Questions re:  GaNfet technology vs other designs.

@orchardaudio ,

I like the approach. Thank you for doing this for the community.

Putting on some popcorn.

@spenav - Perfect answer. GaN is just a component technology. Throwing it into a flawed design is not going to fix the flaws.

Don't let yourself fall into the trap of "musical" / engaging. Mainly that comes from a non-flat frequency response, and that could be correcting something else in your system or just be something you like. There are other easier and more reliable ways to do that.

@invalid ,

I quote,

No matter what speakers we used, every "difference" we thought we had isolated turned out to be there, in equal quantity, when we swapped amplifiers.

He probably could have and maybe just did most of his work with a resistive load adding in a set of speakers for verification. He wasn't going to give away all his secrets.

He did do things with his production models, but it would be good to question what are real sound difference, what is voicing, and what is marketing.

You mean @atmasphere's unsubstantiated claim versus an effect (impedance impact on frequency response) that absolutely is undeniable. I find that most people make 0 correlation between listening results and the underlying cause so assigning cause to a result without knowing what it is does not lead to a valid conclusions. Bob Carver's challenge is a strong data point.

No, my statement is absolutely correct @atmasphere . Read my words carefully. "MAINLY" comes from a non-flat frequency response. That is especially true of most tube amplifiers such as the high output impedance of your own OTL amplifiers. The interaction of that output impedance with a speakers impedance will totally dominate any potential distortion profile your amps may have. Ditto for things like D’Agostino amplifiers. The effect is pronounced because the largest change in speaker impedance curves occurs where the equal loudness lines for the Fletcher Munson curves are close. Most amps have their lowest distortion here and hence not affected.

Tube amps due to typically low feedback would be expected to have mainly compressive distortion which tends to reduce perceived volume, however as we know, even the ability to detect low volumes of distortion, is not great so having much impact to coloration of the tone is questionable and certainly not exceeding the effects of impedance on frequency response in real world implementations.  Contrary to your technically correct comment I have seen you make about feedback levels and high frequency THD, I don’t think you have a lot of evidence to support it based on a multitude of full bandwidth distortion measurements by the likes of Stereophile and ASR showing most amplifiers, "no feedback" or otherwise having low distortion out to 20KHz, even going back some time on Stereophile. You have made claims of very low distortion being audible in other posts, but I can find no evidence provided to support those claims. I am quite certain there is no evidence to support a claim of tonal differences based on these very low distortion levels out to 20KHz. The compressive/expansive effects of distortion in audio electronics would pale compared to what happens intentionally or unintentionally in the recording process (and speakers at reasonable levels).

Keep in mind I am not discounting that distortion can impact perceived tonality, and the nature of the distortion, compressive or expansive will impact that change in perception. I am saying that at the practical levels of distortion, unless the amplifier is very high in distortion, the impact will be significant less than impedance effects and between a lot of amplifiers the distortion is low enough that it will have 0 impact on tonality.

@charles1dad , I would not be so quick to jump on the bandwagon.

@snapsc, have we all forgotten the Carver challenge already? https://www.stereophile.com/content/carver-challenge  -- Bob, took one of his SS amps, a rather run of the mill one, and in 4 days, made it sound exactly what was assumed to be a tube coupled tube amplifier. His goal was to match the transfer functions, or specifically the input to output relationship with a real load, which in laymans terms the frequency response (magnitude and phase) measured at the speaker terminals. Once he did that, they sounded the same. Did he perhaps add in some distortion on his amp to match the other amp? Potentially. He of course did not detail the changes, but the vast majority of the nulling requirement would be to match the frequency response at the terminals of the speaker. That is how you make two amps sound effectively the same. If you drive the amps into clipping, all bets are off. There is no indication that was done in the Carver Challenge, but a comment indicates he had to add lots of impedance at low frequencies to match what was happening on the two amps (drop it from 500 watts to <100 watts).

OTLs are an excellent example as they will have a totally colored response compared to probably virtually any other amplifier type with any given speaker. You can choose one that works best, but no matter what you do, it will sound different with other speakers. A quick review of your posts shows you pairing it with speakers that were not intended to be used as such.

Your interpretation of how the ear perceives harmonics is too simplistic to be useful as a rule. Here is a link to help: https://www.researchgate.net/publication/323914860_THE_EFFECTS_OF_DISTORTION_ON_THE_PERCEPTION_OF_LOUDNESS_IN_LIVE_SOUND Low feedback will result in mainly compressive distortion. Insufficient feedback will result in expansive distortion using their defined parlance.

W.r.t being easy to demonstrate with simple equipment, I agree. I low distortion amp, low distortion amplifier (across 20KHz), and low distortion headphones. Then use a distortion simulation program like Distorter (VST plug in) to simulate various level (and orders) of distortion, or Pkanes Distort. Sure if you did 0.1% distortion of purely higher order for a 500Hz tone, you will hear it before lower order distortion but you and I know that is not a realistic occurrence. And this is detecting with a pure sine tone at optimal frequencies. With real music, no way.

Take a perfectly digital source and low distortion SS amp. Then add in the distortion profile of a typical SET amp especially the 2nd/3rd distortion, at levels one could expect outside clipping. Then blind A/B them with music. It won’t sound warm or cold. On general music you will notice the distortion before you notice any warming ... and yes, I encourage everyone to repeat this experiment themselves if they have a low distortion DAC/AMP and preferably headphones as lower distortion but you could use speakers. Please don’t take my word for it. That whole "warming" due to typical distortion components of tubes is just Philelore.

Yes the brain does have a variety of tipping points, but it also has the ability to extract the underlying information out till things get really bad. It will not pay attention to this distortion tonality more than it will the raw frequency response.

So where should I be looking for this connection between measurements and audibility? Feel free to send a link with some sort of objective listening test of distortion (and reasonable levels) versus tonality, where the only variable changed is distortion. I don't know how you cannot find information explaining why higher order distortion is more audible. This is a readily known thing. Higher order harmonics will not be masked by the primary tone. That is for tones. They will be masked in music.  For pure tones, Fletcher-Munson comes in again, because at low frequencies, the higher order harmonics will be areas of higher sensitivity.  But SS amps typically don't have much distortion at those frequencies. Keep in mind with pure tones, the masked harmonics will themselves mask the higher orders for typical distortion profiles (to a point). That coupled with music again limits the practical audibility of higher order harmonics of real equipment. Philelore.

@sdl4, no one is discounting that higher order distortion is more audible than higher order harmonics. This is not in question and the mechanism is very well understood. It is due to masking functions and the shape of our sensitivity to different frequencies.

Nelson's ideas of very low distortion is probably not what we think of as very low distortion, as well, Nelson discusses feedback in SS state in other papers and how that translates to distortion over frequency and this relates to the claim, I consider false, by Atmasphere that many SS amplifiers with negative feedback suffer from too little feedback at high frequencies. This is a flawed premise on several fronts:

1) This is just not factually true based on an easily done review of tests by Stereophile and Audioscience where even relatively inexpensive amplifiers have low distortion out to 20KHz, and this appears to have been the case for many years.

2) Because of the upward curve of hearing sensitivity, which reduces after about 3.5Khz, distortion at high frequencies becomes harder and harder to hear. If you think about it, at 3.5KHz, the 4th harmonic is 14KHz. Many people here can't even hear that. At 5KHz, the 4th harmonic is 20KHz, inaudible. Where higher harmonics are most audible is at low frequencies where higher harmonics are where our hearing is even more sensitive than the fundamental. 100-1Khz is a prime area. However, all those SS amplifiers with claimed insufficient feedback have tons of feedback at <1KHz (if they have Negative feedback).

This whole amplifiers that measure well and sound poor is nothing but Philelore, started by amplifier companies that make amps that measure poor and then chanted by their customers. If you have an amplifier that measures well, with today's ready suite of measurements, and sounds bad, it is not the amplifier that is the issue, it is your system. If a poor measuring amplifier sounds better, then it is just its poor transfer function complimenting the rest of the flaws in your system. It is one way to achieve a desired end result, but it is 2022 now. There are better ways if audiophiles would just get out of their own way.

In terms of phase from an amplifier, there are readily available metrics for what is audible in phase change. Most modern amplifiers are so far below audible limits w.r.t. phase changes that any discussion of it is a total waste of time. However, go back to those bad measuring amps that sound good, and perhaps you are complimenting a flaw in your system (that should not be there).

Last, keep in mind, that many of these claims of distortion, at the levels of the amps in question causing tonal changes, has never, what 50 years later, be shown to be fully factual even though experiments are readily done. I think most would be rather surprise how hard it is to detect distortion when listening to music. Far far harder than with tones, even when you use higher order distortion products. What is needed is far higher than one would ever expect to see in an amplifier today.

Nelson Pass obviously has a huge following in audio, mainly due to his strong involvement in DIY audio. He is a far more public figure than almost any other audio amplifier designer. However, I suspect few professional amplifier designers (and not just hifi) have spent much time studying his work, where they have spent a ton of time studying Doug Self, or a whole host of people people who listen to audio have never heard of.

Unless you have a speaker with a near flat impedance over frequency, it is impossible for that to sound neutral compared to a high damping factor amplifier.

This statement is simply false- we’ve compared exactly that side by side. I concede that you do have to be careful about the speaker that is being used. But it does not have to have a flat impedance curve! As I pointed out in the article to which I previously linked, the speaker needs to fall into the Power Paradigm of design rules (an example is a speaker designed to work with SETs, which have a similar output impedance to our OTLs).

My statement is absolutely true. Yours is not. I said it is impossible for your OTL amplifier to sound neutral compared to a high damping factor amplifier with any speaker that does not have a flat impedance curve.

Whether the speaker is designed to sound best (flattest) with a constant power, constant voltage of constant current does not make my statement incorrect.  A speaker designed for constant power delivery (uncommon these days) may sound better with your amplifier, but it will not sound the same as a high damping factor amplifier with that speaker. The two will sound much different and have a much different system level frequency response.  You will note I qualified with dynamic speakers of 8 ohms or less. A high impedance speaker will be less impacted by the output impedance of the amplifier obviously.

At a basic level, with your high output impedance OTL amp and a high damping factor amp, with any speaker with a typical varying input impedance, no matter the design rules, it will become almost impossible to level match such that you can even do a proper A/B comparison as the frequency response of the system will be so different.  We are dealing with basic electrical parameters here that cannot be simply had waved away.

Its funny how so many people think the ear stops using its perceptual rules beyond an arbitrary point on a bit of paper.

It is funny how many people arbitrarily make claims that cannot be backed up with demonstrations, but worse, are outside the realm of possibility.  I didn't even call you on the fact you said well over 105db down, I gave you the benefit of 105db down. Well down to most would mean another 10 or 20db lower.

@theaudioamp I see you dropped the idea that a GaNFET amplifier by virtue of GaNFETs must have load dependent frequency response. That's good- Occam's Razor wasn't supporting that idea at all. 

@atmasphere , I never ever said this. Not even close.

We seem to be seeing two messages from you- that harmonic distortion is and at the same time is not audible.

Also a gross misrepresentation of what I said.  I have been very clear.  Distortion is audible when a high enough level, end of story. I have also been clear that distortion is less detectable with music. I have also been clear that frequency plays into distortion audibility. There is no contradiction in anything I have said.

If  you will set aside the handwaving for the time being, the ear's sensitivity to higher ordered harmonics has been known for a very long time-

And the reasons also known, with perfect clarity for a very long time. Is there a point?

Regarding the effects of insufficient feedback, I refer you to Norman Crowhurst, who wrote about how feedback adds distortion of its own (including IMD) back in the late 1950s.

Do you have a point. You bring out this tired argument, it appears time and time again. No one disputes this is the case. However, you fail to provide evidence to support this is at all an issue. As noted, a quick perusal of Stereophile reviews and ASR shows most solid state amplifiers from the last several decades and certainly today having very low distortion out to 20KHz. You also ignored my very salient point that masking of lower order harmonics coupled with hearing degradation and the slope of Fletcher Munson curves means out past 4KHz, if not lower, higher order harmonic distortion becomes less of an issue.

w.r.t. Bruno, I will link his article here:  https://linearaudio.net/sites/linearaudio.net/files/volume1bp.pdf  which is a good dissertation on distortion and why negative feedback is good, which I am certainly not disputing, and which most SS amplifier vendors use enough of (as noted in measurements by Stereophile and ASR). Bruno does not talk a lot about audibility of distortion in this article. Mainly hand waving. He did make an off the cuff comment comparing linear amplifiers to Class-D and typical input stages off linear amps. This is why as engineers off the cuff statements should be avoided as most Purifi amps are front ended by just those same type of circuits buried in op-amps with enough feedback not to be an issue.

You know where Bruno spends a lot of time talking about distortion? Speaker drivers. it is also where he has stated he will spend his time, on speakers, specifically complete amp/speaker combinations. He acknowledges this is where most distortion lies.

I saw stated, and please correct me if I am wrong, that your OTL have output impedance in the 2-4 ohm range?  Unless you have a speaker with a near flat impedance over frequency, it is impossible for that to sound neutral compared to a high damping factor amplifier. It is simply impossible. With typical dynamics speakers at 8 ohms and less, that is going to make a very noticeable change in the frequency response. I personally do not consider that neutral.

@theaudioamp Hm. On that basis we’re on the same page. Since you seem to think we are not, I’m disposed to think that you’ve not stated your position all that clearly in the past. There is also the issue of what is meant by ’when a high enough level’. IME that means well over 105dB down. Most amps can’t do that. The bit that I think gets ignored here is the fact that the ear uses the higher ordered harmonics to sense sound pressure, and due to a +130dB range is far more sensitive to them than most people think.

Now you are just making things up. This is as technically ludicrous as the CD demagnetizer someone reminded me of.

Let us unpack this fully. You are claiming that harmonic distortion 105 db below the fundamental is audible. Let us take an example of a 90db/watt efficient speaker, with 200 w/channel, what most would consider pretty loud. That is 90db/watt at the speaker. Let’s say 8 feet distance in a typical room.

• 6 db gain for 2 speakers
• 8db loss for listening distance
• 3db gain from reflections

Let’s ball park that at 110db peak at the listener and 113 db at the speaker. Sure you can get more efficient speakers and more wattage (be careful of real efficiency versus in-room), but most people would consider 200W/chan with 90db/watt as a setup loud.

Most people’s listening rooms that are not dedicated are probably 40db background, with some lucky to be 35db. Dedicated rooms you may get down to 25db with reasonable sound deadening. Keep in mind simply breathing is 10-20db. Now I realize this is broadband and the spectral energy is spread out and we can detect tones below the noise floor, but it still provides perspective. 105db on top of 25 is 130db (and we only have 110 at our disposal). Realistically you can probably hear down to a 0db tone around 3Khz in a really quiet room. That is a tone, in a really quiet room. I have to have at least a 105db primary tone, at say 300Hz, and that is going to mask a 0db tone at 3KHz.

But let’s go back to where we started. I have a primary, that is limited by my speakers to 110db, at the listening position, but you are claiming that a 5db harmonic is buried in there, that at 110db listening level is going to be heard. Really now? Do you want to walk that claim back a little? A harmonic 105db down is 0.0005% distortion from that harmonic. And you think this is audible in music? I am curious what playback device (speaker or headphone) you managed to do a test where you showed 0.0005% added distortion of a high level harmonics, in music, was audible, when the peak level was 110db to start. Why do I ask that? Because I want to buy those speakers or headphones as they are probably the lowest distorting ever made for home use. Even horn speakers at those levels have audible distortion in controlled circumstances, not to mention the masking and IM distortion induced by real music.

I would not expect anyone to take my word for this. Simply create a file with a single tone at any frequency you want, probably 200-300 Hz is best, then add in another tone 105db down anywhere you want and see if you can hear it. I would suggest about 2-3Khz.

Now seeing your posts on other forums, I see this is a pointless discussion. You will keep making unsupported claims, even if they violate what we know about biology, the spectral composition of real music, the fundamental properties of real world speakers, and the limitation of the volume levels that almost anyone listens to music with.

@atmasphere you are twisting my words to attempt to win an argument. Clearly I said your OTL amplifier cannot be neutral compared to a high damping factor amplifier. You have said nothing that disproves that.

For your other amplifier, are you saying it is difficult to measure because it is low? And you are claiming it sounds "the same" as your OTL amp? I don’t believe you if that is the case. I am sure I could find some high impedance speakers where it may, but for most dynamic speakers, that just will not be the case. The frequency response variations are unavoidably large. In casual listening, you may convince yourself they are similar, in critical listening, they will not be.

Ever hear a system that sounds loud? If you work to really get rid of those pesky higher orders, you’ll find that even at high sound pressure levels on a meter that the system no longer sounds ’loud’.

I see no point in this post. Though I am almost always in ear plugs, I have done enough with pro-audio to know loud. Does it sound loud. Of course it does, even if the distortion is low, but it 0.5% low, which with music is inaudible.

You said that a harmonic 0.0005% down is audible. That is a 5db tone with a 110db primary frequency peak at the listening position. I really should clarify more. That is a 5db tone, when in addition to the 110db primary tone (the limits of our system and most people’s systems), that any non electronic music already has a ton of natural distortion that will itself mask electronic distortion including higher harmonics as the natural higher harmonics will mask other induced harmonics. This is one of the reasons why it is so hard to detect additional distortion in real music as there is already so much natural distortion.

Saying that "loud" is possible does not negate the lack of viability of what you have claimed. Put another way, it does not matter if it is SET or any other technology. You are starting with a conclusion and trying to work back toward a reason without showing the conclusion is true let alone viable.

I'm not twisting your words, simply pointing out that without direct experience you simply are not in a position to know the truth of the matter; i.e. you are wrong. Not a judgement or anything, just didn't get the 'facts' right is all (and still abundantly clear you didn't read the article at the link I provided).

I don't need personal experience to know that 4 ohms of output impedance M-U-S-T have a significant impact on the frequency response of any normal speaker with typical impedance swings. There is no talking your way out of that or trying to say it is my inexperience. Nothing you posted in any way negates anything I am saying or in any way supports what you are saying. I am quite familiar with what you referred to in voltage and power paradigm, and your reference to current source.  IF an amplifier follows a power or current source paradigm, its output versus input transfer function into a varying load impedance MUST be different from a pure voltage source transfer function, and it must be different because the output impedance is effectively different, and must be different to support either constant current (infinite) or constant power (variable) versus constant voltage (0). Your OTL amps appear to effectively be high output impedance, not that variable, and you just said your new GaN are hard to measure, which appears to suggest they are 0 (or low).  If anyone is being not clear hear, it is not me.

The reason the two amps sound similar is the distortion spectra is very similar (although fir entirely different reasons).

No, you are claiming this is the case, while just hand waving away the vastly different frequency response that must exist that will greatly exceed any impact of distortion, which you have provided no evidence of audibility beyond hand waving about concepts that most will not discount while claiming that distortion >> 105 db down is audible which is biologically impossible due to masking, impractical and unlikely due to the harmonics already in music, and physically impossible due to limitations in speaker distortion, and the extreme volume levels that would be required.

I can't help you with not getting the point of my comment about loudness; FWIW a lot of people reading this thread will get it.

You can't help anyone with this as you have provided 0 evidence to support your supposition, and have not even showing viability within the limits of biology, practical music, or physical implementation. You have basically ignored all three practical limits and stated something that fails all 3.  You are making the extreme claim. The onus is on you to provide validity or at a bare minimum, even if possible, let alone audible.