytics are the modern way to go. They are made from Reynolds
aluminum, the up-to-date material also used in the best aeroplanes,
rubbed only on the thighs of certified virgins, and passed within
micrometer-controlled osmotic distance of a unidirectionally drawn
sterling silver bar irridescently verdigris-treated by the secret
Faberfelt process; these exclusive caps are then handfinished by Black
Forest dwarves (the government freed the elves used before, and
leprechauns lack cathodic empathy) in an axial rolling process through
an environmentally-aware nauru-trimmed anechoic chamber to suppress
catalytic resonances, and are finally immersed by pre-tensioned (in case
you're wondering, that means it's bolted both ends) Nylon66
coarse-metric threaded rod in DuPont's RepulseTeflon to ostracize any
transients trying to attach themselves parasitically to your sound.
There is a reasonable if not necessarily small extra charge if you want
them shipped via Japan for the addition of zen or bushido (but not
both--they are mutually exclusive), the zen spirit giving a
fatalistically dynamically-non-discriminating sound very much like that
of one hand clapping, the bushido force working on the forteful primals
like a priest on a quiverful Catholic father of eleven considering
contraception. (That's what the ZS and BG on the cheap Black Gate
consumer brand of electrolytics stands for, "zen spirit" and "bushido
gore", but the Black Gates get only very small quantities of these
desireable elements, whereas the KKKustom KKKaps are exposed to The
Force for the full two seconds it takes to unwrap them and then rewrap
them in the special pink paper of pre-disposed Financial Times which so
well suits the decor of the VIP lounge at the airport in Tokyo where
these rare sonic elements we call The Force are to be found.) The six Ks
in the name of these very special caps refers to the price. You will
have to sell your house and car to pay for such wonderful caps but
what's that to someone truly committed to fine audio?
André Jute
<Thor...@tnt-audio.com> wrote:
> Hi all,
>
> > Tube regulation for any voltage or current required by any tube one
> > wants to use is real easy once one stops thinking it is complicated and
> > baffling oneself by seeking a correspondingly complicated solution.
>
> Good statement. But why use Valve Regulation.... Any Type of regualtion is as
> much a pain to get to work well as is the actual Amplifier, usually more
> so....
I wasn't proposing it to anyone. I was merely responding to someone's
suggestion that we discuss shunt regulation by providing a framework for
the discussion. And what a good discussion it has been!
>
> > You simply use another tube exactly the same (or bigger) as a shunt
> > regulator.
>
> Now that is a good Idea.... But You MUST feed the Shuntreg from a Current
> Source and of couse, Andre' you would not use FET or BJT for that, would you
> now?
I don't know, Thorsten. Whether I would use a transistor would depend on
how available Allen Wright and you are the week I am doing it, so that I
can get the best advice... I'm not dogmatic at all: whatever sounds
right is okay by me. The nice thing about this discussion has been the
pleasant tone and the absence of idiots trying to force everyone to take
sides.
> BTW, there is an intersting reprint of a Sound Practices Article on the web
> on exactly this....
>
> http://www.chimeralabs.com/~CL/KyrieArticles/KyrieArticles.html
That sounds like the guys I mentioned in Texas somewhere. (Maybe you
should take the Word of the Current Source to them, Thorsten, like a
missionary to the New World from the Old. The last time I heard, about
four or five years ago, I don't remember them being into current
sources...)
>
> > For those really obsessed with the sound of a tube type or brand, it is
> > also possible to use any tube as a rectifier:
>
> For Examples look at the Marantz T-1 (a 845 P-P Amp which uses two more 845's
> as Rectifiers....) or Ciro Marzio's Euperte (?) 845 SE Amp (he uses two 211's
> as Rectifier)....
>
> Nothing new under the sun....
That's my point (though your examples are new, not old <G>), and also
made very strongly by Russ Sadd, who had some clever examples from
before any of us were born.
>
> Later Thorsten
>
> -----== Posted via Deja News, The Leader in Internet Discussion ==-----
> http://www.dejanews.com/ Now offering spam-free web-based newsreading
Andre
Bob C
André Jute wrote in message
<1daya1y.qh...@ts01-63.cork.indigo.ie>...
>TubeGarden <tubeg...@aol.com> wrote:
>
>> What could I do to improve
>> the sound?
>
>You have broken the KISS principle. If you listen with a high-res
>stethoscope (Stereophile-recommended MedicPose, of course), you will
<SNIP>
ANDRE JUTE STOP WASTING BANDWITH WITH YOUR STUPID BULLSHIT!!!
John Byrns
Come on Bob, that was the first post from Andre I really enjoyed, he was
clearly in his element with this post. It wasn't wasted bandwidth, it was
a good piece of entertainment.
Regards,
John Byrns
TubeGarden
>The six Ks
>in the name of these very special caps refers to the price.
Yup, I knew that. That's why I shop at Apache Reclamation here in Phoenic. I
can swim in oil at US$3. a can, I can. Actually, I use all those chokes to
reduce Tsunami waves in the oil. Hate to to have a spill on Timpani Al, eh? I
started with one choke, you know, and I liked it, and then I put one on each
channel, and I liked it more, now I'm doing chokes on the 'net to support my
habit. I have this segmented oil slick attached to an SV83 wired triode with
adjustable fixed bias to give me 60mA at 200V and use up the rated 12 watts of
plate. It is to dream, perchance to sleep. I wired my RAT Shack 8" paper (The
$9.99 Lowther) speakers with the Ultra Kleen 10 base 2 RG-68 teflon coax.
(Thirty Nine Cents, the foot) and lay in wonderment listening to this strange
new flesh I've found.
The drivers are mounted to 36" x 5/8" particle board circles with 7" cutouts
near the middle and hang from the ceiling on 3/16" nylon diamond braid.
What are we going to do tonight, Braid?
Same thing we do every night, Pinky. Try and overhear the world.
Stay committed,
John Harper
In article <358E93...@ibm.net>,
and...@ibm.net wrote:
> Um, I think you mean Hexfreds, Dave. They are high speed diodes - and
> they are noiser than your typical 1N4007 - and GZ34. Also, they are
> designed for high-speed switching power conversion. Not line voltage
> rectification.
>
very hard to measure because of high levels of RFI at home.
As far as SWITCHING noise is concerned, Hexfreds are certainly a great
deal better than regular diodes or even regular high-speed diodes. With
the Motorola MUR1100Es I can see clear switching noise, but with Hexfreds
I can see none at all.
What they were designed for seems irrelevant. The 6DJ8/etc was designed as a
cascode VHF/UHF amplifier, but it doesn't stop it being the most widely used
preamp tube (rightly or wrongly). The 417A/5842 was designed as a low-noise
RF front end, but is popular for phone stages. The 6C33C-B was designed as a
regulator pass stage but... well, you get the idea.
John
TubeGarden
And trees were designed for holding up leaves, but that doesn't mean all wooden
musical instruments are basically flawed.
My amp uses solid state for filaments and tubes for B+ and a lantern battery
for bias.
No one wants to discuss batteries?
dang...@earthlink.net
>A small addendum to a really great exchange:
>
>dang...@earthlink.net wrote:
>>
>> On Sat, 20 Jun 1998 22:54:36 +0100, "Russ Sadd" <sad...@aston.ac.uk>
>> wrote:
>> >
>> >Because of this, a good few ideas are thought of as being much more recent
>> >than they are: the 'Japanese production system' concepts were being
>> >implemented in the '30s, <snip>
>>
>> Yes, I agree. The way history rewrites and re-accredits technological
>> achievements can be pretty dissapointing.
>
>The so called 'Japanese production system' concepts were indeed being
>implemented in the '30s.......the 1830s (in the US). It was unique and
>labeled as The American System of Manufacture (as early as London's The
>Crystal Palace Exposition in 1851).
>
>Use of French armaments (and the Uniformity System) during the
>Revolutionary and 1812 Wars, practically demonstrated the advantages of
>the concept of interchangeable parts. Add the inheirent 'tinkering'
>nature of the Americans, combined with the lack of a large skilled labor
>pool (or Class as in Britian), and Congressional mandate in "The Uniform
>Ordinance Act of 1815", and you've got the genesis of all this 'modern'
>technology that seemingly gets (re)-invented every 60 years or so.
>
>-Steve Jones
Interesting point, Steve.
On a more contemporary note, the US process quality control expert,
Denning, was "imported" to Japan, virtually revered as a consultant,
and was arguably the person most responsible for the turn-around in
product quality which led to the tremendous success of Japanese
products in international export markets, beginning circa 1970's.
Only after the Japanese success threatened US manufacturers in the
Automotive, Consumer Electronics, and to a lesser extent,
Semiconductor sectors, did US manufacturers finally adopt Denning's
methods, a change in US corporate policy which could only be
accomplished by threating the internal management structures w/ loss
of employment and corporate bankruptcy under competition from the
"Japanese" manufacturing systems. The very same Denning (US) systems.
Old story, but perhaps worth repeating.
Thanks for the good info, Steve and Russ.
Regards,
Dangerdave
dang...@earthlink.net
I agree John. High end audio design is a hodge-podge of technologies.
Mix and match from all commercial market segments, RF, controls, SS,
whatever, and sort for the components that have the best audio
attributes.
If high end audio had more commercial impact, we would probably have
the same range of "audio only" components of highest quality, that,
for example, digital logic designers have at their disposal.
Regards,
Dangerdave
dang...@earthlink.net
I love batteries. Tough sell for consumer products though, when tubes
are so current greedy.
Using batteries (lead gell cells) I designed a mass spectrometer head
amp that had 75E-15A sensitivity.
That was a fun design. That amp was so sensitive, that you could wave
your arm slowly through the air, from several feet away, and watch the
output meter respond.
Seeing that it's frequency response was about DC to 1 HZ, not quite
the ticket for audio though :-)
To the point though, some of the tubes for old car radios could be
operated off of batteries. I think there were other battery operated
tubes also, but I haven't investigated them.
Regards,
Dangerdave
André Jute
TubeGarden <tubeg...@aol.com> wrote:
> No one wants to discuss batteries?
>
> Al B^}
> Al Marcy
> TubeG...@aol.com
Yes, please. Instant DC!
I have always had this nasty suspicion that a battery used for grid bias
is basically inserted in what is a charging circuit. So I don't use
rechargeable batteries. But can one?
Which kinds of batteries are intrinsically safe in such a circuit
(whether it recharges or not)?
Andy Moss
I've seen this discussion done before. Batteries are not inherently
noise-less by any means.
The chemical reaction generates plenty of hash noise.
John Harper
In article <1db2w8l.s5...@ts01-25.cork.indigo.ie>,
an...@indigo.ie (=?ISO-8859-1?Q?Andr=E9_Jute?=) wrote:
> I have always had this nasty suspicion that a battery used for grid bias
> is basically inserted in what is a charging circuit. So I don't use
> rechargeable batteries. But can one?
>
> Which kinds of batteries are intrinsically safe in such a circuit
> (whether it recharges or not)?
>
through silicon diodes, it remembers they were silicon and there is a silicon
effect - attenuated of course, but still there - as the electricity comes
out again. It will still sound like silicon. Rechargeable batteries are
therefore out.
Batteries are PROBABLY safe if the energy was put there uniquely by
chemical means. For example (simple case, maybe trivial though probably
OK for bias circuits) -- take a piece of zinc and a piece of copper and
stick them in a lemon. You will get around 1V (if memory serves). The
energy was put there, essentially, by a supernova a while ago, when even
the 300B was yet to be a dream. It is probably clean.
(BUT you need a low-noise lemon, at least for a bias supply. This is not
a joke. There was an article about precisely this problem a while back
in MJ. Since citric acid is citric acid whatever the fruit, they concluded
that Satsuma oranges were the best. BUt maybe this was just an attempt
at a non-tariff trade barrier).
There is also the question of the origin of the metal, and its impurities.
Personally I am a cynic in this regard. BUt there are after all people who
claim to be able to hear which mine the silver came from, and the silver
is just a passive conductor whereas the metals in a battery cell are
taking an active part in the circuit. 6N zinc is hard to find. You could
make your own by electrolytic refining BUT you have to think about the
electricity that was used, which becomes circular.
SBench
>The problem is, batteries remember how they were charged. Charge a battery
>through silicon diodes, it remembers they were silicon and there is a silicon
>effect - attenuated of course, but still there - as the electricity comes
>out again. It will still sound like silicon. Rechargeable batteries are
>therefore out.
>
>Batteries are PROBABLY safe if the energy was put there uniquely by
>chemical means. For example (simple case, maybe trivial though probably
>OK for bias circuits) -- take a piece of zinc and a piece of copper and
>stick them in a lemon. You will get around 1V (if memory serves). The
>energy was put there, essentially, by a supernova a while ago, when even
>the 300B was yet to be a dream. It is probably clean.
>
>
You will get a better tuber sound if you use
a potato with copper and zinc rods inserted.
Best Regards,
Steve
Check my web page .. http://members.aol.com/sbench101/
Andy M
Chris Hornbeck wrote:
> In low-level and preamp stages I've always assumed (and been proven
> right enough) that grid current is negligible for use of AA size
> batteries. Use of one of these to bias a 12AX7 is just right, and
> allows solidly bolting the cathode to ground, with many obvious
> benefits. They last forever, too.
> The discussions of battery noise are puzzling; I don't measure
> anything identifiable.
I think extreme measures to get rid of "harmful" noise are absurd.
Batteries are right up there with voodoo and magic as viable solutions
to this "noise" issue.
The only time noise becomes and issue is in MC phono stages - and there
are ways around this using very conventional designs.
Cheers
--
Andy Moss
1800 Sheppard Avenue East, #55104, Willowdale, Ontario M2J 5B9
André Jute
John, Steve:
Please return the wallwarts I sent you for your laptops. I have now
discovered they are lemons and have found a new use for them.
**I don't know if batteries remember by what means they were charged,
but it is well known that rechargeable batteries remember *how far* they
were discharged before reconnection to the charger; that is the point of
all those total dischargers.
You can't call just any old amp a tuber. The name Potato-Amp or
One-Tuber is a WE trademark which they acquired when ITT took over the
Colorado Beetle Company of Denver under the misapprehension that they
were getting into fastback automobiles, and later uniquely applied to
single-tube 417A or 437A amplifiers.
Six-nines zinc is no problem: I have a backyard smelter left over from
my days as a socialist under the influence of Chairman Mao and,
seriously, two working blacksmith's forges within walking and cycling
distance of my home (I live in the *working* country, not the yuppie
dream), and a connection with a statuary foundry that will cast anything
on demand if you have the money to pay for a biggish batch of material.
Please send detailed instructions for the refining process, preferably
in Gaelic as my smelters live in leprechaun country.
Come on, John, confession is good for the soul. You made up that
business about the MJ writer hearing the difference between common or
garden citric acid and Satsumas, didn't you? Anywhere but in high-end
audio I wouldn't even ask... but I've already had a letter from a
Californian audiophile asking (in perfectly good English) if there is a
West Coast stockist of the stuttering KKKustom KKKaps I was recommending
to Al Marcy yesterday. KKKer-razeee.
Andre
**For those who don't have English as a first language, in this letter
only the paragraph marked with the two asterisks is serious. The rest
are jokes. If you knew, I apologize for mentioning it. But when I was
learning English, I often wished the anglophones would send a man with a
red flag before their jokes.
SBench <sbe...@aol.com> wrote:
Chris Hornbeck
On Tue, 23 Jun 1998 20:11:19 +0100, an...@indigo.ie (André Jute)
wrote:
>I have always had this nasty suspicion that a battery used for grid bias
>is basically inserted in what is a charging circuit. So I don't use
>rechargeable batteries. But can one?
>
>Which kinds of batteries are intrinsically safe in such a circuit
>(whether it recharges or not)?
In low-level and preamp stages I've always assumed (and been proven
right enough) that grid current is negligible for use of AA size
batteries. Use of one of these to bias a 12AX7 is just right, and
allows solidly bolting the cathode to ground, with many obvious
benefits. They last forever, too.
The discussions of battery noise are puzzling; I don't measure
anything identifiable.
Chris Hornbeck, Guyville
Will Rogers on the stock market:
"Don't gamble. Take all your savings and buy
some good stock and hold it till it goes up
then sell it. If it don't go up don't buy it."
TubeGarden
Charlie at AES told me before WWII (the big one) that radio manufacturers
routinely used a "grid bias cell" because it was cheaper than cathode bias. The
circuit IS essentially a charger and the little wonders lasted the life of the
radio.
I am using a 6V lantern battery on my output SV83 grids because I have 50K pots
to make my fixed bias adjustable and there is always a little current flowing.
I like being able to dial in the plate current. I have meters in the circuit
between the OT and the final filter cap.
I do not hear any noise, and residual (inputs shorted) at the OT at 2V into
4ohms is < 0.02 mV measured on my HP334A AC. That is some pretty finely ground
hash...
Happy Ears!
TubeGarden
>The only time noise becomes and issue is in MC phono stages - and there
>are ways around this using very conventional designs.
IMHO the only time noise is an issue is when you find out the sound is better
after you remove it. The great thing about building you own stuff is you do not
need the approval of some committee to make a design decision. If I think the
sound is better, it is a good design decision. If my meters notice, fine, if
not, so what? I am not playing the Music for my meters. I find meters useful in
making sure my dipshit design techniques have not pushed some component out of
its specified operating range, not to tell me if I am enjoying the Music more
or less.
>Batteries are right up there with voodoo and magic as viable solutions to this
"noise" issue.
"Sound" is the only issue, science and technology exist only to help achieve
good sound. Batteries are not "extreme" in any sense. They are cheap, easy to
work with and reliable. They have been used successfully for many decades.
Cathode bias is cheaper, even easier to manufacture, and allows greater
variations in tube performance to still work in the circuit. Better sound is
not in the list.
Anything which detracts from my enjoyment of the Music is "harmful". If I
choose to use low value resistors on the AC legs of my ss bridge rectifier to
lower the level of switching noise on my filament supply, so be it. There is no
way to make my perception of Music absolute and universal. Yours, neither. Ears
and brains are organic, living mechanisms constantly growing and changing,
subject to the whims of chance and environment. Trying to find some absolute
meaning in this life stuff is "Voodoo".
Tubes are wonderful little packages of technology which are currently so cheap
it is wonderful. The "Big Game" has moved on. Let us help one another enjoy
playing with these leftovers of the twentieth century instead of demanding that
we all play with them "the proper way".
dang...@earthlink.net
>"Sound" is the only issue, science and technology exist only to help achieve
>good sound.
I agree entirely.
The bottom line is: what does it take to make the design sound better?
The design is "optimized" when you can no longer make improvements in
sound quality.
FYI: reference the concurrent post, "Listening Test Methods"
I faced this dilemia often when working for/with electronics
corporations. It was often the case that many trained, academic, EE's
resisted using "soft" evaluation criteria, because they argued such
methods were not "proper" or "rigourous".
I once had an assignment to train EE's in these methods, and I can
tell you, it was a bitch, and often I felt like I was arguing w/ brick
walls.
( Appologies in advance to the many, many trained EE's and dedicated
experimenters, in RAT, who love good sound and already do "whatever
it takes" to make sound better ).
Bottom line though is sound quality. Real test is what the
consumer/user perceives as good sound. Real challange is how to
systematically employ "soft" experimental listening criteria to
optimization of "hard" circuit parameters.
Regards,
Dangerdave
André Jute
TubeGarden <tubeg...@aol.com> wrote:
> IMHO the only time noise is an issue is when you find out the sound is
> better after you remove it. The great thing about building you own stuff
> is you do not need the approval of some committee to make a design
> decision. If I think the sound is better, it is a good design decision. If
> my meters notice, fine, if not, so what? I am not playing the Music for my
> meters. I find meters useful in making sure my dipshit design techniques
> have not pushed some component out of its specified operating range, not
> to tell me if I am enjoying the Music more or less.
Hear, hear.
You can't be too "dipshit" if you get residual noise at the OPT down to
0.02mV.
Up the Star Chamber.
Andre
Bob C
André Jute wrote in message
<1db4y6v.1wk...@ts02-46.cork.indigo.ie>...
>
>Hear, hear.
>
>
>Up the Star Chamber.
>
Is that your call to the Cork Boys!!!!!!!!!!!!!!!!!!!!!!!!!!
Ned Carlson
On Wed, 24 Jun 1998 19:09:50 +0100, an...@indigo.ie
(=?ISO-8859-1?Q?Andr=E9_Jute?=) wrote:
>Up the Star Chamber.
Man, you are a reactionary...the British ditched
Star Chamber back in the 17th century, didn't they?
What's next, you're gonna call for Charles I's
head to be reattached and the reimpostion of forced
loans & ship taxes?
Ned Carlson Triode Electronics,2225 W Roscoe Chicago, IL, 60618 USA
ph 773-871-7459 fax 773-871-7938
12:30 to 8 PM CT, (1830-0200 UTC) 12:30-5 Sat, Closed Wed & Sun
http://www.triodeel.com
Text file catalogs:Catalog 'Bot at cat...@triodeel.com
Jamie
Yeah, Bob, lighten up! After all Andre is clearly in his element here!
As long as he sticks to his true calling, writing fiction, I have no qualms.
Jamie Campbell
John Byrns wrote in message ...
>In article <VaBj1.62$601.88648@pm01nn>, "Bob C" <bo...@mci2000.com> wrote:
>
>> André Jute wrote in message