Some Facts on the McIntosh Legend
Anton Elron
subject that are not known to everyone, evidently. Many people
consider McIntosh almost synonymously with tube hi-fi. Keep in mind
the following:
1.McIntosh vacuum tube power amplifiers use a unity coupled output
circuit with a specially designed output transformer. They operate in
Class AB2 to Class B2 with a lot of feedback and a cathode follower or
(in early sets) a interstage transformer to provide relatively low
impedance drive to the final tubes' control grids.
This places them at odds with the general trend in "high end"
practice which has shied away from global feedback and toward Class A
operation of tubes.
2. McIntosh products, possibly with rare exceptions, were constructed
with strictly consumer grade components and with the intent ,
successfully realized, of keeping build costs low. Mc tube amps are
built much more like a Fender than a Tektronix, Collins, or telco box.
Forty years on we see this in things like the almost universally
corroded chrome tops-they were bent out of prechromed steel rather
than triple plated after bending. While one could strip, replate, and
rescreen this isn't done because " it would devalue the piece"(also,
cost money) so Japanese collectors instead get entirely new tops
clandestinely fabricated from prechromed material, which, no doubt,
will rust once again in the Japanese climate.
3. The McIntosh legend was constructed largely by McIntosh as they
channelled a reasonable amount of the gross margin of their
products-which was spectacular because they were an entry level
employer in a small town and did everything inhouse-into things like
dealer education and discipline (you had to be or employ a competent
bench tech, buy test equipment and keep it in cal, and would be
terminated if caught discounting)and the service clinics. All the
principals were into General Semantics, also.
4. While there are American McIntosh collectors and buffs, they don't
set the market. The Japanese do. The Japanese have been buying this
stuff for years and will continue to do so as it is a matter of
prestiege and face to own such a thing, particularly a MC275 in mint
original condition. Maybe this is compensation for certain national
feelings of inadequacy, as one can see for oneself in Araki's "Tokyo
Lucky Hole"(Taschen 1997) and as described a little obliquely in the
autobiography of Gregory Boyington (the real model for Robert Conrad's
TV show "Baa Baa Black Sheep").
Bob Groschen
"Anton Elron" <anton...@hotmail.com> wrote in message
news:ea82a75b.02011...@posting.google.com...
> 1.McIntosh vacuum tube power amplifiers use a unity coupled output
> circuit with a specially designed output transformer. They operate in
> Class AB2 to Class B2 with a lot of feedback and a cathode follower or
> (in early sets) a interstage transformer to provide relatively low
> impedance drive to the final tubes' control grids.
This is an interesitn statement Anton, how much would you classify as
"..a lot of feedback.." Based on my calaculations, they have no more or
less
negative feedback then any other tube amp of the era.
> 2. McIntosh products, possibly with rare exceptions, were constructed
> with strictly consumer grade components and with the intent ,
> successfully realized, of keeping build costs low. Mc tube amps are
> built much more like a Fender than a Tektronix, Collins, or telco box.
Another interesting statement. In what way are the components "..strictly
consumer grade?" I've done designs for spaceborne applications and
I can tell you the passive components used were the same you'll find in
AA5s. Just the derating was diferent.
> Forty years on we see this in things like the almost universally
> corroded chrome tops-they were bent out of prechromed steel rather
> than triple plated after bending.
Well, my MC-30 chassis are no thing of beauty, but that's because they were
stored in someone's garage for decades. Yet we see many many cases
of MC-30s, MC-40, etc that have perfect chrome plating even after all these
years. That just goe sto show that it's the storage environment that
determines
the level of corrosion, not the fact that they are plated.
> 3. The McIntosh legend was constructed largely by McIntosh as they
> channelled a reasonable amount of the gross margin of their
> products-which was spectacular because they were an entry level
> employer in a small town and did everything inhouse-into things like
> dealer education and discipline (you had to be or employ a competent
> bench tech, buy test equipment and keep it in cal, and would be
> terminated if caught discounting)and the service clinics. All the
> principals were into General Semantics, also.
A rose by any other name is still called marketing. So McIntosh did it?
So what? Their best tool tool for many years was their "Clinics". Even
Marantz copied this idea. As a former gear peddler in my engineering
school days, I can tell you quite emphatically that this was probably the
best, most powerful marketing tool ever devised. The idea that one could
bring a McIntosh amp in and have it gone over by factory techs for
nearly *free* made one hell of a lot of customers for life. Try that with
your ARC or Conrad Johnson!
> 4. While there are American McIntosh collectors and buffs, they don't
> set the market. The Japanese do. The Japanese have been buying this
> stuff for years and will continue to do so as it is a matter of
> prestiege and face to own such a thing, particularly a MC275 in mint
> original condition. Maybe this is compensation for certain national
> feelings of inadequacy, as one can see for oneself in Araki's "Tokyo
> Lucky Hole"(Taschen 1997) and as described a little obliquely in the
> autobiography of Gregory Boyington (the real model for Robert Conrad's
> TV show "Baa Baa Black Sheep").
And this is a problem? As I see it, it's a market just *waiting* for
Mcintosh
Labs to fill. Nothing quite like exporting *to* Japan instead of importing
*from*.
--
Best Regards,
Bob Groschen
Manufacturing Development Engineer
Agilent Technologies, Inc.
Colorado Springs, CO
Patrick Turner
Bob Groschen wrote:
> "Anton Elron" <anton...@hotmail.com> wrote in message
> news:ea82a75b.02011...@posting.google.com...
>
> > 1.McIntosh vacuum tube power amplifiers use a unity coupled output
> > circuit with a specially designed output transformer. They operate in
> > Class AB2 to Class B2 with a lot of feedback and a cathode follower or
> > (in early sets) a interstage transformer to provide relatively low
> > impedance drive to the final tubes' control grids.
>
> This is an interesitn statement Anton, how much would you classify as
> "..a lot of feedback.." Based on my calaculations, they have no more or
> less
> negative feedback then any other tube amp of the era.
I worked it out once upon a time, that the MC 75 had 20 dBof loop NFB from the
output secondary back to the input.
Then the output stage is indeed "unity coupled", so that
the gain is reduced from about 10 down to one, so there is
a shirtload of local NFB in this stage, as half of the total power producing
primary voltage is fed back to the cathode.
It is like a giant concertina phase inverter,
which is a case of high feedback use.
The end result is that the Mac has far more NFB than most other amps
of the era, yet did not achieve such fabulous distortion figures.
This was because a very high drive voltage was required
for the output stage, which was a class B2 concoction.
However, because of all the feedback, they argued all is OK,
and THD was not all that high.
But Zout was fabulously lower than most other amps of the day,
but ther was no great need.
To me, the Mac is an example of two steps forward, and one back,
to get one step forward.
But if you set a Mac up in class A, well, that would really be something,
not much to complain about then....
> > 2. McIntosh products, possibly with rare exceptions, were constructed
> > with strictly consumer grade components and with the intent ,
> > successfully realized, of keeping build costs low. Mc tube amps are
> > built much more like a Fender than a Tektronix, Collins, or telco box.
>
> Another interesting statement. In what way are the components "..strictly
> consumer grade?" I've done designs for spaceborne applications and
> I can tell you the passive components used were the same you'll find in
> AA5s. Just the derating was diferent.
I know not all about yankee amps.
I have a friend who badgered GIs to let the odd 6L6 fall outthe back of trucks
passing slowly through narrow
streets. Many 6L6s were given out, but only one to each kid,
and it really started something, which of course was the
Cult Of The Single Ended Tube Amp.
Ambitious Japanese kids trailing around after GI trucks leaving
radio stations yelled out "211!!!, 211!!!", and the old
ones got passed out the back.
With small timber houses, japanese tube hifi flourished,
while in US, everyone had to have huge numbers of 6550s and lots
of NFB.
After 35 years of secret enjoyments of SET sound, many US
and other western peoples became all misty eyed and very nostalgic about
what they may have missed out on, simple living music zen experience,
and reinvented SET all over again, and we have Sound Practices
magazine to illustrate this phenomena, and hardly any PP
within SP mag.
And of course culture warping goes both ways, and some Japanese have to have
a symbol of success on the mantel peice, such as MC275,
or if real sophistication and class was attempted to proove,
then a pair of mint Leaks was pride and joy.
This matches the jag in the garage, and tweed jacket.
>
>
> And this is a problem? As I see it, it's a market just *waiting* for
> Mcintosh
> Labs to fill. Nothing quite like exporting *to* Japan instead of importing
> *from*.
I think Quad did well with their reissue of a limited number ofspecial edition
Quad II amps.
I ask some folks who adore SS amps when their favourite maker
is releasing a new tube model, and they are very puzzled.
Patrick Turner.
Bob Groschen
"Patrick Turner" <in...@turneraudio.com.au> wrote in message
news:3C3ECB09...@turneraudio.com.au...
> I worked it out once upon a time, that the MC 75 had 20 dBof loop NFB from
the
> output secondary back to the input.
That's about what I remember. I don't have my notebook in front of me at
the moment.
> It is like a giant concertina phase inverter,
> which is a case of high feedback use.
> The end result is that the Mac has far more NFB than most other amps
> of the era, yet did not achieve such fabulous distortion figures.
"Fabulous distortion figures'? Patrick what in the world are you talking
about?
The MC-30 claims to provide 30 W at 0.3% distortion which *for that era*
was (to use your phrase) a "fabulous distorion figure". Name another amp
from the 50s that could do that!
> This was because a very high drive voltage was required
> for the output stage, which was a class B2 concoction.
Careful here Patrick! While the basic topology remained the same,
small changes in bias point were made from model to model. Just
compare the MC-30 to the MC-40 for example.
> However, because of all the feedback, they argued all is OK,
> and THD was not all that high.
> But Zout was fabulously lower than most other amps of the day,
> but ther was no great need.
> To me, the Mac is an example of two steps forward, and one back,
> to get one step forward.
Hindsight is always 20-20 Partrick. We can criticize all we want *now*
but let's remember this design work was done before I was born.
> But if you set a Mac up in class A, well, that would really be something,
> not much to complain about then....
Isn't that exactly what VanDerVeen suggested in his book on the Plitron
transformers?
> I have a friend who badgered GIs to let the odd 6L6 fall outthe back of
trucks
> passing slowly through narrow
> streets.
Well hey! I'd badger them too! :-)
> Ambitious Japanese kids trailing around after GI trucks leaving
> radio stations yelled out "211!!!, 211!!!", and the old
> ones got passed out the back.
Gee, I'd try "2A3" or "807" too! :-)
> I think Quad did well with their reissue of a limited number ofspecial
edition
> Quad II amps.
As did Marantz IIRC.....
> I ask some folks who adore SS amps when their favourite maker
> is releasing a new tube model, and they are very puzzled.
Shame on you Patrick! Asking hard questions of the clueless! :-)
MYanchick
1) Mcintosh amplifiers have no merit or realistic value.
2) The desirability of Mcintosh tube amps is ultimately rooted in some sort of
Japanese inferiority complex.
Opinions are one thing, but I find it funny that someone would actually
have such contempt for some old amplifiers.
Mike
John Byrns
<in...@turneraudio.com.au> wrote:
> I worked it out once upon a time, that the MC 75 had 20 dBof loop NFB from the
> output secondary back to the input.
> Then the output stage is indeed "unity coupled", so that
> the gain is reduced from about 10 down to one, so there is
> a shirtload of local NFB in this stage, as half of the total power producing
> primary voltage is fed back to the cathode.
Why does this notion, that the gain of the McIntosh output stage is about
one, persist like some kind of urban legend? The gain of the McIntosh
output stage is about two, not one.
> It is like a giant concertina phase inverter,
> which is a case of high feedback use.
> The end result is that the Mac has far more NFB than most other amps
> of the era, yet did not achieve such fabulous distortion figures.
> This was because a very high drive voltage was required
> for the output stage, which was a class B2 concoction.
> However, because of all the feedback, they argued all is OK,
> and THD was not all that high.
> But Zout was fabulously lower than most other amps of the day,
> but ther was no great need.
> To me, the Mac is an example of two steps forward, and one back,
> to get one step forward.
>
> But if you set a Mac up in class A, well, that would really be something,
> not much to complain about then....
If you ran the output stage of a McIntosh amp class A, people would
complain about the poor reliability, the glass would melt!
Regards,
John Byrns
Surf my web pages at, http://www.enteract.com/~jbyrns/index.html
John Byrns
<bob_gr...@agilent.com> wrote:
> "Patrick Turner" <in...@turneraudio.com.au> wrote in message
> news:3C3ECB09...@turneraudio.com.au...
>
> > It is like a giant concertina phase inverter,
> > which is a case of high feedback use.
> > The end result is that the Mac has far more NFB than most other amps
> > of the era, yet did not achieve such fabulous distortion figures.
>
> "Fabulous distortion figures'? Patrick what in the world are you talking
> about?
> The MC-30 claims to provide 30 W at 0.3% distortion which *for that era*
> was (to use your phrase) a "fabulous distorion figure". Name another amp
> from the 50s that could do that!
What about the Leak amps, with their 0.1% distortion? I remember lusting
after the Leak Stereo 20 during my high school days in the late 1950's,
thinking that when I got some money, that was the amp I would buy. By the
time I got the money, I had forgotten about the Stereo 20, and bought
something else, although the reasons were not entirely a negative
reflection on Leak. It seems to me, IIRC, the earlier Leak mono amps also
boasted 0.1% distortion ratings.
JOHN L STEWART
Not only done by McIntosh ( about 12 db) but in varying degree
by the Circlotron, Quad & the ultralinear connexion.
With a lot less gain variation in the output stage life gets
easier for the application of full loop feedback.
Many don't seem to realize that.
The triode by it's nature is inherently fedback.
Wish I had A McIntosh!! John L Stewart
BFoelsch
The Mac amps sounded pretty much as good as the best then available, but I
liked the 50W-2 because it was unconditionally stable under any reasonable
conditions (and many unreasonable conditions.) The disc cutting heads were
difficult loads, and standard practice was to make them more difficult by
connecting them to the amplifier through an RC network that gave you the LF
crossover you wanted. No problem for the Mac, but a lot of generic NFB
amplifiers had troubles.
FWIW, the standard recording amplifiers of the era were PP triodes, without
NFB. Often 2A3s or triode connected 807s. In the really high quality stuff
even the voltage amplifiers and equalizer stages were PP.
As far as their construction, I would have to agree that the Mac components
were nothing great. Better than an AA5, but not as good as Tektronix. The
Mac stuff definitely met its published spec, however, which was uncommon in
that era..
To me the distinguishing factor of the Mac power amps was that they just
plain worked well. You turned them on and they ran, no adjustments, no fuss,
no complaint. If you accidentally hooked a 4 ohm load to the 16 ohm tap you
probably couldn't tell the difference without test equipment. If you ran it
all day with no load nothing happened.
It may not be the best stuff in recorded history, but Mac, along with a very
few others, certainly set a new ( and very necessary) standard during the
1950's and 60's.
"Bob Groschen" <bobgr...@peoplepc.com> wrote in message
news:u3tnuq...@corp.supernews.com...
Bob Groschen
"John Byrns" <jby...@enteract.com> wrote in message
news:jbyrns-1101...@216-80-74-106.d.enteract.com...
Correct me if I'm wrong John, but didn't the leak Point One occur *after*
the MC-30 & MC-40?
Anton Elron
A reductio ad absurd......No, they're fascinating. Not worth big
money though. A mint 275 brings about eight grand in Japan.
John Byrns
<bobgr...@peoplepc.com> wrote:
Don't know, sounds reasonable, but either way it's irrelevant as you
didn't say anything about which came first, you just asked someone to
"Name another amp from the 50s that could do that!", which I did.
JOHN L STEWART
8 to 15. If you assume a gain of 10 & then reduce it to 2 as is
done in the McIntosh, then you have 14 db of local feedback.
The Circlotron beats that a bit by cutting gain in the output stage
to less then one. So now you have around 20 db local feedback.
It works well. I have one here (my own recipe) that manages
damping factor of around 3 without external feedback.
John L Stewart
John Byrns
> Gain of commonly connected pentode power amps is in the range
> 8 to 15. If you assume a gain of 10 & then reduce it to 2 as is
> done in the McIntosh, then you have 14 db of local feedback.
"8 to 15" seems a tad on the low side for a modern pentode output stage.
> The Circlotron beats that a bit by cutting gain in the output stage
> to less then one. So now you have around 20 db local feedback.
> It works well. I have one here (my own recipe) that manages
> damping factor of around 3 without external feedback.
I'm not sure what you mean by this statement? The circlotron and the
McIntosh circuit are identical from and AC standpoint. If you short out
all the power supplies and etc., and draw the AC equivalent circuits of
the both the McIntosh and the circlotron, you will find they are
identical, and both have an output stage gain of about two, not one!
Anton Elron
papers along with the commentary of the time. Norman Crowhurst has
some commentary on this as do some IRE papers in the late 50's. There
was a lot of professional interest in audio power amplifiers in those
days, whereas it is now mostly of concern to hobbyists. Pros just buy
them and plug 'em in. (The PA world of course has to have someone to
design their amplifiers but,largely, they have gone to operating
topologies and classes of no interest to the audiophile market)
As I recall, Mc's patent application and write-ups for nascent AES
papers (Mac doesn't go to AES anymore...) specifically cite the
fundamental objective as to achieve the highest possible ratio of
primary to leakage inductance to permit the most NFB. At least
publicly, they held this to be their chief concern in the use of their
output section.
McIntosh held numerous patents on their amplifiers, which enable the
modern student to derive a lot of information from the PTO on their
designs.
guy
guitar amps.
I haven't heard these but the MC60 (using 6550's) sure sounded
fabulous, with tremendous depth compared to SS amps.
So how is the design 2 steps forward, 1 step back?
How would one design a state of the art tube PP amp today, use
balanced design, current mirrors ala OP-AMP style?
Guy
Terry DeWick
tube McIntosh pass across the bench, IMHO the build quality and
performance can not be beat !! Just put filter and bias caps in a pair
of re chromed 30's all other parts were original, tested out at 30W
with .05 THD !!! For an old Mac to not beat it's specs it will have
been abused and or have tired tubes........otherwise the sound can't
be beat - especially the MC225 with the new EH7591's !!!!
The Black Beauty caps they used may have been consumer grade but
they were derated well enough to hold up 40 years !! Most of the
problems I see are the 30+ year old filter capacitors are bad, easy
fix compared to other tube amps that I work on !! I still get MC40's
and 240's with the original McIntosh 6L6's in them, they still meet
spec !! how many other brands can claim such good tube life.....
Seems strange to me about the Japanese buying up old McIntosh, we
haven't sold any into Japan in several years, and we do sell a bunch
of used McIntosh, maybe they can't read my web site like the other
130000+ people have......
Terry WB4NHB
Tubie
> news:ea82a75b.02011...@posting.google.com...
>
> > 1.McIntosh vacuum tube power amplifiers use a unity coupled output
> > circuit with a specially designed output transformer. They operate in
> > Class AB2 to Class B2 with a lot of feedback and a cathode follower or
> > (in early sets) a interstage transformer to provide relatively low
> > impedance drive to the final tubes' control grids.
>
> This is an interesitn statement Anton, how much would you classify as
> "..a lot of feedback.." Based on my calaculations, they have no more or
> less
> negative feedback then any other tube amp of the era.
>
since I was thinking about that matter for a long time I like to throw
my statement in here:
Someone else in this thread statet, that output stage gain is reduced
from 10 to 2 due to unity coupling. I agree.
But isn't the cross-coupling of driver supplies in fact increasing
driver gain by about the same factor ? If I vary the supply of an
anode load in phase with the signal, this will (almost) add to the
output signal.
IMHO this is a positive feedback reducing negative feedback, so
overall gain stays the same ?
Best regards,
Michael aka Tubie
Bob Groschen
> > Correct me if I'm wrong John, but didn't the leak Point One occur
*after*
> > the MC-30 & MC-40?
>
> Don't know, sounds reasonable, but either way it's irrelevant as you
> didn't say anything about which came first, you just asked someone to
> "Name another amp from the 50s that could do that!", which I did.
As near as I can tell, the Leak Point One evolved from the TL-12 and did so
around the same time as Frank McIntosh was producing the MC-30 (1950s).
The TL-12 put out a "nominal" 12 watts (2X 6L6/KT-66 PP-triode) at less than
0.1% distortion
and because of the triode output, had a *very* respectable damping factor.
And, for the record,
it claims to use 26dB of negative feedback.
Bob Groschen
"guy" <g...@nothome.com> wrote in message
news:3c3fb708...@news.telus.net...
> The MC30 and MC40 uses the 6L6's which is usually associated with
> guitar amps.
And your point is? Are you implying that guitar amps are now in the same
league
as McIntosh amps just because they use 6L6's????
> So how is the design 2 steps forward, 1 step back?
That's just Patrick. Don't worry about it. :-)
> How would one design a state of the art tube PP amp today, use
> balanced design, current mirrors ala OP-AMP style?
Until someone can design a tube that uses 'hole' emission as opposed to
electron emission, I think it's impossible. ;-)
JOHN L STEWART
the McIntosh & other circuits famous in the tube era.
Hope to hear the opinions of others regards these competing
topologies, not so much as to which sounds better, but rather
the arrangement of the various components. Each seem to have
some advantages along with problems.
The common formula used for calculating the gain of a pentode
stage, transconductance times load impedance is an approximation,
but is accurate to a few percent when the load is much less then the
plate resistance. That is usually the case, especially in a power amp.
Taking a couple of common pentodes-
For the 6K6 at 250 volts plate & screen 2.3 X 7.6 = 17.5
& For the 6L6 at 250 volts plate & screen 6 X 2.5 = 15
In a PP output stage you would normally use a lower load impedance
then 2X those figures in order to reduce the 3rd harmonic distortion.
The resulting gain would be proportionately less.
The increased 2nd harmonic in each tube of the PP pair will be
canceled in the output transformer.
All of the commonly used output stage topologies with the exception of
the Futterman & Circlotron drive the load as if the output tubes were
connected in series. The Futterman & Circlotron are fundamentally
different in that the output tubes are connected in parallel so far
as the load is concerned. I like to picture the Circlotron as a pair
of cathode followers connected back to back across the load.
If you were to draw the AC equivalent circuit, that is what you
would see. That still gets you a gain of somewhat less the one.
All of the gain of a Cathode Follower is applied as local feedback.
So 20 log 10 is 20db (local).
From the above I don't think you could rearrange the parts of
a Circlotron to end up as a McIntosh. The series vs. parallel
connexion to the load makes it unlikely.
There are quite a few out there questioning the bootstrap since it
is positive feedback. I usually use something like a 6SN7 as a driver.
The gain can never be more then the mu of the tube. I've
typically measured gain increase of about one db.
What is does is increases the available plate swing of the driver
by about 2X. Thats what you need to drive the output stage
of a McIntosh or a Circlotron without benefit of an
interstage transformer.
Henry Pasternack
news:3C406470...@sympatico.ca...
> All of the commonly used output stage topologies with the exception of
> the Futterman & Circlotron drive the load as if the output tubes were
> connected in series.
I'm late to this discussion, so forgive me if I miss the point, but it
seems to me that the conventional push-pull output stage has the two
tubes connected in parallel through coupling of the two halves of the
primary winding.
-Henry
JOHN L STEWART
You don't have to have the best product to be successful.
What you need is the best possible combination of product,
innovation, sales support before & after the transaction,
documentation & competetive pricing.
I never had any dealings with McIntosh or their dealers but from
Bob G's comments I get the feeling that McIntosh had all of those things.
After spending 35 years in hitech sales (HP & R&S)
I realize how important that is to the customer. John L Stewart
guy
>> The MC30 and MC40 uses the 6L6's which is usually associated with
>> guitar amps.
>
>And your point is? Are you implying that guitar amps are now in the same
>league
>as McIntosh amps just because they use 6L6's????
>
saturation, distortion levels and winding geometry.
BTW, does anyone know how McIntosh transformers are wound?
The 6L6 tubes are considered harsh on overload compared to 6550 and
EL34.
The advantage of the 6L6 is that its much cheaper than the 6550.
Guy
Anton Elron
> >> The MC30 and MC40 uses the 6L6's which is usually associated with
> >> guitar amps.
There were no electric guitars when the 6L6 was invented. It was
fifteen years before they were even heard of and then it was someone
with a lap steel playing Hawaiian music in a music store orchestra.
That was how amateur musicians aspired to aduation and nookie back
then. Times have changed...(?)
> >And your point is? Are you implying that guitar amps are now in the same
> >league
> >as McIntosh amps just because they use 6L6's????
> >
> no because guitar amps uses very different criterion for transformer
> saturation, distortion levels and winding geometry.
Until the boo-teek poofies hit the linen _nobody_ designed an output
transformer specifically for guitar amplifier use. Fender bought
Triads that were as close to a hi-fi transformer as he was willing to
pay for, then Schumacher wound them as close to a Triad as they could
on gang winders using the same setup as the 12V power xfmrs that go in
the Sears battery chargers they make.
Remember,nobody wanted distortion. They wanted the clean sound. Rock
and rollers cranking them up and overdriving them...an accident. A
guitar amp was just a PA or utility grade amplifier that was packed in
a case with a speaker and had a guitar maker's name instead of a movie
projector or tape recorder or whatever.
>
> BTW, does anyone know how McIntosh transformers are wound?
Besides Mc? Yeah. You gotta dig and/or take one apart. Some guy at the
Univ. of Cincinnati built his own version circa 1956. Exact winding
details are in the paper.
> The 6L6 tubes are considered harsh on overload compared to 6550 and
> EL34.
>
> The advantage of the 6L6 is that its much cheaper than the 6550.
Two 6550s that go two thousand hours are a lot cheaper than four 6L6s
that go a thousand in a lot of applications.
Bob Groschen
> BTW, does anyone know how McIntosh transformers are wound?
All the gory details are listed in Frank McInstosh's patent (2,477,074).
It's quite the read,
and I can't make any claims to finishing it yet with complete understanding.
> The 6L6 tubes are considered harsh on overload compared to 6550 and
> EL34.
It seems to me that this is *highly* implimentation dependent, so your
statement,
as it stands, is suspect.
> The advantage of the 6L6 is that its much cheaper than the 6550.
Just about all the 6L6 variants and the EL-34s are cheaper then the
6550/KT-88.
I believe it's just a market volume thing.
John Byrns
> Hello John & others following this thread. Here is my take on
> the McIntosh & other circuits famous in the tube era.
> Hope to hear the opinions of others regards these competing
> topologies, not so much as to which sounds better, but rather
> the arrangement of the various components. Each seem to have
> some advantages along with problems.
>
> The common formula used for calculating the gain of a pentode
> stage, transconductance times load impedance is an approximation,
> but is accurate to a few percent when the load is much less then the
> plate resistance. That is usually the case, especially in a power amp.
>
> Taking a couple of common pentodes-
> For the 6K6 at 250 volts plate & screen 2.3 X 7.6 = 17.5
> & For the 6L6 at 250 volts plate & screen 6 X 2.5 = 15
> In a PP output stage you would normally use a lower load impedance
> then 2X those figures in order to reduce the 3rd harmonic distortion.
> The resulting gain would be proportionately less.
> The increased 2nd harmonic in each tube of the PP pair will be
> canceled in the output transformer.
Hi John,
It's interesting that the common pentodes you choose for your example all
came in at or above the top of your gain range of 8 to 15, this suggests
to me that the actual range of gain for pentode output tubes is higher
than 8 to 15, perhaps more like 15 to 40. Consider these additional
common output pentodes.
For the 25L6 at 200 Volts plate & 125 Volts screen 8 X 4 = 32
For the EL84 at 250 Volts plate & screen 11.3 X 5.2 = 58.76
For the EL34 at 300 Volts plate & screen 11 X 3.5 = 38.5
> All of the commonly used output stage topologies with the exception of
> the Futterman & Circlotron drive the load as if the output tubes were
> connected in series. The Futterman & Circlotron are fundamentally
> different in that the output tubes are connected in parallel so far
> as the load is concerned. I like to picture the Circlotron as a pair
> of cathode followers connected back to back across the load.
> If you were to draw the AC equivalent circuit, that is what you
> would see.
It is correct that the circlotron is like two tubes connected back to back
across the load, but it is not true that they are operating as pure
cathode followers.
> That still gets you a gain of somewhat less the one.
This conclusion is incorrect, and is based on an incomplete understanding
of the operation of the circlotron circuit. If you look at the circlotron
circuit carefully, you will see that the output stage drive is applied
between the grid of each tube, and the center tap of the output
transformer, this causes the output stage gain to be about two, not one.
The function of the center tap on the circlotron output transformer, is to
provide this reference point for the output stage drive Voltage, output
stage currents do not flow through the center tap, and it is sometimes
replaced by a pair of resistors.
> All of the gain of a Cathode Follower is applied as local feedback.
> So 20 log 10 is 20db (local).
>
> From the above I don't think you could rearrange the parts of
> a Circlotron to end up as a McIntosh. The series vs. parallel
> connexion to the load makes it unlikely.
Perhaps it is unlikely, but it's true nevertheless. With the power
supplies shorted, the McIntosh transformer has two center tapped windings,
one in the cathode circuits, and one in the plate circuits, with both
center taps grounded for AC. For audio frequencies, where the leakage
inductance is insignificant, the two windings are effectively in parallel,
with the plate of one tube in phase with the cathode of the other tube.
The two center tapped windings can conceptually be replaced with one
center tapped winding, as in the circlotron, for audio frequencies below
the point where transformer leakage inductance enters the picture. In
fact some other circuits of this type actually use capacitors to tie the
ends of the two windings together, to help mitigate the effects of leakage
inductance. Re drawn this way, the circlotron and McIntosh circuits are
seen to be identical from an AC perspective, at frequencies below the
point where the transformer leakage inductance comes into play.
> There are quite a few out there questioning the bootstrap since it
> is positive feedback. I usually use something like a 6SN7 as a driver.
> The gain can never be more then the mu of the tube. I've
> typically measured gain increase of about one db.
> What is does is increases the available plate swing of the driver
> by about 2X. Thats what you need to drive the output stage
> of a McIntosh or a Circlotron without benefit of an
> interstage transformer.
This sounds pretty much on the money. Only the plate resistor, and
sometimes the following grid resistor are bootstrapped, the internal plate
resistance of the driver tube is not affected by the bootstrap, so the
effect s similar to replacing the plate resistors with constant current
sources. Is a constant current source load "positive feedback"? In any
case, the fact that the driver gain may be increased by bootstrapping
doesn't alter the fact that the output stage gain by itself is still only
two.
John Byrns
"BFoelsch" <BFoe...@no.spam.att.net> wrote:
> I became acquainted with McIntosh when I was doing lacquer disc recording.
Would it be getting too personal to ask what time period you are talking about?
> The Mac amps sounded pretty much as good as the best then available, but I
> liked the 50W-2 because it was unconditionally stable under any reasonable
> conditions (and many unreasonable conditions.) The disc cutting heads were
> difficult loads, and standard practice was to make them more difficult by
> connecting them to the amplifier through an RC network that gave you the LF
> crossover you wanted. No problem for the Mac, but a lot of generic NFB
> amplifiers had troubles.
What was the nature of these RC networks? They don't sound like they
would be qualitatively much different than the 300 Hz or 500 Hz frequency
dividing network for a high frequency horn speaker. I wonder why one was
a problem for a lot of amps, when the other apparently wasn't? Any idea
why a low level network wasn't used to achieve the desired recording
crossover? Seems like that would even save a little amplifier power, and
give more headroom? But I suppose it was something gross, like the desire
to connect a monitor speaker directly across the output of the cutting
amp?
John Byrns
anton...@hotmail.com (Anton Elron) wrote:
> g...@nothome.com (guy) wrote in message
news:<3c4123d1...@news.telus.net>...
> > >> The MC30 and MC40 uses the 6L6's which is usually associated with
> > >> guitar amps.
>
>
> There were no electric guitars when the 6L6 was invented. It was
> fifteen years before they were even heard of and then it was someone
> with a lap steel playing Hawaiian music in a music store orchestra.
> That was how amateur musicians aspired to aduation and nookie back
> then. Times have changed...(?)
I don't know anything about "electric guitars", or when they were
invented, although 15 years after the 6L6 sounds a little late, as
"electric guitars" were well entrenched not long after that. I do know
that amplifiers were commonly used on acoustic guitars around the time of
the 6L6.
JOHN L STEWART
McIntosh & Circlotron.
You are of course correct in pointing out that there are some pentodes capable
of more gain then the
examples I chose. However, you won’t find many 25L6’s or EL84’s in either of
the subject circuits
altho they would work OK. You would need to cut those gain figures by about 25%
when used in a PP circuit as well. EL 34’s for sure must have been used at some
point. With the gains you show there would be even more local feedback then I
had indicated. That’s a plus in my opinion.
The requirement of a vastly different ( 4 to 1 ) load impedance is another
clue that the circuits a quite different.
Anyway, my career path did not bring me in contact with a McIntosh let alone
own one. I often thought it
would be interesting to try the circuit but back in those days money was short.
I couldn’t afford the fancy
output transformer, so here we are. I like many others assumed that the output
stage gain would be two (2).
That is because it looks a lot like a split load phase inverter.
After reading your reply I wasn’t sure anymore, so I did several simulations
using Electronics Workbench
software. Sure beats the soldering iron when I screwup!! Besides, I still don’t
have that transformer.
The simulations are as follows-
1) Standard PP Output
2) Circlotron
3) McIntosh
4) SE Split Load
In every case I used one section of a 6BX7 for each of the tubes required.
There are no pentodes in my software.
They are all supplied by 250 volts & biased at 18 volts.
I avoided needless complication of a driver stage by using 1:1 input
transformers.
Look carefully & you will see that the load in every case is connected to a 1:1
transformer
whose primary & secondary AC potential are equal.
There are no multiple winding transformers in my software.
The Mac transformer is simulated by the two transformers with their secondaries
connected.
If you don’t think it will work in the case of the McIntosh, think again.
Crowhurst published at least three circuits in the 50’s based on this topolgy.
I have copies of them all here. I intend to build my version this spring.
Results-
Gain of the push pull stage is about 8X as you would expect.
Gain of the Circlotron shows about 0.8X……..( 799.1 / 997.7 ) , so there is our
20 db nfb. [ 20 log (8 / 0.8)]
Gain of the McIntosh shows about 0.824X……( 822.2 / 997.3 ) , about the same as
the Circlotron.
HOW COME!!! That last number was a surprise to me. After checking all my
connexions,
I checked another simulation using 2A3’s. The outcome was the same.
I had done an article for AudioXpress covering a line amp that was different
(June 2001).
It used a SE Split Load output, a seldom seen arrangement. It is like having
one half of a Mac
output stage. A simulation of that also showed a gain of less then one as I had
seen
during development of the article. In the schematic the parallel coupling of
the secondaries
does the same thing as magnetic coupling in an ideal transformer.
I think what has happened here is that many including myself had looked at the
Mac
output stage & made some incorrect assumptions.
The only coupling in a split load phase inverter is thru the tube.
Here we have a circuit where there is magnetic coupling as well.
That gets us negative voltage feedback rather then negative current feedback.
The outcome is vastly different.
I have posted the four simulations on alt.binaries.radio.pictures
& alt.binaries.pictures.radio for all to see.
Don't know why but I can't seem to get schematics onto that binaries location.
Comments from anyone please. John L Stewart
Russ Sadd
> "Fabulous distortion figures'? Patrick what in the world are you talking
> about?
> The MC-30 claims to provide 30 W at 0.3% distortion which *for that era*
> from the 50s that could do that!
1947. Williamson amplifier. <0.1% distortion @ 15W.
1948. Acoustical Manufacturing Co. (Quad)QA/12P. <0.1% distortion @
10W.
1948. Leak TL/12 "Point One". <0.1% distortion @ 15W.
1952. Acoustical Manufacturing Co. (Quad) Q.U.A.D. <0.25% distortion @
12W.
1953. Leak TL/25A. <0.1% distortion @ 25W.
1957. Quad II. <0.1% distortion @ 15W.
Just to name a couple of popular amplifiers from this time! Plus, of
course, all of these were operating in Class A (not Class B2 like the
McIntosh), and featured the minimum 20-20KHz +/- 0.5dB or better
frequency range.
Best regards,
Russ Sadd
(Heidelberg, Germany)
Russ Sadd
> Correct me if I'm wrong John, but didn't the leak Point One occur *after*
> the MC-30 & MC-40?
The MC-30 came out in 1954, and the MC-40 came out in 1962. These may
be compared to the advent of the "Quad" QA/12P, Leak "Point One", and
Williamson amplifier from the late 1940s. Although if we're talking
*professional* rather than *domestic* amplifiers, I could also cite
something like the Leak 15W Type 15 from 1945, also with <0.1%
distortion.
By the time of the MC-40, Quad had already moved on to solid-state,
and we were nearing the end of the Leak valve amplifiers (up to the
Stereo 50 or TL/50+, for example).
Patrick Turner
Russ Sadd wrote:
Well, the Mac 275 put out 75 watts of B2 power, and hadabout 0.2 % of THD, and what tubed Quad or Leak came near
that?
Maybe you could say the mono Leak 50? watt did,
I repaired one which made 65 watts class AB1.
They had 550 v B+, and fixed bias.
The MC75 could be expected to make very tiny amounts of THD
at 15 watts, or at 2 watts, which is where many folks used them for
domestic sound.
But in commercial applications, Macs were popular.
It is possible to get far more power from class B2.
Audio Research make amps with 16 x 6550 and make 600 watts.
This is possible because one can get 75 watts from two 6550 pretty easily,
even in AB1.
It is possible, and was often done that 2 x EL34 were set up with 900
plate volts, and 450 screen volts, and 20 mA of idle current, and about
12,000 ohms RL, and you get about 100 watts AB1 from 2 tubes.
It is listenable, even enjoyable for PA. Not really all that practical
for domestic hi-fi.
The class A sound is of course better than struggling class B sound.
And a Mac with lower plate voltage and more idle current
for increased class A might be hard to beat.
All the ancient Leaks that I have tested are indeed able to
get THD down to 0.1%, but only on a sunny day,
with the wind behind them.
When you test one that's done a few years, and needs a service,
expect far worse THD results.
The amount of loop NFB is high to be able to get down to 0.1%.
Not much has to be wrong for the figure to be 0.3%.
At a watt, though, they ain't bad, even when a bit old,
and this is due to the class A action.
The problem I have with old Leaks is the stability is nothing like what
some of the reviews I have seen say. I have found it to be somewhat
non optimal.
I think it is easy to make a better sounding amp than a Quad II.
I compared a mint pair with a simple triode amp I made using very
plain parts, and 6CM5 output tubes in class AB, not all class A,
and the CM5 amp was better sounding.
Members of my father's generation who were into DIY
routinely ran up excellent tube amps based on 807s and
locally made aussie made A&R or Fergusson output transformers.
Nobody ever cared a hoot about measuring distortion.
Speakers of the day were sensitive, and so a 25 watt amp
was as powerful as a 100 watt amp today, to get deafening sound levels.
The Kriesler Radiogram mum bought in 1963 had two SE EL84 pentode
stages, SFA feedback, pehaps 8 dB, and provided very definitely
SE type of sound, which was a little short on quality above a low level, but
that's where mum liked it because her hearing was razor sharp
and she knew what us kids were up to even when we were outa sight.
For dinner music, the record playing ability of such junk was entirely
adequate. Ella Fitzgerald and Satchmo sounded great to us.....
Every joke the Goons came out with on the AM radio was clear as a bell....
If only we'd stop laughing, so we'd get the next one.....
I still have the chassis, completely rewired, and it sounds very well,
but the awful Rola speakers with detatchable magnets are gone into a box
at the back of the shed someplace, insects ate the cones a bit....
The brandnames of yesterday never ever had a monopoly on good
amplifer performance, it was always realisable by DIY.
Quite a few blokes built their own tubed B&W TVs
when TV started in 1956, as new TVs were darned expensive,
but not impossible for the enthusiast to make,
along with oscilliscopes, and a whole range of radio gear.
But nobody here did much with FM, as public programme transmissions
didn't start until 1972, so no FM radios were built by DIY folks.
By 1972, the early tube era was kaput, and we were flooded by
cheap Japanese gear when the tariffs were abolished, which took out
the local industry.
I remember completely gutting and rebuilding an old
ex army tranciever to my own simpler design in about 1962,
when I was at early high school. That beat doing school homework,
and radio got me into all sorts of troubles.
The Home Brew amps allowed excellent sound from the new LP source.
And really keen members of that generation built a second amp
when stereo appeared, although with reluctance sometimes,
because the care of the baby boomer generation was in full swing,
and TV came in, and the rot of flopping passively in front
of some live idiot box began, when I Luv Lucy, and sit coms began.
One cannot help but ramble on about the nostalgic past sometimes.
Patrick Turner.
John Byrns
Hi John,
Sorry, I didn't realize you were only talking only about the gain of
pentodes that have been used in the circlotron circuit. In my defense, I
would point out that the EL84 was used in the ElectroVoice A15 circlotron
amplifier, and it has a gain of 58.76 using your formula, considerably
above the 8 to 15 gain range you mentioned. You also quoted the gain of
the 6K6, what circlotron design uses this tube?
Your simulations were very interesting, and while they contain several
inconsistencies, they do point out the reason I claim the
Circlotron/McIntosh output stage has a gain of about two, while you claim
a gain of about one for the Circlotron/McIntosh output stage. Our basic
difference is that I am calculating the gain considering the input to be
the Voltage between one grid and ground, and the output to be the Voltage
measured between the plate and cathode of one tube, while you consider the
input to be the Voltage from grid to grid of a push pull stage, and the
output to be different things for different circuits. Measuring the input
grid to grid, instead of grid to ground seems just as legitimate as my
method for a push pull stage, accounting for the factor of two difference
in our results. My gain calculation is based on the single ended
equivalent of the circuit, which in the case of both the circlotron and
McIntosh is what you are calling the "SE Split Load" circuit.
Now, some of the problems I see with your simulations. First, we need to
make a note that while we were talking about pentode amplifiers, you used
a triode in you simulations, which will impact the results to some
extent. A bigger problem is that your circuits are not consistent in the
load they place on the tubes, using three different values of load
resistance. Each tube in the "Standard PP Output" sees a 5k Ohm load
resistance, while each tube in the "Circlotron" and "McIntosh" circuits
sees a 20k Ohm load resistance, and the tube in the "SE Split Load"
circuit sees a 40k Ohm load resistance.
Your simulation of the "SE Split Load" is seriously flawed by the fact
that you are only measuring the output Voltage across half the output
transformer primary winding, which makes no sense at all for a "SE"
circuit, and cuts the gain to half of what it actually should be.
Your "McIntosh" circuit is conveniently drawn in a way that makes it easy
to see how it is identical with the circlotron, from an AC signal point of
view, when leakage inductance is neglected. You can see this by first
shorting out the power supply, which doesn't change the circuit from the
AC point of view, connecting the center tap of your right-hand output
transformer to ground. Next, since the secondaries of the two
transformers are already connected in parallel, the corresponding ends of
the primaries of the two transformers will be at the same AC potential,
ignoring leakage inductance effects, and can also be paralleled. Once we
have both the primaries, and the secondaries of the two transformers
connected in parallel, we can replace the two transformers with a single
transformer. At that point the McIntosh and Circlotron output stages are
seen to be identical from an AC point of view.
The McIntosh version of this circuit has the advantage over the circlotron
of only requiring one power supply, while a stereo circlotron would
require four seperate power supplies for the output stage alone. The
McIntosh circuit has the disadvantage of requiring a complex output
transformer with low leakage inductance, which is not required by the
circlotron, where there is only a single primary winding, and the center
tap does not carry any output stage currents, even in class B operation,
and can be replaced by a pair of resistors if desired.
Regards,
John Byrns
In article <3C424261...@sympatico.ca>, jh.st...@sympatico.ca wrote:
> Hello John- Thankyou for your thoughtful response to my short comparison =
> of the
> McIntosh & Circlotron.
> You are of course correct in pointing out that there are some pentodes ca=
> pable
> of more gain then the
> examples I chose. However, you won=92t find many 25L6=92s or EL84=92s in=
> either of
> the subject circuits
> altho they would work OK. You would need to cut those gain figures by abo=
> ut 25%
> when used in a PP circuit as well. EL 34=92s for sure must have been used=
> at some
> point. With the gains you show there would be even more local feedback th=
> en I
> had indicated. That=92s a plus in my opinion.
>
> The requirement of a vastly different ( 4 to 1 ) load impedance is anothe=
> r
> clue that the circuits a quite different.
>
> Anyway, my career path did not bring me in contact with a McIntosh let al=
> one
> own one. I often thought it
> would be interesting to try the circuit but back in those days money was =
> short.
> I couldn=92t afford the fancy
> output transformer, so here we are. I like many others assumed that the o=
> utput
> stage gain would be two (2).
> That is because it looks a lot like a split load phase inverter.
>
> After reading your reply I wasn=92t sure anymore, so I did several simula=
> tions
> using Electronics Workbench
> software. Sure beats the soldering iron when I screwup!! Besides, I still=
> don=92t
> have that transformer.
>
> The simulations are as follows-
> 1) Standard PP Output
> 2) Circlotron
> 3) McIntosh
> 4) SE Split Load
>
> In every case I used one section of a 6BX7 for each of the tubes required=
> =2E
> There are no pentodes in my software.
> They are all supplied by 250 volts & biased at 18 volts.
> I avoided needless complication of a driver stage by using 1:1 input
> transformers.
> Look carefully & you will see that the load in every case is connected to=
> a 1:1
> transformer
> whose primary & secondary AC potential are equal.
> There are no multiple winding transformers in my software.
> The Mac transformer is simulated by the two transformers with their secon=
> daries
> connected.
> If you don=92t think it will work in the case of the McIntosh, think agai=
> n.
> Crowhurst published at least three circuits in the 50=92s based on this t=
> opolgy.
> I have copies of them all here. I intend to build my version this spring.=
>
>
> Results-
> Gain of the push pull stage is about 8X as you would expect.
> Gain of the Circlotron shows about 0.8X=85=85..( 799.1 / 997.7 ) , so the=
> re is our
> 20 db nfb. [ 20 log (8 / 0.8)]
> Gain of the McIntosh shows about 0.824X=85=85( 822.2 / 997.3 ) , about th=
> e same as
> the Circlotron.
>
> HOW COME!!! That last number was a surprise to me. After checking all my
> connexions,
> I checked another simulation using 2A3=92s. The outcome was the same.
>
> I had done an article for AudioXpress covering a line amp that was differ=
> ent
> (June 2001).
> It used a SE Split Load output, a seldom seen arrangement. It is like hav=
> ing
> one half of a Mac
> output stage. A simulation of that also showed a gain of less then one as=
> I had
> seen
> during development of the article. In the schematic the parallel coupling=
> of
> the secondaries
> does the same thing as magnetic coupling in an ideal transformer.
>
> I think what has happened here is that many including myself had looked a=
> t the
> Mac
> output stage & made some incorrect assumptions.
> The only coupling in a split load phase inverter is thru the tube.
> Here we have a circuit where there is magnetic coupling as well.
> That gets us negative voltage feedback rather then negative current feedb=
> ack.
> The outcome is vastly different.
>
> I have posted the four simulations on alt.binaries.radio.pictures
> & alt.binaries.pictures.radio for all to see.
> Don't know why but I can't seem to get schematics onto that binaries loca=
> tion.
>
> Comments from anyone please. John L Stewart
>
John Byrns
<in...@turneraudio.com.au> wrote:
[Snip]
>
> The MC75 could be expected to make very tiny amounts of THD
> at 15 watts, or at 2 watts, which is where many folks used them for
> domestic sound.
> But in commercial applications, Macs were popular.
> It is possible to get far more power from class B2.
This raises a question I have often wondered about, do the MC's actually
run in B2, or even AB2 if you will? That wimpy 12AX7 cathode follower
doesn't seem like it could drive the grids very far positive, making the
amp effectively B1, or AB1, depending on where you draw the line between
them?
Regards,
John Byrns
Patrick Turner
John Byrns wrote:
> In article <3C42CB54...@turneraudio.com.au>, Patrick Turner
> <in...@turneraudio.com.au> wrote:
>
> [Snip]
> >
> > The MC75 could be expected to make very tiny amounts of THD
> > at 15 watts, or at 2 watts, which is where many folks used them for
> > domestic sound.
> > But in commercial applications, Macs were popular.
> > It is possible to get far more power from class B2.
>
> This raises a question I have often wondered about, do the MC's actually
> run in B2, or even AB2 if you will? That wimpy 12AX7 cathode follower
> doesn't seem like it could drive the grids very far positive, making the
> amp effectively B1, or AB1, depending on where you draw the line between
> them?
As far as I know, the peak current ability of the AX7 is way above twice
theidle current, and would be able to drive the grids positive, and with the
force
of the boosted error voltage behind them,
from the action of the global NFB, the amp eventually
goes into AB2 action for the last few watts.
Tim De Paravicini designed an amp with 10 x 12AX7 for a power amp,
operating in the ridiculously termed "enhanced triode mode",
ie, AB2, and got 25 watts, which is like getting 2.5 watts AB2 per tube.
Next time you get a chance, build a simple driver amp for a couple
of 6L6 tubes in triode, and include a direct coupled CF stage,
and you will find that where you may only get 18 watts class AB1,
with AB2, you will get 32 watts, as much as if you used AB1 UL, or pentode.
With AB2 pentode, 80 watts is available with 6L6 with B+at 600 v,
and EG2 at 300.
Patrick Turner.
Russ Sadd
> Well, the Mac 275 put out 75 watts of B2 power, and hadabout 0.2 % of THD, and what tubed Quad or Leak came near
> that?
Very likely one of the PA / cinema variants, but this is beside the
point. As far as *domestic* amplifiers go, the goal is first and
foremost sound quality rather than output power. As long as an
amplifier has enough Class A power output to comfortably drive good
speakers in the home, we don't need to start compromising quality with
PA output stages.
> I think it is easy to make a better sounding amp than a Quad II.
Of course, the Quad II is a commercial design and therefore quite
possible to better by throwing money at it. Bearing in mind, though,
that Quad IIs are rather good amplifiers in their own right, and this
kind of "betterment" will take some serious thought and effort.
> The brandnames of yesterday never ever had a monopoly on good
> amplifer performance, it was always realisable by DIY.
As you already know, I've gone down that route myself, since I regard
the Quad II's output stage as being particularly elegant, and
therefore used it - and the Williamson design (which is still superb
by modern standards) - as the basis for my own ideas. For example,
replacing the valve phase-splitters with a coupling transformer, using
over-designed output transformers, and a fully-regulated power supply
of the fiendishly optmised kind from Richard Sears. Going overboard
with polished stainless steel for the chassis was a nice finishing
touch.
And yes, it sounds damn good indeed! But I should add that this kind
of build quality didn't come cheap, and would certainly place this amp
in a comparative price bracket to the new Quad II-40s or the Mac 275
reissues (not that I'm terribly impressed with either and can see far
too many shortcomings and cost-cutting happening in both). Just
kludging together some off-the-shelf transformers might be a very
cost-effective alternative, but my point is, that a seriously good DIY
amp is hard to do and soon becomes quite expensive purely from the
cost of components (bearing in mind that everything in mine is pretty
much 'trade' price, not 'over the counter').
For some superb examples of what can be done with carefully
thought-out designs and good engineering, take a look at
http://members.tripod.com/richard984 - and things that leave my own
over-the-top approach looking decidely minimalist!
Patrick Turner
Russ Sadd wrote:
> Patrick Turner <in...@turneraudio.com.au> wrote in message news:<3C42CB54...@turneraudio.com.au>...
> > Well, the Mac 275 put out 75 watts of B2 power, and hadabout 0.2 % of THD, and what tubed Quad or Leak came near
> > that?
>
> Very likely one of the PA / cinema variants, but this is beside the
> point. As far as *domestic* amplifiers go, the goal is first and
> foremost sound quality rather than output power. As long as an
> amplifier has enough Class A power output to comfortably drive good
> speakers in the home, we don't need to start compromising quality with
> PA output stages.
>
> > I think it is easy to make a better sounding amp than a Quad II.
>
> Of course, the Quad II is a commercial design and therefore quite
> possible to better by throwing money at it. Bearing in mind, though,
> that Quad IIs are rather good amplifiers in their own right, and this
> kind of "betterment" will take some serious thought and effort.
Nope, the point I have been making in postings often is thatit is easy for DIY to get better sound than a Quad II
provides.
I do question the rightness of the design, when with a little effort,
it could have been streets ahead.
I use some of their circuit ideas quite freely, but the way Quad
implemented their product leaves a lot to be desired.
Any discerning examination of the output tranny the Quad II
is lumbered with bears out what I am saying.
Given some iron, wire, sheet metal, and a few tubes,
one can easily do better with DIY, especially when we way
up the cost of a new Quad II.
I am very nostalgic and sentimental and fond of these
lovely old british bangers, and I don't want to upset
fans of this type of hardware, but too often I have heard better.
> > The brandnames of yesterday never ever had a monopoly on good
> > amplifer performance, it was always realisable by DIY.
>
> As you already know, I've gone down that route myself, since I regard
> the Quad II's output stage as being particularly elegant, and
> therefore used it
I find myself thinking the output stage was a way to get onto the triodebandwagon by using NFB with pentodes.
Quad's implementation is to me a very poor version of
local CFB. When you use the 8 ohm tap connection, there is 17 %
of copper losses, and this is quite crummy.
Many of these amps exibit very out of balance DC in the output tubes,
exacerbated by the high value of grid bias resistors,
and then there's a common cathode resistor, itself an idea
I don't see much merrit in.
Leaks' idea of seperate cathode resistors, and bypass caps
regulates the DC in the output tubes far better.
The Quad OPT is just far too much of a toy to be regarded
as something magnificent.
Onset of saturation is at 40 Hz at full power....
Performance into a cap load, not too good...
> - and the Williamson design (which is still superb
> by modern standards) - as the basis for my own ideas. For example,
> replacing the valve phase-splitters with a coupling transformer, using
> over-designed output transformers, and a fully-regulated power supply
> of the fiendishly optmised kind from Richard Sears. Going overboard
> with polished stainless steel for the chassis was a nice finishing
> touch.
Agreed.I find the Williamson driver amp as originally configured to be
very easy to get right for the DIY person, and it is a very
fast performer, with higher bandwidth than most other
types of drive amps, as the concertina is a good buffrer between
the input stage and the balanced amp driver.
This driver has a reasonably low output Z when each
individual side of the driver is considered, compared to the
Mullard 520, and the Leaks, or modern types as my own
with a CCS in the LTP tail.
The output Z of the LTP is only low when both tubes
in it see the same load, and some would say this is not the case
at HF , say above 20 kHz when a cap load is used on the amp.
Maybe they split hairs, but high current drivers are supposed to sound better,
as they are more immune to capacitive effects.
In an amp I have with 4 x 6550 per channel, with 12.5 % of
local CFB a la Quad, as well as UL screen taps, I have gone to using
EL84 in triode with 15 mA per tube to drive the output tubes.
The operation observed on the CRO is little different to the use I used to have
with a 6CG7 instead of the EL84.
But why not try things?
> And yes, it sounds damn good indeed! But I should add that this kind
> of build quality didn't come cheap, and would certainly place this amp
> in a comparative price bracket to the new Quad II-40s or the Mac 275
> reissues (not that I'm terribly impressed with either and can see far
> too many shortcomings and cost-cutting happening in both). Just
> kludging together some off-the-shelf transformers might be a very
> cost-effective alternative, but my point is, that a seriously good DIY
> amp is hard to do and soon becomes quite expensive purely from the
> cost of components (bearing in mind that everything in mine is pretty
> much 'trade' price, not 'over the counter').
Well, good luck if you can get "name brand" type of pricesfor your gear. Prices for Quad, Mac, CJ, Audio Research,
do seem to have little relation to the cost of materials and a margin
for wages.
>
>
> For some superb examples of what can be done with carefully
> thought-out designs and good engineering, take a look at
> http://members.tripod.com/richard984 - and things that leave my own
> over-the-top approach looking decidely minimalist!
Very interesting.Unfortunately I found trying to see the schematics was barred from me
by my capricious PC, or some other link gremlin,
so I cannot comment on text claims and circuit operation.
Patrick Turner.
Bob Groschen
"Russ Sadd" <gri...@ntlworld.com> wrote in message
news:279deb62.02011...@posting.google.com...
Hi Russ,
Thanks for the info. I've never heard of Acoustical Manufacturing but I
*have*
heard of Quad. BTW, while your list is impressive, none of these amps come
close to 30 W at 0.3% distortion. The Leak TL/25A comes closest at 25 W.
Just measuring my MC-30's I can tell you that over most of the frequency
range,
the MC-30 will produce 30w at less that 0.1%. It's only below about 100 Hz
that the distortion rises to the 0.3% level or beyond. TL/25A may just
equal the
MC-30, if driven to 30W but not owning one, I can't make those tests.
Bob Groschen
> The amount of loop NFB is high to be able to get down to 0.1%.
I visited the Leak website and was amused to see Leak litteraly
"boast" of the large amount of NFB contained within their three
loops. I thought this would be antithema for the NFB crowd.
> The Kriesler Radiogram mum bought in 1963 had two SE EL84 pentode
> stages, SFA feedback, pehaps 8 dB, and provided very definitely
> SE type of sound, which was a little short on quality above a low level,
but
> that's where mum liked it because her hearing was razor sharp
> and she knew what us kids were up to even when we were outa sight.
Is this a latter-day confession Patrick? :-)
> I remember completely gutting and rebuilding an old
> ex army tranciever to my own simpler design in about 1962,
> when I was at early high school. That beat doing school homework,
> and radio got me into all sorts of troubles.
Ah ha! There *is* a little bit of rabble-rouser in old Patrick after all!
:-)
> One cannot help but ramble on about the nostalgic past sometimes.
That's Ok Patrick. At least yours are usually informative.
Choky
various legends.....................;))
--
Choky,Prodanovic Aleksandar,YU
my reply address is tweaked!!
Bob Groschen <bobgr...@peoplepc.com> wrote in message
news:u498gem...@corp.supernews.com...
>
***********carefully snipped*****************
Russ Sadd
> *have*
> heard of Quad.
The "Acoustical Manufacturing Co." changed its name to "Quad" in the
1950s. *grin*
> BTW, while your list is impressive, none of these amps come
> close to 30 W at 0.3% distortion. The Leak TL/25A comes closest at 25 W.
> Just measuring my MC-30's I can tell you that over most of the frequency
> range,
> the MC-30 will produce 30w at less that 0.1%. It's only below about 100 Hz
> that the distortion rises to the 0.3% level or beyond. TL/25A may just
> equal the
> MC-30, if driven to 30W but not owning one, I can't make those tests.
My point, however, is that these were *domestic, high-fidelity, Class
A* amplifiers. I would call the Mac a "public address" design (is the
MC-30 Class A?) - and I've got quite a nice Wharfedale-made 50W Class
A high-fidelity amp intended for PA use from circa 1954! (The Leak
TL/25A is pure Class A as well and conservatively rated.) Power output
isn't really that important to go on, and I still hold with the
Williamson design and a massive regulated power supply as the ideal.
Bob Groschen
> My point, however, is that these were *domestic, high-fidelity, Class
> A* amplifiers. I would call the Mac a "public address" design (is the
> MC-30 Class A?)
No, the McIntosh topology isn not Class A. Ironically, I measured the
idle current through the 6L6's but misplaced the piece of paper I made
the notes on (there's a lesson for you: *always* use your notebook!)
So I can't tell you exactly what class they're running, but Class B seems
a little extreme.
OTOH, we'll have to agree to disagree about the "public address"
term. The McIntosh amps all came with two taps on the power transformer:
One for 117 V and one for 125 V. Consumer applications used the 117 V
tap while commercial applications used th 125 V tap. I seriously doubt
that the mains were running that high back then so it turned out to be a
neat way to "derate" the amplifier for more abusive environments.
Class A holds no magic for me and I find your affection for it curious.
> - and I've got quite a nice Wharfedale-made 50W Class
> A high-fidelity amp intended for PA use from circa 1954! (The Leak
> TL/25A is pure Class A as well and conservatively rated.) Power output
> isn't really that important to go on, and I still hold with the
> Williamson design and a massive regulated power supply as the ideal.
I readily concede your point on the Williamson amp, but I'll have to
temper your comments about the Leaks with Patrick Turner's observations.
Patrick Turner
Bob Groschen wrote:
> "Patrick Turner" <in...@turneraudio.com.au> wrote in message
> news:3C42CB54...@turneraudio.com.au...
>
> > The amount of loop NFB is high to be able to get down to 0.1%.
>
> I visited the Leak website and was amused to see Leak litteraly
> "boast" of the large amount of NFB contained within their three
> loops. I thought this would be antithema for the NFB crowd.
Its always good to be able to boast about loop after loopof feedback, ( if the
date is 1955 )
Probably wise to shut up about how many feedback tricks one tries
to perform today.
But the loops Leak had were included the two from output transformer
to output tube screens, ie normal plain garden variety ultralinear,
and the global feedback path, used by 95 % of makers then, and ever
since, and in SS amps.
The amount of total feedback expressed in terms of gain reduction
in Leaks, Quads, and Dynaco, is far less than the typical
Mac, which had a lot of NFB in the output stage,
and then another lot around the global circuit.
No wonder they measure fair.
>
>
> > The Kriesler Radiogram mum bought in 1963 had two SE EL84 pentode
> > stages, SFA feedback, pehaps 8 dB, and provided very definitely
> > SE type of sound, which was a little short on quality above a low level,
> but
> > that's where mum liked it because her hearing was razor sharp
> > and she knew what us kids were up to even when we were outa sight.
>
> Is this a latter-day confession Patrick? :-)
Of course not, I was well behaved at all times.Bob, how could you think such
farnarcling suspicious thoughts.
But seriously, folks back in 1960 people didn't need much volume,
their hearing wasn't so challenged as many folks today.
But in all aspects of life, they got by on less,
as their experience was tempered by the war years, and
the Depression before that....
Things only changed when Buddy Holly was let loose in studios,
and somebody got the idea that bass wasn't sinful.
> > I remember completely gutting and rebuilding an old
> > ex army tranciever to my own simpler design in about 1962,
> > when I was at early high school. That beat doing school homework,
> > and radio got me into all sorts of troubles.
>
> Ah ha! There *is* a little bit of rabble-rouser in old Patrick after all!
> :-)
I better not say from which large cache of nearly pristineold cathedral and
other radios we got all our spares from....
Every time my mate's dad had a bother with a radio, under the house it went.
He was a bank manager, and he could easily afford the rampant luxury
of new radios, and nice cars even, and a nice house, in a nice suburb.
He deserved it, he'd fought hard in Tripoli, and in New Guinea....
Now that I think about that, I don't remember telling
the priest in confession about it, and
I never ever thought I should have.
Tellya what, I'll go down the local church, and sort it all out.
Trouble is, the padre down there likes old radios,
and it'll take a year to say my penance.
Patrick Turner.
Patrick Turner
Bob Groschen wrote:
We all know about the Williamson design, it is in the RDH4,with all the
trannie details.
Now just who, if anyone, in the olde worlde of tube amp makers,
actually built a really and truly accurate Williamson?
I suspect Williamson's design was a little too much for most of the
bean counters of the day.
Then there are all the preamps he designed.
They were rather full of switches and do-das & tone
shaping controls, which now
lead me to my side cutters.....
But people wanted gear to allow replay of shellac records,
which were noisy, such was the hunger
for some musical diversion of an evening.
I just cannot remember anyone needing a preamp
for a low output MC.
If silver cabling was mentioned, they would phone
for the guys at the asylam to come pick you up....
Patrick Turner.
Bob Groschen
> > Is this a latter-day confession Patrick? :-)
>
> Of course not, I was well behaved at all times.Bob, how could you think
such
> farnarcling suspicious thoughts.
Oh, of *course* you were! A perfect angel. NOT! :-)
> I better not say from which large cache of nearly pristineold cathedral
and
> other radios we got all our spares from....
We can keep a secret! Can't we guys (and gals)? :-)
> Every time my mate's dad had a bother with a radio, under the house it
went.
> He was a bank manager, and he could easily afford the rampant luxury
> of new radios, and nice cars even, and a nice house, in a nice suburb.
The truth comes out!
> Now that I think about that, I don't remember telling
> the priest in confession about it, and
> I never ever thought I should have.
> Tellya what, I'll go down the local church, and sort it all out.
> Trouble is, the padre down there likes old radios,
> and it'll take a year to say my penance.
Well Patrick, the sooner you start, the sooner you'll finish. :-)
Patrick Turner
Bob Groschen wrote:
Do you think I should include the bit about shooting at thebulbous rare hard
to get tubes we found, with an air rifle?
or should we just stare into space when asked, "and is that all?",
by the disgusted padre.
Patrick Turner.
John L Stewart
Go to alt.binaries.radio.pictures or alt.binaries.pictures.radio for
schematics.
The following text is included there as well.
First of all let me tell you I have no favorite circuit. I've tried many.
My interest is in circuit design. Here is some information I hope you will
find useful.
In order to make a valid comparison I kept the differences of
components used to an absolute minimum.
As a result in each circuit we have the following-
I used triodes in all the simulations in order to keep things simple.
Every tube used is one-half of a 6BX7.
They are all supplied by 250 volt power supplies & all biased at -18 volts.
A check of the cathode currents shows them all to be about 38 mA.
They are all driven by a 10 volt, 1000 Hertz signal source, through a 1:1
input transformer.
The total load for each one-half 6BX7 in series with it's power supply is
4K. (Two 2K loads)
Loads are all connected to the transformer primaries. Some may question
this.
However, that's how the tube manufacturers measure power output of their
products.
All of the output transformers are 1:1 so there are no circulating currents
in the interconnected
secondaries. Since the secondaries are all open circuited they have no
bearing on the final result.
A comparison of the common PP output schematic with the McIntosh reveals a
definite similarity.
If you look carefully at the two schematics you will see that they match
exactly except for a few
connexions at the output. That is because I generated the schematics, one
from the other to save time
& to better illustrate their similarity. Also notice that all of the
simulations work.
If there were an error in the circuit the simulations would produce odd
results. These didn't.
In both the common PP & the McIntosh the load is driven by the output tubes
in a series connexion.
The loop passes through both output tubes & the four 2K loads in a complete
loop.
In the common PP case all of the load is in the plate circuit & yields a
gain of 7.66.
The windings are in parallel for the transfer of POWER, not the signal
current & voltages.
In the McIntosh one-half of the load is in the cathodes. This provides local
negative voltage feedback.
The resulting gain to each of the anode & cathode circuits is 0.793 for a
total of 1.586.
A careful inspection of the circuit will show that all four of the 2K loads
are still in series with the tubes,
even though one-half of the loads are in the cathodes.
In both the common PP stage & the McIntosh, the DC from the power supply
follows a parallel
path through the tubes & the output transformer. DC flux in the output
transformer is cancelled.
In sharp contrast to this the Circlotron circuit is quite different. The
pair of half 6BX7's are
connected to the output as cathode followers. All of the resulting signal
has been applied as
local negative voltage feedback. The resulting gain is 0.794.
It is also easy to see that those two half 6BX7's are driving their two 4K
loads in parallel.
I included both 4K resistors in the schematic to illustrate that. So the
load for this circuit is 2K.
That is one-quarter the load used in the other PP circuits I have covered.
It makes no difference to this last observation whether triodes or pentodes
are used.
We need two independent HV supplies, since there is no common except at the
input.
As a matter of fact, each of the two HV supplies is bodily driven by the
full audio voltage
developed at each side the output.
Another result is that no DC current flows in the output transformer.
To follow the DC current path, imagine a figure 8 laid on it's side.
The DC current of one of the half 6BX7's cancels an equal but opposite
current from the
other half 6BX7 in the output transformer.
Some of you might find these results hard to believe or perhaps in
opposition
to what you have heard. Better believe them. They are real & valid.
Did I leave anything out? Probably, but I think this makes the point I
brought up last week.
If anyone would like to challenge this, be sure to support your line of
reason with schematics
& wherever it would help with calculations as I have done here.
Remember the saying " A picture is worth a thousand words". So is a
schematic or drawing.
Why would I know all this? I have a Circlotron here which I designed from
the ground up.
It works very well. There is an article in the works that illustrates even
more ways to
hook up the output transformer.
As well I've built many other different amps over the years.
It was implied that I didn't understand how these circuits work.
Looks to me like it's the other way around. Sorry about that.
John L Stewart
Bob Groschen
> Do you think I should include the bit about shooting at thebulbous rare
hard
> to get tubes we found, with an air rifle?
Gasp! You didn't!?
> or should we just stare into space when asked, "and is that all?",
> by the disgusted padre.
That's my take on it. :-)
Philip Lawrence
How could the early Leak {TL12} be compared to the Mcintosh.
The output transformers were very poor quality, they had so much feedback they
were hardly stable. The output tubes were so close to each other they nearly
touch.
phil
Bob Groschen wrote:
> "John Byrns" <jby...@enteract.com> wrote in message
> news:jbyrns-1101...@216-80-74-2.d.enteract.com...
>
> > > Correct me if I'm wrong John, but didn't the leak Point One occur
> *after*
> > > the MC-30 & MC-40?
> >
> > Don't know, sounds reasonable, but either way it's irrelevant as you
> > didn't say anything about which came first, you just asked someone to
> > "Name another amp from the 50s that could do that!", which I did.
>
> As near as I can tell, the Leak Point One evolved from the TL-12 and did so
> around the same time as Frank McIntosh was producing the MC-30 (1950s).
>
> The TL-12 put out a "nominal" 12 watts (2X 6L6/KT-66 PP-triode) at less than
> 0.1% distortion
> and because of the triode output, had a *very* respectable damping factor.
> And, for the record,
> it claims to use 26dB of negative feedback.
Patrick Turner
Philip Lawrence wrote:
> Hi
> How could the early Leak {TL12} be compared to the Mcintosh.
> The output transformers were very poor quality, they had so much feedback they
> were hardly stable. The output tubes were so close to each other they nearly
> touch.
> phil
I don't regard nearly touching output tubes as a major defect.But yes, most Leak
amps are not real stable, but
then someone must have realised that capacitor loads were
not going to be used often with Leak amps.
In fact most speakers offer an ultimately inductive load,
left in their raw form, without Zobel networks added,
to make the load purely resistive, at HF.
McIntosh took another road, and used a lot of local NFB in the output
stage, to linearise it before adding global NFB.
McIntosh then required some form of stabilisation, as
it normally would have been impossible
to apply the same total amount of global NFB + local NFB
in a conventional Williamson, or Leak, or Quad II amp.
His answer was to wind the OPT with bifilar windings, so that
although the cathode and plate windings are at potentials which are
400 volts apart, they have the same signal voltages and these are locked
together well with extremely low leakage inductance,
which then won't cause instability problems, due to HF phase shift,
at too low a frequency, like it does in most conventional amps.
The McIntosh way of winding trannies is painful for the DIY person.
So do it the GE way, and have equal windings seperated by insulation,
and where the signal voltages are the same, but DC potentials different,
just bypass the ends of the windings with suitable
10 UF high voltage plastic caps, which will shunt any leakages.
One still has the challenge of having the drive amp produce
120 volts rms to each output tube grid, and so McI used bootstrapping,
which tends to act as positive voltage feedback in the way he used it.
One could try using a very high supply voltage, and resistors,
and small power tubes, such as the EL86, or EL84 in triode.
Nobody seems to have tried taking the cathode voltages in a
McIntosh style output stage to the screens of an ultralinear driver stage.
It may be a new way worthy of a trial, stability permitting.....
I figured that such a scheme would not need any global NFB,
and with driver tubes as EL84, with their high screen transconductance,
the THD and Zout would be nice and low.
The feedback path to the driver tube screens is short,
and around only 1 tube stage, and stability should be fine.
Patrick Turner.
John Byrns
<in...@turneraudio.com.au> wrote:
> McIntosh took another road, and used a lot of local NFB in the output
> stage, to linearise it before adding global NFB.
> McIntosh then required some form of stabilisation, as
> it normally would have been impossible
> to apply the same total amount of global NFB + local NFB
> in a conventional Williamson, or Leak, or Quad II amp.
> His answer was to wind the OPT with bifilar windings, so that
> although the cathode and plate windings are at potentials which are
> 400 volts apart, they have the same signal voltages and these are locked
> together well with extremely low leakage inductance,
> which then won't cause instability problems, due to HF phase shift,
> at too low a frequency, like it does in most conventional amps.
Hi Patrick,
I could be wrong, but I always had the impression that the reason McIntosh
went with the low leakage inductance design was not primarily for
stability reasons, but was because they run most of their amps deep into
class AB, if not class B, and the low leakage inductance was needed to
eliminate notch distortion, so that they could call it high-fidelity.
You better be careful with that talk about "the cathode and plate
windings" having "the same signal voltages and these are locked together
well with extremely low leakage inductance". That's saying the McIntosh
circuit is equivalent to the circlotron circuit from an AC signal point of
view, and John Stewart won't like that!
> The McIntosh way of winding trannies is painful for the DIY person.
> So do it the GE way, and have equal windings seperated by insulation,
> and where the signal voltages are the same, but DC potentials different,
> just bypass the ends of the windings with suitable
> 10 UF high voltage plastic caps, which will shunt any leakages.
Now you have really gone and put your foot in it, actually tying the
plates and cathodes of the opposite tubes together for AC signals with a
large capacitor, now it's clear you are trying to say that the McIntosh
and circlotron are equivalent for AC signal purposes. You better duck
quick.
John Byrns
Stewart" <jhste...@hotmail.com> wrote:
> Read on & learn the real goods on the McIntosh vs. Circlotron topology!!
[much snippage]
> It was implied that I didn't understand how these circuits work.
> Looks to me like it's the other way around. Sorry about that.
Hi John,
It looks like you missed your calling, you should have been writing for
Glass Audio. Your simulations even look like the ones that I used to see
in Glass Audio while I was still a subscriber. Unfortunately, I am not an
up to date designer, equipped with that same simulation package, but your
simulations and schematics provide plenty of fodder to look at and
discuss. Did you ever hear the old saw, "there are lies, damn lies,
statistics, and then there are simulations"? You are not reasoning, you
are creating simulations, and attempting to make them look like they prove
your point.
A couple of issues with your test and simulations, first can you
explain/justify your claim that "In both the common PP & the McIntosh the
load is driven by the output tubes in a series connection. The loop passes
through both output tubes & the four 2K loads in a complete loop"? This
is clearly not true literally, because both the "common PP & the McIntosh"
topologies can be operated in class B, which clearly wouldn't work if the
tubes were literally in series, only pure class A would work in a series
arrangement. I believe Henry Pasternack posted earlier to the effect that
the tubes in the "common PP" circuit effectively operate in parallel. I
would also point out that the only real difference, from the AC point of
view, at frequencies below where the effects of leakage inductance become
important, between the "common PP", and the McIntosh and circlotron
circuits, is where on the load the reference point for the grid drive
signal is placed, the "common PP" referencing the grid drive to the
cathode end of the load, while the McIntosh and circlotron reference the
grid drive to the center point of the load.
Your simulations are still messed up. The McIntosh and circlotron are OK,
but you seem to draw the wrong conclusion about the gain of the McIntosh,
which you claim in your post is "a total of 1.586" for the McIntosh
circuit, and the "gain is 0.794" for the circlotron, although this
conclusion isn't supported by your simulations, which I will return to
shortly. The other problem is an Apples and Oranges thing with the
"common PP" simulation vs. the McIntosh and circlotron circuit
simulations. You have modified your simulation so that the tubes in all
the simulations now see the same load, which is good, although the tubes
in the "common PP" simulation may have slightly different gains because
they are being driven harder, which will affect the effective
transconductance and plate resistance of the tubes, although I will ignore
that, if for no other reason than that you haven't disclosed the nature of
your 6BX7 model. What you have not done is measure the output across a
given fixed resistance in all three simulations, as is required to make
comparable gain measurements. You have measured the gain of the ""common
PP" across an 8k load, and the gain of the circlotron across a 2k load.
It is not clear where you measure the McIntosh gain, because your
simulation shows it being measured across a 2k load, but then in the text
of your post you double this gain value to effectively make it as if it
were measured across an 8k load. The result of all this is an overstating
of the gain of the "common PP" vs. the circlotron by a factor of two. You
present a more confusing picture of the gain of the McIntosh circuit
because your simulation gives a different answer than the text of your
post.
A better way of looking at this gain issue, might be to look at the power
output from each circuit for a given input Voltage, 10 Volts in the case
of your simulations. Here is a table of the results of your simulations
presented this way, power out for 10 Volts input.
Circuit Output Voltage Load Resistance Output Power
common PP 76.42 V 8k Ohms 730.0 mW
circlotron 7.913 V 2k Ohms 31.3 mW
McIntosh per simulation 7.912 V 2k Ohms 31.3 mW
McIntosh per post 15.86 V 8k Ohms ?? 31.4 mW
If we normalize the "common PP" gain to a 2k load, then the gain drops to
3.83 from the gain of 7.66 shown in the formual on your simulation for the
"common PP". Any way you slice it the McIntosh is seen to have the same
gain as the "circlotron".
It is worth noting that the reason that "the tube manufacturers measure
power output of their products" directly across a resistor in the plate
circuit, and not through a transformer, is mainly because real
transformers have significant losses, and they want to get the biggest
number possible, not to mention how much loss should the transformer have
if they used one. This consideration doesn't apply to your simulations,
where you are presumably using loss less transformers.
With respect to the fact that the McIntosh and circlotron circuits are
really the same basic topology morphed in different ways, Patrick Turner
made an interesting post earlier today, where he in essence stated that in
the McIntosh circuit the plate of one tube is locked to the cathode of the
other tube by the bifilar winding of the transformer. Just as the plate
of one tube is locked to the cathode of the other in the circlotron by a
piece of wire. Patrick went on to mention an amplifier design that used
capacitors from the plate of one tube to the cathode of the other to ease
the leakage inductance requirement on the output transformer, I believe I
also mentioned this same capacitor trick in an earlier post of mine.
To understand the equivalence of the McIntosh and circlorton circuits
consider these four steps. I have presented this argument before, and you
have failed to disagree with any of the steps in the argument, you only
disagree with the conclusion, which is not valid unless you point out the
flaw in the argument.
1. Take a McIntosh amplifier, and connect each plate to the opposite
cathode with a large high quality capacitor. The circuit still works.
2. For AC analysis purposes, remove the power supply, and replace it with
a short.
3. Replace the capacitors installed in step #1 with shorts, as these
capacitors are large, and are effectively shorts for AC purposes.
4. Replace the two wires in the bifilar transformer windings with a single
piece of wire since they are now shorted, and are effectively a single
winding.
5. Look at the resulting morphed schematic of the McIntosh circuit, it is
identical to the schematic of the circlotron, with the power supply
shorted for AC analysis purposes.
A perhaps more convincing way of morphing the McIntosh circuit, if you are
uncomfortable with removing the power supply for AC analysis purposes, is
to replace the single McIntosh power supply with two parallel power
supplies, and then translate one through each plate winding so that it is
connected between the plate of each tube and the transformer winding. The
amplifier will still actually work at this point, and the rest of the
above argument can be applied to the resulting amplifier to morph it into
the circlotron topology. These details are left for the student. In
fact, as I pointed out earlier, the only real difference between these two
circuits, the McIntosh and circlotron, and the common PP, is where on the
load the reference for the grid drive is placed.
If it is true that the circlotron is really a fundamentally different
circuit than the McIntosh, you should be able to poke a hole in the above
argument, have at it, but please remember I am ignoring leakage
inductance, as your simulations also do.
John L Stewart
What you describe is how Crowhurst did it in the 1950's. John L Stewart
JOHN L STEWART
John Byrns wrote:
In article <PzL18.21708$jt2.3...@news20.bellglobal.com>, "John L
Stewart" <jhste...@hotmail.com> wrote:Read on & learn the real goods on the McIntosh vs. Circlotron topology!!
[much snippage]It was implied that I didn't understand how these circuits work.
Looks to me like it's the other way around. Sorry about that.
Hi John,
It looks like you missed your calling, you should have been writing for
Glass Audio. Your simulations even look like the ones that I used to see
in Glass Audio while I was still a subscriber. Unfortunately, I am not an
up to date designer, equipped with that same simulation package, but your
simulations and schematics provide plenty of fodder to look at and
discuss. Did you ever hear the old saw, "there are lies, damn lies,
statistics, and then there are simulations"? You are not reasoning, you
are creating simulations, and attempting to make them look like they prove
your point.
computer you are sitting in front of. Without considerable knowledge in each case the tool is useless.
I stand by everything I said in that post. I could as easily came to the same conclusion without
the simulator. It did provide a very convenient way of making the schematics. Not long ago
they were all done on a drawing board.
My experience comes from considerable actual contact with the stuff I talk about.
If there were inconsistencies in the results they would be obvious to me.
I've had enough hands on & design experience to recognize which circuit is which.
Talk is cheap. John, you are grasping at straws.
If you want to prove your point to me & others, lets see some evidence,
in the form of schematics & calculations illustrating your proof.
And by the way, I have had several articles published in Glass Audio & Electronics World (UK).
At this time there are several other articles at AudioXpress, proofed & ready to go when space permits.
One of them happens to be a 50 watt Circlotron.
John L Stewart PEng
John Byrns
> John- What planet are you on? A simulator is a tool, just as a hammer is
> to a carpenter or that
> computer you are sitting in front of. Without considerable knowledge in
> each case the tool is useless.
> I stand by everything I said in that post. I could as easily came to the
> same conclusion without
> the simulator. It did provide a very convenient way of making the
> schematics. Not long ago
> they were all done on a drawing board.
>
> My experience comes from considerable actual contact with the stuff I
> talk about.
> If there were inconsistencies in the results they would be obvious to me.
> I've had enough hands on & design experience to recognize which circuit
> is which.
>
> Talk is cheap. John, you are grasping at straws.
> If you want to prove your point to me & others, lets see some evidence,
> in the form of schematics & calculations illustrating your proof.
OK John, let's take off the gloves and get down to brass tack's, we
disagree about the nature of the McIntosh circuit. You don't seem to want
to discuss your own simulations, which show that your concept of the
McIntosh circuit gain is wrong, and that mine is correct, specifically
that the gain is the same as the circlotron. You are trying to put over
your point, that the gain of the McIntosh circuits is 1.586, by bluff and
bluster, when your very own simulation shows that it to be 0.79, as with
the circlotron. Did you say you were in sales?
This is the critical point, and one you should be able to use to settle
this argument one way or the other. Your simulations for both the
circlotron and the McIntosh circuits show a gain of about 0.79, which
supports my contention that the circuits are equivalent, yet in the text
of your post you claim that the gain of the McIntosh output stage is
1.586, but you provide no evidence to back that claim up. What accounts
for this difference between your simulation results, and your claims? If
the gain really is 1.586 as you claim, then you should be able to connect
one of those Voltage indicator boxes in your simulator between two points
that will show an output Voltage that supports your claim. Yet for some
reason you haven't included this one simple measurement in your
simulation, why? I claim it is because you found that there is no such
pair of points in the McIntosh circuit, and decided to try and win the
battle by bluff and bluster. I am calling your bluff, you have it in your
power to end this argument by presenting the measurement I am asking for.
But I don't expect you will, because your own simulation shows a gain of
only 0.79, not 1.586 as you claim in your post.
> And by the way, I have had several articles published in Glass Audio &
> Electronics World (UK).
> At this time there are several other articles at AudioXpress, proofed &
> ready to go when space permits.
> One of them happens to be a 50 watt Circlotron.
All the articles in the world are no substitute for getting it right. Is
Glass Audio a peer reviewed journal, if not what is the significance of
having had an article published there? Electronics World (UK) is probably
better, but has been known to have published a few questionable articles.
I don't know anything about "AudioXpress", beyond the fact that you have
been hyping it here quite a bit recently.
Patrick Turner
John Byrns wrote:
> In article <3C47DD98...@turneraudio.com.au>, Patrick Turner
> <in...@turneraudio.com.au> wrote:
>
> > McIntosh took another road, and used a lot of local NFB in the output
> > stage, to linearise it before adding global NFB.
> > McIntosh then required some form of stabilisation, as
> > it normally would have been impossible
> > to apply the same total amount of global NFB + local NFB
> > in a conventional Williamson, or Leak, or Quad II amp.
> > His answer was to wind the OPT with bifilar windings, so that
> > although the cathode and plate windings are at potentials which are
> > 400 volts apart, they have the same signal voltages and these are locked
> > together well with extremely low leakage inductance,
> > which then won't cause instability problems, due to HF phase shift,
> > at too low a frequency, like it does in most conventional amps.
>
> Hi Patrick,
>
> I could be wrong, but I always had the impression that the reason McIntosh
> went with the low leakage inductance design was not primarily for
> stability reasons, but was because they run most of their amps deep into
> class AB, if not class B, and the low leakage inductance was needed to
> eliminate notch distortion, so that they could call it high-fidelity.
No, you are not wrong.Indeed the McIntosh design addresses the problem caused
by high leakage inductance with a PP amp trying to operate in low bias current
class AB.
They were indeed trying to get more power easily, cheaply,
and effectively, at low distortion figures, and with low output impedance.
They succeeded at all these things.
But I have never been in any hurry to copy a McIntosh,
as I would maintain you can get all the good sound you want
with a less elaborate OPT and simpler circuit.
The "notch" distortion is , I assume the crossover distortion
which occurs when one PP tube cuts off suddenly, and the other takes
over during the +ve and -ve parts of the cycles.
Both halves of the primary should be coupled tightly
together to enable this to happen as smoothly as possible.
In fact, it is difficult to wind an ordinary OPT with
the half primaries to be close to each other,
but they are locked well together due to the secondaries being
intermingled throughout all the primaries.
The McIntosh method uses primary sections of cathode and plate windings
wound as bifilar windings, so the cathode winding for one tube is beside the
plate winding of the other tube, and vice versa.
I suspect they divide the two lots of dual primaries into a total
of 4 sections, so 5 secondaries can be used.
It is an effective way to wind a tranny.
But, as I said before, GE found a way which I would suspect is
just as effective, and that has at least four equal sections
of primaries, which could be set up as a non bifilar McIntosh amp,
and the same tight coupling could be achieved with shunt caps
across coils of different DC potentials, but same signal phase.
The GE book from 1957 which has schematics of
17 amps from 5 watts to 1,100 watts gives details of all this.
>
>
> You better be careful with that talk about "the cathode and plate
> windings" having "the same signal voltages and these are locked together
> well with extremely low leakage inductance". That's saying the McIntosh
> circuit is equivalent to the circlotron circuit from an AC signal point of
> view, and John Stewart won't like that!
There are similarities, but there are differences too.
> > The McIntosh way of winding trannies is painful for the DIY person.
> > So do it the GE way, and have equal windings seperated by insulation,
> > and where the signal voltages are the same, but DC potentials different,
> > just bypass the ends of the windings with suitable
> > 10 UF high voltage plastic caps, which will shunt any leakages.
>
> Now you have really gone and put your foot in it, actually tying the
> plates and cathodes of the opposite tubes together for AC signals with a
> large capacitor, now it's clear you are trying to say that the McIntosh
> and circlotron are equivalent for AC signal purposes. You better duck
> quick.
I didn't set out to say that McIs and Circs are the same.There are
differences.
The circlo has two supplies riding up and down with
two phases of the signal, McI uses one fixed supply.
Circlo uses local NFB in the output stage, and needs a
high drive voltage to the output tube grids.
But both use tubes in PP, which individually act the same way.
One can build direct coupled SS circlotrons,
and so are they all that different to any other Class AB
low bias current amp?
I prefer good old class A for most of my listening.
But one could make a McIntosh and set it up in class A
There would then be little need for any Global NFB,
as THD and Zout of the output stage would be pretty
good, especially if the tubes were in triode mode.
But the drive amp has to produce the grid to cathode voltage
of about 25 volts for most output tubes, in addition
to the half of the plate to cathode voltage, which is about 120 volts,
so a total of 145 volts rms has to be found, at low THD
to drive the McIntosh.
Hence the need for global NFB to "cure"
the added distortion of the drive amp, as well
as the output stage.
It is a bit like being in the land of taking 2 steps
forward, and one step back, to get one step ahead.
Plain old fashioned AB plate loaded PP amps well done seem
to me to be an effective way of building an amp.
However, because something can be done, someone might do it,
and it all adds to the tapestry of life, and man is an ingenious creature.
Most professional transformer winders vomitted into my lap when I
gave them the details of my output transformers.
Such simple things, they are, too.
But they would have heart failure if I asked them to wind a
McIntosh OPT.
And I have to draw a line someplace for things
I might wind myself, and that line is with bifilar primaries.
While I am on bifilar, remember the capacitance between the
bifilar wires is high, so what would be so bad
about seperate windings, and large caps?
I always steer clear of any secondaries using bifilar too.
And as yet, I have seen no need to use
flat rectangular wire, wound on its edge...
No doubt some folks might grumble that I am
never going to achieve the same sound as a McIntosh.
I don't really care, the methods I use work, and sound well,
and who knows, might be better than a McIntosh.
Patrick Turner.
John Byrns
<in...@turneraudio.com.au> wrote:
> John Byrns wrote:
>
> > Hi Patrick,
> >
> > I could be wrong, but I always had the impression that the reason McIntosh
> > went with the low leakage inductance design was not primarily for
> > stability reasons, but was because they run most of their amps deep into
> > class AB, if not class B, and the low leakage inductance was needed to
> > eliminate notch distortion, so that they could call it high-fidelity.
>
> No, you are not wrong.Indeed the McIntosh design addresses the problem caused
> by high leakage inductance with a PP amp trying to operate in low bias current
>
> class AB.
> They were indeed trying to get more power easily, cheaply,
> and effectively, at low distortion figures, and with low output impedance.
> They succeeded at all these things.
> But I have never been in any hurry to copy a McIntosh,
> as I would maintain you can get all the good sound you want
> with a less elaborate OPT and simpler circuit.
> The "notch" distortion is , I assume the crossover distortion
> which occurs when one PP tube cuts off suddenly, and the other takes
> over during the +ve and -ve parts of the cycles.
Yes, notch distortion "occurs when one PP tube cuts off suddenly", as a
result of the transformer leakage inductance. At the point where the
notch occurs, the other PP tube may have already been conducting for some
time, it isn't necessary for a switch to occur from one tube to the other
as in pure class B, which can cause what I would call crossover
distortion, because it occurs in the "crossover" region around zero
Volts. Notch distortion on the other hand may not appear until relatively
high power levels, and occurs fairly high up on the Voltage wave form. I
would say Notch distortion is a class AB thing, while "crossover"
distortion is a class B thing.
> Both halves of the primary should be coupled tightly
> together to enable this to happen as smoothly as possible.
> In fact, it is difficult to wind an ordinary OPT with
> the half primaries to be close to each other,
> but they are locked well together due to the secondaries being
> intermingled throughout all the primaries.
> The McIntosh method uses primary sections of cathode and plate windings
> wound as bifilar windings, so the cathode winding for one tube is beside the
> plate winding of the other tube, and vice versa.
> I suspect they divide the two lots of dual primaries into a total
> of 4 sections, so 5 secondaries can be used.
> It is an effective way to wind a tranny.
> But, as I said before, GE found a way which I would suspect is
> just as effective, and that has at least four equal sections
> of primaries, which could be set up as a non bifilar McIntosh amp,
> and the same tight coupling could be achieved with shunt caps
> across coils of different DC potentials, but same signal phase.
> The GE book from 1957 which has schematics of
> 17 amps from 5 watts to 1,100 watts gives details of all this.
This same capacitor trick was used in the National Horizon 20 amplifier,
which used a single ended totem pole type output circuit, but the two
tubes were only stacked in the totem pole configuration for AC signals,
they were connected in parallel for the DC supply, eliminating the need
for twice the normal supply Voltage. To accomplish this, two windings
were required on the output transformer, and a capacitor was used to tie
the two windings together for AC. The thing I worry about with this
capacitor scheme, is that it seems possible that the capacitor(s) could
resonate with the leakage inductance in the audio range, resulting in
negative sonic consequences. It wouldn't surprise me if this wasn't one
of the reasons that McIntosh went with the more expensive low leakage
inductance transformer design, rather than using the cheap capacitor
trick.
> I didn't set out to say that McIs and Circs are the same.There are
> differences.
> The circlo has two supplies riding up and down with
> two phases of the signal, McI uses one fixed supply.
That's exactly the point, McIntosh and Electrovoice (circlotron) made a
different set of tradeoffs in implementing basically the same fundamental
AC circuit topology. McIntosh choose a simple power supply, and expensive
output transformer, while Electrovoice choose an inexpensive output
transformer, and more expensive power supply arrangement. The McIntosh
tradeoff was probably a better choice for a stereo amplifier, while the
circlotron approach was impractical in a stereo amplifier, four power
supplies being required vs. McIntoshes single supply. Perhaps that is one
of the factors leading to the disappearance of the circlotron from the
commercial amplifier scene at about the time stereo came out.
> Circlo uses local NFB in the output stage, and needs a
> high drive voltage to the output tube grids.
> But both use tubes in PP, which individually act the same way.
Both circuits use exactly the same amount of local NFB in the output, and
require the exact same drive Voltages with a given set of output tubes.
This is easily seen if you actually do the math, I mean actually write out
and solve the gain equations, not just wave your hands trying to make it
look like a "simulation" that shows the gains to be equal, says they
aren't. If you compare the driver stages of the Electrovoice amplifiers
with the McIntosh 225, you will find that they are virtually identical,
using the same bootstrap technique and everything. I lost the URL a while
back, but there is a web site that goes through the gain calculations for
many output stage topologies, including the circlotron, a search might
locate it.
BFoelsch
instantaneous microgroove recordings.
The RC network was pretty much the standard recommendation in that era, just
a parallel RC circuit in series with the head. This network was generally
built into the recording table. I never saw it installed in the amplifier.
The common "professional grade" recording heads, Presto 1D, Fairchild 541,
RCA MI11850C, etc. were rated with an "impedance", usually at 400 Hz, but as
the X/R ratio was much higher than that of a standard voice coil, the
impedance
varied
much more widely over the frequency range in question. The RC network
expedient was to serve the dual purpose of adjusting the turnover frequency
and also of setting a low limit to the load impedance presented to the
amplifier as the frequency declined. Some recorders, like the RCA 73B,
provided adjustable networks (via jumpers, of course) which also assumed the
function of calibrating the sensitivity of the heads, which commonly varied
by +/- 2
or 3 dB from sample to sample. Equal sensitivity was important so that long
programs, which extended over several discs, would play back uniformly.
Standard practice was to use a single recording amplifier to feed
several turntables via a changeover switch. However, adjusting the
individual RC networks to correct a 3dB level imbalance
resulted in having significantly different loads presented to the amplifier,
depending on which head was selected!
Many feedback designs of that era did not care for the task, I assume due to
the net capacitive reactance of the load over part of the audio spectrum.
These RC networks made no attempt at all to maintain constant impedance in
either
direction.
The real trouble was at the other end of the spectrum, where the load
impedance was high, and the amplifier was lightly loaded. The recording
heads were designed to give a constant recorded velocity for a constant
voltage. Add to
this the standard NAB pre-emphasis of 16dB at 10 kc, plus a few more dB for
diameter compensation and correction of miscellaneous HF losses, and you get
to a situation where you need lots of output voltage into a very high
impedance load at a high frequency. Never would you need 50W of power out of
the
amplifier, but you would need the voltage a 50W amplifier could provide into
basically an open circuit! The average feedback amplifier did not like this
either.
Some of the more successful designs look strange by today's standards. For
example, the Presto 90B recording console used PP 6L6s in pentode mode, but
took no feedback around the output transformer. NFB was applied from each
6L6 plate to the cathode
of its respective driver, but that was it. The Fairchild 620, on the other
hand,
used global NFB from the 8 ohm output tap to the cathode of the first stage,
but the output tubes were
2A3/6B4/300B depending on the serial number.
I tried to get the American Williamson circuit to work (Stancor A-8054 OPT),
but I had problems with LF instability, unless I swamped the output with
resistors. I tried some home-brew amplifiers and a few other commercial
designs, with no outstanding
results. The Mac 50W-2, however, was the first general-purpose amplifier I
found that both
approximated a perfect voltage source and had no particular stability
problems with the type of loads discussed above, and
it also had a considerably higher power rating than the specialty recording
amplifiers
of the time.
I have no doubt that a "non-Mac" amplifier could have been built to suit
these requirements, but just about that time the concept of feedback around
the recording head was developed by Emory Cook, and from that point forward
the design of the head and amplifier
was based on a system approach.
Your point about the crude design of these systems is well taken. I will
comment that these recording systems were largely based on empirical data,
and engineering analysis was sketchy at best. The RC network to which I
refer is the topic
of much discussion in the literature, but the idea of providing a
constant-impedance
network to perform this task is rarely if ever discussed. Indeed, the
hardest part of getting that stuff
to work was getting any degree of repeatability at all. The age of the disc,
the temperature of the disc, and the room, and the recording head were all
significant factors.
The program level was an issue, because higher levels would heat the head
and change its
characteristics. The exact depth of the groove was critical in that it
changed the loading on
the moving stylus point.Then, of course, the styli varied wildly from sample
to
sample, both as to noise level and frequency response, and to the required
cutting angle. To be
honest, the best you could hope for was a recording that sounded OK. If you
tried to
measure anything you would be shocked. Actually, if you listen to some of
the earliest LPs, which were recorded on this caliber of equipment, you will
be shocked!
"John Byrns" <jby...@enteract.com> wrote in message
news:jbyrns-1301...@216-80-74-247.d.enteract.com...
> In article <ozI%7.438971$C8.31...@bin4.nnrp.aus1.giganews.com>,
> "BFoelsch" <BFoe...@no.spam.att.net> wrote:
>
> > I became acquainted with McIntosh when I was doing lacquer disc
recording.
>
> Would it be getting too personal to ask what time period you are talking
about?
>
> > The Mac amps sounded pretty much as good as the best then available, but
I
> > liked the 50W-2 because it was unconditionally stable under any
reasonable
> > conditions (and many unreasonable conditions.) The disc cutting heads
were
> > difficult loads, and standard practice was to make them more difficult
by
> > connecting them to the amplifier through an RC network that gave you the
LF
> > crossover you wanted. No problem for the Mac, but a lot of generic NFB
> > amplifiers had troubles.
>
> What was the nature of these RC networks? They don't sound like they
> would be qualitatively much different than the 300 Hz or 500 Hz frequency
> dividing network for a high frequency horn speaker. I wonder why one was
> a problem for a lot of amps, when the other apparently wasn't? Any idea
> why a low level network wasn't used to achieve the desired recording
> crossover? Seems like that would even save a little amplifier power, and
> give more headroom? But I suppose it was something gross, like the desire
> to connect a monitor speaker directly across the output of the cutting
> amp?
JOHN L STEWART
prepared drawings
of the development of both the Circlotron & the McIntosh from simple
Single Ended Amps.
See the binaries NG's we usually inhabit.
These are looked at from an electrical point of view rather then
historical.
The circuits as shown are known in textbooks by many as the "AC
Equivalent Circuits".
There is no DC to get in the way with this kind of analysis. It is used
by Colleges &
Universities to teach circuit theory, as many of you are aware.
For the McIntosh, I started out with a SE Amp, then to the PP version &
then showed
how connecting one half of the load in the cathodes resulted in the
McIntosh.
For the Circlotron, I began with a cathode follower since it more
readily fits.
The Circlotron appears as a pair of back to back connected cathode
followers
driving a common load in parallel.
AC current flow in each case is shown by an arrow, just as you would see
it in a text book. If anyone wants to take the time to do the math they
will
find the resulting numbers for gain will be very close the those shown
on my
earlier simulations. Don't expect a three figure correlation.
Tube specs are approximate at best.
Cheers John L Stewart
Patrick Turner
John Byrns wrote:
I thought radio engineers had bothers with buzzing soundsin time with the programme,
when notch distortion was bad
in the tube amp class B modulators for radio transmitters.
I don't remember hearing such phenomena.
But at one old station, the engineer couldn't go home at night
until he'd made sure the distortion in the monitor receiver was less
than 2%.
Class AB, where there is a large part, say 50% of power in
class A, and the balance in class B does create a gain change in the amp.
With triodes, and UL, this is a gradual thing, but with pentodes
it is a sudden change, when one tube out of the two cuts off.
The transfer curve becomes like a kinked line, and
the THD content is appalling.
With triodes though, the load each tube in the AB circuit
changes gradually from half the nominal a-a RL,
to a quarter of a-a RL, when one tube is cut right off.
Therefore the load lines are curved, not straight, for
each individual tube.
Triodes have increasing Gm at higher plate currents, which is part of their
nature when making 2H distortion in a SET amp.
So when they are used PP, the change down in gain from A to AB
is lessened by their good nature.
Combined with a gradual cut off behaviour, class AB
triode PP amps are generally pretty darn good.
Well, it is easy to work out what resonances you may end up with.The shunt C and LL
of most OPTs will resonate at some Fo
and the higher this Fo, the better.
Typical shunt C might be 0.001uF, and LL might be
20mH seen at the primary.
Fo = 5,035 divided by 2 x 3.14 x (square root of [ C x L ] ).
If the Fo is above 50 kHz, there is not much to fear from
resonances were it not that they seriously upset stability
when NFB is used, and so critical damping and
a reduction of phase shift and gain at HF is required.
>
>
> > I didn't set out to say that McIs and Circs are the same.There are
> > differences.
> > The circlo has two supplies riding up and down with
> > two phases of the signal, McI uses one fixed supply.
>
> That's exactly the point, McIntosh and Electrovoice (circlotron) made a
> different set of tradeoffs in implementing basically the same fundamental
> AC circuit topology. McIntosh choose a simple power supply, and expensive
> output transformer, while Electrovoice choose an inexpensive output
> transformer, and more expensive power supply arrangement.
Agreed.
> The McIntosh
> tradeoff was probably a better choice for a stereo amplifier, while the
> circlotron approach was impractical in a stereo amplifier, four power
> supplies being required vs. McIntoshes single supply. Perhaps that is one
> of the factors leading to the disappearance of the circlotron from the
> commercial amplifier scene at about the time stereo came out.
Not even mass production could get over these obstacles.
>
>
> > Circlo uses local NFB in the output stage, and needs a
> > high drive voltage to the output tube grids.
> > But both use tubes in PP, which individually act the same way.
>
> Both circuits use exactly the same amount of local NFB in the output, and
> require the exact same drive Voltages with a given set of output tubes.
> This is easily seen if you actually do the math, I mean actually write out
> and solve the gain equations, not just wave your hands trying to make it
> look like a "simulation" that shows the gains to be equal, says they
> aren't. If you compare the driver stages of the Electrovoice amplifiers
> with the McIntosh 225, you will find that they are virtually identical,
> using the same bootstrap technique and everything. I lost the URL a while
> back, but there is a web site that goes through the gain calculations for
> many output stage topologies, including the circlotron, a search might
> locate it.
>
> Regards,
>
> John Byrns
Yes, well, ask the Output Tubes if they notice any difference goingon outside
themselves, and I think they would tell you....
"We seem to be in a class aB amp, with low bias
currents, and we make a few mistakes, with distortion,
and we are forced to be like semi conductors,
only half turned on in our job, but this owner of us all
has the NFB whip in hand and is applying
to our backsides to keep us on the straight and narrow,
and we do try to stay cool and out of bother by producing
lots of power, at distortion which
he says is tolerable.
We here upon the chassis are not alone,
and we have many little friends providing
our high drive voltages, and the owner whips them too with
NFB, and these little friends whip us with correction voltages.
We would prefer to be in a nice old Williamson;
we would give anything to be outa here."
Patrick Turner http://www.turneraudio.com.au
John Byrns
<in...@turneraudio.com.au> wrote:
Hi Patrick,
I think you missed the point here, I wasn't talking about resonance's of
the leakage inductance with the shunt C in the transformer, but resonance
of the transformer leakage inductance with the much larger capacitor you
were talking about using to parallel two windings, as per "GE".
As an aside, I ran out to the bookseller yesterday to buy the GE book you
mentioned, and found they only used the trick with driver transformers, I
didn't see it used in the output transformer of a single design in the
book, probably because they weren't into the Macintosh style topology.
JOHN L STEWART
Patrick Turner
John Byrns wrote:
Ah, well, I assumed it would not be a problem.Say you had an OPT with a primary
divided
up into four equal sections placed between 5 secondaries.
It is possible to use such an arrangement to provide the necessary
4 windings for a McIntosh amp with caps connecting ends
of windings with equal audio signals.
Now the leakage in such an arrangement between the coils in question
will not be so very high, less than the total P to S leakage,
a few mH, and if the cap is 10 uF,
then the resonance would be in the audio band theoretically,
but I suspect it is of little consequence.
Take C = 0.001uF, and LL = 5 mH.
then Fo will be 71 kHz.
By the way, the formula I hastily quoted for Fo was wrong,
it should be 5,035 divided by the square root of LL x C, where
LL is in mH, and C is in uF.
I use this formula for speaker crossover values often.
The 2 x 3.14 is already in the figure of 5,035.
But 10 uF and 1 mH will have resonance at 1,592 Hz, which
is a very low Fo, but the impedances of the resonant structures
at such low Fs will be small compared to the load which I would
say would utterly damp the Q of the resonance.
We use say 50 uF and 1 mH to drive a midrange speaker, and
there are no worries about resonance between the two
reactive elements.
10 uf has an impedance of 0.8 ohms at 20 kHz, and gets
to 0.08 at 200 kHz, so the windings become very locked together as F rises.
> As an aside, I ran out to the bookseller yesterday to buy the GE book you
> mentioned, and found they only used the trick with driver transformers, I
> didn't see it used in the output transformer of a single design in the
> book, probably because they weren't into the Macintosh style topology.
I know you are correct, I have the book,but the principle of locked together
windings
is the same. I used it in a superlinear amp I had with 25% of the total
primary voltage fed back to the cathodes, and also to the screens.
So part of the tranny was locked together.
Where the cathode signal voltage was equal to the cathode voltage
of the other tube, I bypassed it with a cap.
The amp was marginally more stable.
In the McIntosh, the tubes are working as real pentodes.
The signals at the cathodes are the same as the the screens.
Now there are a few possiblities with how you arrange all
these various windings, which are most effectively done
when they are all on the same core for good LF coupling,
rather than on seperate similar transformers,
or else the size of C coupling caps would seem to me to have to be large,
maybe 100 uF, so electros come to mind.
I would have thought that if you took the screens to the B+,
and had them at a fixed voltage, like Quad II does,
then you'd have a defacto
UL amp with 50% of the plate voltage fed to the screens.
The relative voltages would be the same, and the power
output would be the same as a 50% UL amp, and distortion
would be reduced.
I honestly dunno how a Leak "50 watt" mono amp
with 2 x KT88 or 6550 would compare.
I got 65 watts at clip with one I fixed a few years back.
The cathode feedback is a more effective way of applying screen feedback,
IMHO, so the direct comparison between a 50% UL amp
and a mac with its screens taken to the B+ isn't a fair comparison.
But I see no reason to believe it wouldn't be a bad idea to
do this. The down side is that UL amps need more drive than pentodes,
so a marginally even higher drive voltage is required.
And why bootstrap from the output stage?
I have never favoured this, but I guess it works.
I would have thought some way of producing drive without
bootstrapping would always be better.
But 160 vrms needed isn't so easy, unless one uses
a step up tranny, and power tubes such as EL84 to drive
the thing. And then stability with feedback
goes out the window, with all these reactive elements involved.
At the end of the day, all we need is a good damping factor,
ie, speaker Z / amp Ro = better than 10.
If we wanted extra low Ro, then there is a case for the McIntosh,
but I can't see the point of going to all that trouble.
What I have found is that 12.5% of cathode feedback
is all you can have before the drive amp voltage begins to become high,
and the drive amp distortion increase is as bad as the output stage distortion,
and when you have got the to the McIntosh position,
with 50% of cathode feedback, then the although you have
low Ro even with no global NFB connected, the THD
will be several %, and include some crossover components,
if the output stage is in class AB with low bias current.
So, My superlinear amp with 4 x 6550 per channel
still manages to pump out 90 watts with very low THD,
and as there is a bagful of class A working, the sound is utterly blameless.
I had a large 200 watt musician's bass amp come in for a minor repair today,
and it has 4 x KT88. Now this simple amp has a splendid appeal to its owner,
but 200 watts from only 4 output tubes is a bit over the top.
And so is the complex elegance of McIntosh.
Somewhere between the two extremes is a middle path and better road,
and it involves more class A, and tube numbers, and a retention
of the simplicity in the drive amp.
This is the path I take.
For the brave souls who try to do their own McIntoshes,
I do wish them well.
I invite anyone to cap couple the windings, instead of bifilar winding
everything, and tell us how you get on.
Patrick Turner.
Patrick Turner
JOHN L STEWART wrote:
> Better look at my post of Crowhurst's
> circuit. John L Stewart
But exactly which binaries group?Or do you have the
schematics posted somewhere.
Patrick Turner.
JOHN L STEWART
alt.binaries.schematics.electronic
alt.binaries.radio.pictures
alt.binaries.pictures.radio.
There may be others, but those are where I posted Crowhursts circuit
& a lot of other stuff relating to the McIntosh & Circlotron.
I see posts from others in those NG's from this NG as well,
but I heard that some can't receive them through their ISP.
If you can't get the binaries let me know.
I will send the stuff straight to your mailbox.
Regards John L Stewart
John Byrns
>
> John Byrns wrote:
>
> >Hi Patrick,
> >
> >I think you missed the point here, I wasn't talking about resonance's of
> >the leakage inductance with the shunt C in the transformer, but resonance
> >of the transformer leakage inductance with the much larger capacitor you
> >were talking about using to parallel two windings, as per "GE".
> >
> >As an aside, I ran out to the bookseller yesterday to buy the GE book you
> >mentioned, and found they only used the trick with driver transformers, I
> >didn't see it used in the output transformer of a single design in the
> >book, probably because they weren't into the Macintosh style topology.
> >
> Better look at my post of Crowhurst's circuit.
Hi John,
OK, I've got it, is there anything in particular I should be looking for
in Crowhurst's circuit? You also mentioned that Crowhurst built a total
of three similar amplifiers in the 1957 thru 1960 period. Could you
provide references to where the articles on the other two were published?
Patrick Turner
JOHN L STEWART wrote:
> Hi Patrick- The binaries people on this NG seem
> to favour are the following-
> alt.binaries.schematics.electronic
I just checked this last binaries to which I
subscribe, andthere was no sign of anything from
you, or anything titled about tube circuits.
so I guess the ISP I have is slack.
> alt.binaries.radio.pictures
> alt.binaries.pictures.radio.
I have just subscribed to these two, and after group
download is completed,I will be able to see what is
there.
> There may be others, but those are where I posted
> Crowhursts circuit
> & a lot of other stuff relating to the McIntosh &
> Circlotron.
> I see posts from others in those NG's from this NG
> as well,
> but I heard that some can't receive them through
> their ISP.
> If you can't get the binaries let me know.
> I will send the stuff straight to your mailbox.
I have not a great deal by Mr Crowhurstin my info
slections, and I would like to read more.
Thanks John,
Patrick Turner.
Patrick Turner
JOHN L STEWART wrote:
> Hi Patrick- The binaries people on this NG seem
> to favour are the following-
> alt.binaries.schematics.electronic
> alt.binaries.radio.pictures
> alt.binaries.pictures.radio.
>
> There may be others, but those are where I posted
> Crowhursts circuit
> & a lot of other stuff relating to the McIntosh &
> Circlotron.
> I see posts from others in those NG's from this NG
> as well,
> but I heard that some can't receive them through
> their ISP.
> If you can't get the binaries let me know.
> I will send the stuff straight to your mailbox.
>
> Regards John L Stewart
Seems like I can't get all 3 alt binaries
contentscontaining pictures.
I guess I should phone my server, and see why.
There are plenty other alt binaries available, but
none with pics of radios or electronic schematics.
Patrick Turner. http://www.turneraudio.com.au
John Byrns
<in...@turneraudio.com.au> wrote:
> John Byrns wrote:
>
> > Hi Patrick,
> >
> > I think you missed the point here, I wasn't talking about resonance's of
> > the leakage inductance with the shunt C in the transformer, but resonance
> > of the transformer leakage inductance with the much larger capacitor you
> > were talking about using to parallel two windings, as per "GE".
>
> Ah, well, I assumed it would not be a problem.Say you had an OPT with a
primary
> divided
> up into four equal sections placed between 5 secondaries.
> It is possible to use such an arrangement to provide the necessary
> 4 windings for a McIntosh amp with caps connecting ends
> of windings with equal audio signals.
> Now the leakage in such an arrangement between the coils in question
> will not be so very high, less than the total P to S leakage,
> a few mH, and if the cap is 10 uF,
> then the resonance would be in the audio band theoretically,
> but I suspect it is of little consequence.
Hi Patrick,
I agree that the resonance is probably of little consequence in a modern
audiophile style class A amplifier, in which case there is probably little
point in including the capacitor in the first place. My worry is that the
resonance could cause nasty sonic problems in a class AB or class B output
stage of the type for which these topologies were originally designed.
> Take C = 0.001uF, and LL = 5 mH.
> then Fo will be 71 kHz.
>
> By the way, the formula I hastily quoted for Fo was wrong,
> it should be 5,035 divided by the square root of LL x C, where
> LL is in mH, and C is in uF.
> I use this formula for speaker crossover values often.
> The 2 x 3.14 is already in the figure of 5,035.
>
> But 10 uF and 1 mH will have resonance at 1,592 Hz, which
> is a very low Fo, but the impedances of the resonant structures
> at such low Fs will be small compared to the load which I would
> say would utterly damp the Q of the resonance.
> We use say 50 uF and 1 mH to drive a midrange speaker, and
> there are no worries about resonance between the two
> reactive elements.
> 10 uf has an impedance of 0.8 ohms at 20 kHz, and gets
> to 0.08 at 200 kHz, so the windings become very locked together as F rises.
>
>
> I would have thought that if you took the screens to the B+,
> and had them at a fixed voltage, like Quad II does,
> then you'd have a defacto
> UL amp with 50% of the plate voltage fed to the screens.
> The relative voltages would be the same, and the power
> output would be the same as a 50% UL amp, and distortion
> would be reduced.
Yes, that is also my take on the QUAD II circuit, although of course QUAD
didn't use a 50-50 split of the winding, but the same reasoning applies.
Some of the QUAD faithful seem to bristle at the idea that the QUAD II is
a closet UL amp, probably because the QUAD marketing machine tried to
portray the UL circuit in a negative light.
Patrick Turner
John Byrns wrote:
I am not sure, but if C values were high, then maybe no problem.
The Quad II operating conditions are just like if you hadabout 10% of screen
feedabck, and so the tubes put out
nearly as much power as pure pentode operation.
But the use of CFB has a far greater effect on the effective plate
impedance than 10% of feedback to the screens if the tubes were
plain UL connected.
Quad's idea was "pentode power, triode sound!"
Remember that in 1953 people had shooting wars
over whether triodes or pentodes were best.
Then UL, and CFB became items, and even more bodies
were found, and it didn't stop until some idiot
improved transistors for use in amps.
The Quad II Ra with CFB is like a triode amp,
or like tetrodes connected as triodes.
The idea is basically sound, but it is different to plain UL.
I like a combination of both techniques,
but I don't use it on all my amps, and I prefer plain UL
for most amps, and I rely on wide bandwidth low loss
OPTs to do the work of UL without all the claimed
problems that a poor UL amp can have, the main one being
instability.The original Hafler and Keroes articles on CFB, and UL, et all,were
published in Wireless World in about 1953.
See if you can get to read that, if you can get to a collection
of WW mags in a library.
Generally most good universities will have copies, often poked
away in a basement somewhere.
I looked through all WW mags from 1917 onwards,
and they are a really darn good read, both in the language used,
presentation, effort put in, and information given, about radio,
and audio, and a thousand other old ways of doing things,
mostly forgotten now.
Patrick Turner.