SET Asylum

Single Ended Triodes (SETs), the ultimate tube lovers dream.

Return to SET Asylum

Message Sort: Post Order or Asylum Reverse Threaded

A tale of two amps

98.144.114.174
Posted on February 14, 2021 at 12:31:15
PaulF70
Audiophile

Posts: 1403
Location: Midwest
Joined: June 30, 2006



I've got a couple really great amplifiers here now - a Croft Series 7R and a pair of Antique Sound Labs 845 SETs.

They are both fed directly from a PS DirectStream DAC and driving Art Dudley's (RIP) Quad 57s (fully rebuilt by the master at Electrostatic Solutions).

(I sometimes bi-amp with a Marchand XM66 and two large REL subs, but did not for these comparisons).

For those not familiar with the Croft, it is an evolution of his groundbreaking OTL all-tube designs. A zero-feedback triode input stage drives a single pair of MOSFETs per channel.

I cannot pick a "winner." Many of the differences are what you'd expect from raw specs, especially the far wider bandwidth of the Croft. The Croft is faster and has a somewhat lower noise floor. Tonally, they are most similar using the ASLs' four-ohm transformer taps. Even then, the Croft has "more" treble, or sounds it.

In almost all ways, the Croft seems superior. And, believe me, it's a fabulous amp.

But the SET has two qualities that continue to draw me in:

- Midrange palpability - and the ability to differentiate textures - seems slightly superior

- The amps' treble seems a bit more pure.

Now, naysayers will say that the SETs have loads more low-order THD which makes mids sound rich, and that I simply prefer a rolled-off treble. But I see issues with such analysis.

First, the midrange is not really "thicker," but just a bit more - alive.

Regarding the treble, the Quad ESL has about the most pure and natural treble one will ever hear. That membrane has no bad habits. So it doesn't seem to wash that more extended treble would sound unpleasant or poor.

I am tempted to believe that there will always be some artifact of push-pulling a signal and that treble is where it will be most audible.

Two things are for sure: The human brain's auditory system is extremely sophisticated, and this hobby has nearly inexhaustible potential for interest.

 

Hide full thread outline!
    ...
RE: A tale of two amps, posted on February 15, 2021 at 01:13:12
morricab
Distributor or Rep

Posts: 9177
Location: switzerland
Joined: April 1, 2005
 Not a too surprising outcome. Try the ASLs on the 8 ohm tap. I have found with nearly every SET I have owned that the 8 ohm tap sounds better, regardless of the speaker load. More open...might bridge the gap you perceived there.

That palpability is hard to get away from because if you sit reasonably close to live instruments in real space you hear just that.

It would be interesting to compare the ASL (an amp I have considered many times but for some reason never pulled the trigger) to the real OTL Croft and not a hybrid. The simple truth is that push/pull MOSFETs will never sound like a single big or small tridoe output tube. Their transfer functions are simply different and the subsequent distortion produced not the same pattern.

I guess the Croft output stage is Class AB? It would also be interesting if it were true Class A as Mosfets make a lot less distortion in Class A.

I once had an even more interesting amp that could be compared directly to a SET. It was a NAT Symbiosis SE, which was a single ended hybrid with just ONE giant MOSFET on the output stage. It had huge tower heat sinks to dissipate the massive 800 watts it consumed just sitting there. It also had a regulated output, which meant two more giant MOSFETs on two more huge heat sinks (4 in all). Even then, it STILL didn't sound quite like a SET or an OTL (of course it had no output transformers) but it did deliver an almost psychadelic tonality and soundstaging when fully warmed up (took over 2 hours to get there though). I sold it because I still preferred SET and the 2+ hours to get really good was too annoying.

Anyway, look forward to hearing more about this comparison in time.

 

Took the SET Koolaid a couple of years ago., posted on February 18, 2021 at 17:22:12
The Killer Piglet
Audiophile

Posts: 4780
Location: FL
Joined: January 2, 2002
 Korneff 45 and Omega SAM's, 1.75 watts of audio catnip.

Still totally enthralled.

KP

 

RE: A tale of two amps, posted on February 18, 2021 at 20:35:11
PaulF70
Audiophile

Posts: 1403
Location: Midwest
Joined: June 30, 2006
 I am actually not sure if the Croft is biased into Class A. Little info on their site. (Like so many boutique brands in our hobby, it's basically one guy in England from what I understand.)

I had been preferring the Croft. Tonight, I hooked up the xover - the Marchard XM66 set at 60 Hz feeds the subs and the amp(s). Tonight, I definitively preferred the SETs. The low-level nuance feature (not a bug!) somehow took on greater prominence. The higher noise floor receded - is it 60 hz hum neutered by rolling off the bass?

Tonight both Art Petter and Rush were very good indeed.

16 ohm taps.

 

The math, posted on February 19, 2021 at 13:01:01
Ralph
Manufacturer

Posts: 4769
Location: Minnesota
Joined: April 24, 2002
 I am tempted to believe that there will always be some artifact of push-pulling a signal and that treble is where it will be most audible.

Its a bit trickier than that! If things go wrong, yes, that happens.

If the circuit is fully differential rather than just push-pull at the output, then it will express what is known as a 'cubic non-linearity', as opposed to the quadratic non-linearity of single-ended circuits.

An amp with the former will generate the 3rd harmonic as the dominant distortion product (which the ear treats exactly as it does the 2nd) although at a level slightly lower than seen in an SET. As the order of the harmonics are increased, their amplitudes drop off at a higher rate than in the latter (SETs). The result is a more neutral sound (more musical) since the ear assigns a tonality to all forms of distortion.

Now if the circuit shares single-ended and push-pull circuitry single-ended input, push-pull output), both non-linearities are present. The algebraic sums result in a fairly prominent 5th harmonic and the distortion drops off fairly slowly (which is why amps of this sort often employ loop negative feedback), compared to an amp with only a cubic non-linearity. This seems to me to be the main objection SET purists have to push-pull operation, but is unfair since not all push-pull amps are made this way!

IOW the topography of the amplifier makes a difference!

Masking is also playing a role- in an SET, the 2nd and 3rd are masking the presence of the higher orders, so the amp sounds smooth even though it has more higher ordered harmonic distortion than most solid state amps! Solid state amps often sound bright and harsh due to the fact that the 2nd and 3rd are of insufficient magnitude to mask those higher orders, so they stand out, because the ear is keenly sensitive to higher ordered harmonics, as it uses them to sense sound pressure.

A bit of a discussion about feedback is in order here too- a hybrid amp with MOSFET outputs is clearly also using feedback. The thing about feedback is while it does suppress innate distortions in the circuit, it also adds its own through the process of bifurcation, and can add IMD at the feedback node. This causes harshness and brightness. This can be avoided if enough feedback is used, but it has to be a lot: 35-40dB at a minimum, and quite simply no hybrid amplifier will ever have enough gain bandwidth product to allow for that kind of feedback, and even if it did it would likely just go into oscillation sooner or later.

There are ways around this problem, but the point here is that you are making a bit of a leap by accusing all push-pull amps of the issue you describe.

 

RE: The math, posted on February 23, 2021 at 16:38:55
PaulF70
Audiophile

Posts: 1403
Location: Midwest
Joined: June 30, 2006
 "If the circuit is fully differential rather than just push-pull at the output, then it will express what is known as a 'cubic non-linearity', as opposed to the quadratic non-linearity of single-ended circuits."

Please explain that. :) What does it mean?

(Nelson Pass once told me that all push-pull amps suffer from some form of crossover distortion. Do you disagree?)

"...a hybrid amp with MOSFET outputs is clearly also using feedback." Please do explain that as well. I ask out of ignorance. I thought it was possible to use a valve for voltage gain and a transistor following with current gain only, with no feedback. Is that wrong?

I ask seriously. Most of us don't design circuits and don't really know this shit in the details. I certainly don't.

 

RE: The math, posted on February 24, 2021 at 08:59:29
Ralph
Manufacturer

Posts: 4769
Location: Minnesota
Joined: April 24, 2002
 You may have also read Nelson commenting on the fact that a fully differential amplifier tends to have lower overall distortion- that its harmonics are all at a lower amplitude than those harmonics are seen in a single-ended circuit. IOW, as you increase the order of the harmonic, they drop off at a faster rate. That is really what you need to know about a cubic non-linearity.

How it works is that even ordered harmonics are cancelled at each stage throughout the amplifier, unlike a conventional push-pull amp where the even orders are only cancelled at the output. This means that distortions are not compounded from stage to stage throughout the amp. Because of algebraic sums, this means that the pesky odd orders are at a lower amplitude as well.

(Nelson Pass once told me that all push-pull amps suffer from some form of crossover distortion. Do you disagree?)

If we are talking measurable, no. There might be a slight amount that can't be seen on an oscilloscope (an instrument that displays waveforms) and if that is the case, the result would tend to skew the distortion spectra towards the lower orders, IOW making it very hard to hear!

"...a hybrid amp with MOSFET outputs is clearly also using feedback." Please do explain that as well. I ask out of ignorance. I thought it was possible to use a valve for voltage gain and a transistor following with current gain only, with no feedback. Is that wrong?

More or less, yes. You would really have to jump through some hoops to make a circuit like that be linear. MOSFETs that are commonly available tend to not be very linear at all- most of them are made for switching these days. In addition, they tend to have rather high amounts of gate (input) capacitance, which can be difficult for a tube to drive.

Decades ago there was a kind of power FET known as a Static Induction Transistor or Vertical Field Effect Transistor (VFET). Sony made these famous for a while in the 1970s. They have linearity similar to triode power tubes and are even capable of soft clipping. If you had some of those you could make a zero feedback amplifier as you describe and people have done just that (take a look at the link below!). But those devices are rare these days so no-one makes products based on them (except from time to time Nelson Pass has made limited production kits from them, although from what I hear, there won't be any more unless he encounters another stash). This means that you have to use the Horizontal Field Effect Transistors (also known as MOSFETs), which don't have that sort of linearity. So they'll need feedback.

 

RE: The math, posted on March 15, 2021 at 18:12:39
PaulF70
Audiophile

Posts: 1403
Location: Midwest
Joined: June 30, 2006
 "You may have also read Nelson commenting on the fact that a fully differential amplifier tends to have lower overall distortion- that its harmonics are all at a lower amplitude than those harmonics are seen in a single-ended circuit. IOW, as you increase the order of the harmonic, they drop off at a faster rate. That is really what you need to know about a cubic non-linearity."

I don't know, but I do know that he told me personally that all push-pull circuits have *some* crossover distortion (which is evidently why he developed his single-ended bias circuit).


"Decades ago there was a kind of power FET known as a Static Induction Transistor or Vertical Field Effect Transistor (VFET). Sony made these famous for a while in the 1970s. They have linearity similar to triode power tubes and are even capable of soft clipping. If you had some of those you could make a zero feedback amplifier as you describe..."

Yes, I know. Everyone knows about SITs now thanks to the First Watt amps that use them. But this has very little to do with the Croft amp since it uses tubes for voltage gain.


"More or less, yes. You would really have to jump through some hoops to make a circuit like that be linear. MOSFETs that are commonly available tend to not be very linear at all- most of them are made for switching these days. In addition, they tend to have rather high amounts of gate (input) capacitance, which can be difficult for a tube to drive."

Is anyone able to jump through these hoops?

As I noted, the Croft circuit uses valves for voltage gain and MOSFETs for output (current). Do you *know* that such a circuit must use feedback? Local or global? How much?

 

RE: The math, posted on March 15, 2021 at 18:35:07
rivervalley817
 'Vertical Field Effect Transistor (VFET). Sony made these famous for a while in the 1970s'

Ralph,

have you seen anything on Vertical Organic Field‐Effect Transistors?

apparently they're realizing some potential lately

with regards,

 

RE: The math, posted on March 16, 2021 at 08:36:55
Ralph
Manufacturer

Posts: 4769
Location: Minnesota
Joined: April 24, 2002
 "More or less, yes. You would really have to jump through some hoops to make a circuit like that be linear. MOSFETs that are commonly available tend to not be very linear at all- most of them are made for switching these days. In addition, they tend to have rather high amounts of gate (input) capacitance, which can be difficult for a tube to drive."

Is anyone able to jump through these hoops?

As I noted, the Croft circuit uses valves for voltage gain and MOSFETs for output (current). Do you *know* that such a circuit must use feedback? Local or global? How much?

My paragraph above yours explains how I know it uses feedback, and your amp apparently is one that jumps through those hoops :) It would be loop feedback as degenerative feedback would be insufficient. Hard to say how much but because it uses tubes likely not more than 20dB as the phase margin would prevent any more due to oscillation.

Regarding Nelson's comments, if so the crossover distortion is unmeasurable in many amps and probably manifests as a lower ordered harmonic.

 

RE: I've not- looks interesting., posted on March 16, 2021 at 08:41:47
Ralph
Manufacturer

Posts: 4769
Location: Minnesota
Joined: April 24, 2002
 No comments about linearity that I've seen so far.

 

RE: A tale of two amps, posted on May 8, 2021 at 08:12:23
PaulF70
Audiophile

Posts: 1403
Location: Midwest
Joined: June 30, 2006
 I came across something I found very interesting on another forum recently. Here it is:

"Some time ago, I had my Cary CAD-211AE amplifier (110W push-pull) on my engineer's bench to do some repairs. We had a very interesting conversation. It went something like this:

Him: "Look at this. The voltage rail on the push part of the circuit measures at 904V. The voltage rail on the pull part of the circuit measures 890V"

Me: "You are saying that the push voltage and the pull voltage are different? Will that affect the sound?"

Him: "Yes it will"

Me: "Why are the push and pull voltages different?"

Him: "Because of valves. Most likely the valves on the push side are of a different specification to the pull side""

He goes on to note that this is unavoidable. Even valves that start life equivalent will drift.

Things I'm wondering:

- How could ANY push-pull (or "differential") amp avoid such a problem?

- This has to have some audible effect on the signal - especially low-level signals. (Hmm, that's just where SE amps really shine, isn't it?)

Thoughts?

 

RE: A tale of two amps, posted on May 12, 2021 at 18:24:39
PaulF70
Audiophile

Posts: 1403
Location: Midwest
Joined: June 30, 2006
 Ralph, any comment on this?

 

RE: A tale of two amps, posted on May 22, 2021 at 12:37:21
tube wrangler
Manufacturer

Posts: 2484
Location: USA
Joined: January 29, 2007
 A push/pull, and/or a balanced scheme is
a signal processor.

In studio recording, and in the field, you
NEED this processor-- mike cables get long
(and noisy), hum is picked up along the way
by Single-Ended circuits. You process it out.

To record, we need balanced circuits and
push/pull architecture precisely because
we need the cancellation of Common-Mode
signals that are unwanted-- they're picked up
by wiring and circuits.

In PLAYBACK, we don't need any further signal
processing (normally).

Compensation for the effects of balanced
circuits and push/pull systems was taken care
of by the recording studio's minor adjustments.
The studios deliver what they can with what
they use.

If we get rid of balanced or push/pull
processing for playback, we get the whole recorded
sonic picture unaltered-- at least in the
playback electronics. The LISTENING is far
superior-- the MEASUREMENTS are not superior.

SE amps will normally have poor distortion
measurements, and poor overall musical
bandwidth. Extreme efforts taken by the best
SE amp builders improve these quite a bit--
at extra expense, but measurements will never
approach those of even an ordinary push/pull.

My choice is SE amps with zero feedback because
I like to have music do what it is doing without
fuss or alteration. It's just plain better, period.

To LISTEN TO, not to measure.

There are two worlds out there, and the proponents
don't necessarily agree on much of anything.

SE amps require sensitive, highly accurate speaker
systems to play at anything near what the amp is
capable of sonically. The best of these high-
efficiency transducers were designed for movie theatres
where three things were considered paramount: (1)
a sense of reality and presense that sold tickets.
(2) High efficiency transducers in order to
PROJECT INTO a large space filled with people, and
burn less amplifier power, and (3) allow
smaller conductors to be used in theatre wiring..

These transducers sound best with SE amps because they
are transparent, and readily show-up balanced circuit
or push/pull processing as a sonic perversion of the
playback delights that were placed by the studio into the
recording being played back.

-Dennis-

 

RE: A tale of two amps, posted on May 22, 2021 at 12:59:24
tube wrangler
Manufacturer

Posts: 2484
Location: USA
Joined: January 29, 2007
 Many push/pull amps had bias adjustments
to keep idle currents identical with tubes
that differed.

This gave way to automatic bias tracking.

At first, this kept idle currents equal. If
you take two different tubes and force equal
idle currents, then when a signal hits the
two tubes, one has a different bias than the
other. So now, the SIGNAL is thrown off under
driven condition.

This gave way to automatic tracking of the signal
AND the bias. And sounded better yet.

One can see where all this is leading up to--
the thing is being forced into operating at
equal currents, and is being forced into
balancing the signal. The amp is also much
more reliable.

What's not to like? The !!@@##$$ SOUND!!!

With good, hi-eff speakers, you can hear all
of the correction mechanisms---- well, correcting.

Voila! Amp-corrected "music"!

Kind of like building a racing car with different
kinds of tires, and correcting each to a standard.

The question is: would you like to drive it in
racing?

-Dennis-



 

RE: The math, posted on May 24, 2021 at 12:15:30
cpotl
Audiophile

Posts: 1002
Location: Texas
Joined: December 6, 2009
 "Nelson Pass once told me that all push-pull amps suffer from some form of crossover distortion. Do you disagree?"

I wonder if he might perhaps have been alluding to the phenomenon known as gm doubling (or transconductance doubling), which is potentially a feature of all class AB push-pull amplifiers. In the low-signal regime where the amplifier is effectively operating in class A for the entire waveform, both output devices (tube or semiconductor) are driving the speaker, whereas once the signal is large enough to take the amplifier out of class A operation (i.e. in the part of the waveform where one or other output device is in cutoff), only one device is driving the speaker and the effective transconductance is halved.

John Broskie, Robert Cordell and Douglas Self, among others, have written extensively on this.

It could be thought of as a kind of crossover distortion, in the sense that it potentially implies a distortion associated with transitioning out of the overlap regime to the regime where one or other output device is non-conducting.

gm doubling would not be an issue for a genuine class A push-pull amplifier, though. (i.e. in an amplifier where the change of current flow with change of signal voltage remains in step for both output devices throughout the entire audio waveform cycle; one current goes up and the other goes down by the same amount.)

 

RE: A tale of two amps, posted on May 28, 2021 at 18:37:25
PaulF70
Audiophile

Posts: 1403
Location: Midwest
Joined: June 30, 2006
 More importantly, the vast majority of push-pull (or "differential") tube amps do not have any such real-time auto-correcting circuitry.

That means the pushin' and the pullin' ain't happenin' equally. That means distortion - crossover distortion - right where it hurts the most: where signals are fading to nothing.

SE tube amps + ESL speakers (NOT efficient) has given me more musical bliss, for the past year, than anything else, and I've tried everything else, many times.

 

RE: A tale of two amps, posted on June 3, 2021 at 17:39:05
PaulF70
Audiophile

Posts: 1403
Location: Midwest
Joined: June 30, 2006
 .

 

Page processed in 0.029 seconds.

[ Top Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2024, Audio Asylum, all rights reserved ]
[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]