• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Distortion discussion from mosqueto Hybrid thread

Gentlemen (and any ladies present) please continue here:

I see several problems in this discussion.

First off we don’t seem to be talking the same language or terms.

I’ll define a hypothetical scenario and hope we can go from there. Or just take off as you were but please keep it in this thread.

I design three amps. I’m a competent engineer or DIY hobbyist who understands tube theory and can read the tube specifications so I do a good job. I use the same tube in all three amps, and my output transformers are wonderful.

1. SE 6L6GB 6.25W at 10%thd

2. PP 6L6GB class a 12.5W at 2%thd

3. PP 6L6GB class AB1 25W at 2%thd

They are driving nearly identical speakers in that the frequency response, distortion, etc are identical, only the efficiency varies (by 3dB/WM) such that all three produce the same acoustic pressure levels with the same input signal.

Which has the lower distortion?

Which sounds better?


Personal ramble starts here so jump to the next post as you wish.

In the early 70s as I remember it, bigger was always better. Bigger engines, bigger amplifiers, bigger speakers, bigger everything. Then the gas shortage in 74 pushed people to start thinking (at least temporarily) about efficiency and power consumption.

Push pull amps were bigger (higher power output) and therefore they were better. They used less power for the same output as a SE amp (of which there were very few commercial versions) so from an efficiency standpoint they were better.

Distortion was reduced to a meaningless number by GNFB.

Stereo sales persons were told to crank up the volume to impress the customer, and mess with their hearing. (My room mate at Clemson University worked for one of the local up scale audio stores and had a lot of interesting stories to tell.)

Bose 901s are great because they take lots of power:( (no wrech imacon).

The fact is that PP amps naturally produce more odd harmonic distortion than SE amps. While the total THD (without GNFB) may or may not be less than a SE amp, the undesirable distortion is worse.

I believe the real reason big tube PP amps ruled the late 60s and early 70s was that most speaker efficiency was not that great. To get the loud levels that rock and later disco (remember that fad?) promoted, one needed lots of power.

Transistor amps were not necessarily better as much as they allowed music to be played louder. It was all driven by marketing and floor sales people.

Out of probably 30 friends at Clemson, I only know of one who had what would be today good speakers. That was my room mate Phil. He convinced me to get my first set of Heresy speakers and taught me a lot (what little I know) about stereo and sound reproduction.

I think I had the smallest amp of any of my friends (Sony STR5800SD 56wpc). But I had the most efficient speakers so I didn't care. I seriously doubt that any of them still have any of their stereo equipment. I still have both the Heresys and the Sony. Although the only reason I use the Sony now days is for it's phono input to play records into my computer to digitize them.
Which is lower distortion is an incomplete question and open to interpretation. To use your automotive analogy,m it's like asking which car is more efficient. The answer lies in the definition of efficiency. Is it miles per gallon? If the motor requires higher octane fuel, the answer needs a correction factor. Efficiency could also be horsepower per pound of fuel consumed, miles per passenger gallon (taking the advantage away from motorcycles and two-seaters). cost can even enter into the equation.

Back to audio, what is distortion? Just THD? How about a low frequency or high frequency rolloff? Not a "classic" definition, but it is a "distortion" of the original signal, if only part of the signal is played back. Ditto for low level information retrieval; if we don't reproduce all of the transients and hall sounds, etc., the sound is "distorted".

So when we quote distortion, any listening comparison will tell you, we are not measuring and reporting all of the distortions. How else to explain that a SE amp with 10% distortion is even listenable?

On your three amp comparison, I submit that the PP AB version is the lowest distortion. The amps have to be compared at a selected power level that all can achieve. So, 6.25 watts is the maximum comparison level. The SE amp will benefit in the comparison if a still lower level comparison is chosen.

I know it's crude, but I just check for visible clipping levels, crossover distortion and visible waveform distortions of other forms, then evaluate "distortion" with my ears.

I purposely defined all three as driving different efficiency speakers so as to produce the same audio pressure levels at the same input drive voltage. Presume also that all three amps (including OPTs) have the same frequency response, phase margin, etc.

This is done so that we are not disadvantage one amp (SE) over another (PP) solely based on output level capability.

Therefore we are comparing all three at the same relative output, but not the same absolute output.

All three are compared at clipping for max distortion.
First let me say that I have never built an SE tube amp. However, I built a 50% cathode follower push-pull amp and as a result of compromises I made in the design of the last voltage amplifier stage I got some extra second harmonic distortion.

Many go on and on about how second harmonic is a euphonic distortion, but what many also overlook is that the non-linearity in the transfer function that produces the second harmonic distortion will also produce IMD when two or more tones mix (Hey, that occasionally happens in music). That will happen with any non-linearity in the transfer function. The best amp is the one with the straight transfer function. You just have to minimize overall distortion, make sure you aren't producing too much of the really nasty-sounding harmonics, ensure slew rates aren't limiting things, make sure transients get reproduced accurately, make frequency response flat, and reach your target damping factor, making trades as necessary. It's really quite simple.

By the way, my amp sounds awesome except when the singer makes an "s" or a "t" sound with any sort of musical accompaniment, then I cringe. Oh, well. Adjusting out the imbalance should be easy enough.


  • 1WdistchanB.jpg
    236.7 KB · Views: 286
Hats off to you TheGimp.
Hello to the Rest,
Back in the good old days of working through school there was a company called Dynaco or Dynakit. I bought a Stereo 35 assembled by a tech at Aerojet. I also used a PAS 3. The speakers were JBL D130’s and 075 tweeters installed in homemade scoop horn cabs (JBL mailed the plans) made by me. I wish that I still had this stuff, it was all stolen. Yes it was about playing Led Zeppelin’s Heart Breaker loud. In those days a SET (to me) was the low power console HiFi used by my grandfather.
These days it is playing with SET headphone amplifiers, Salas’ RIAA JFET preamplifier and AudioTester 3.0.
Next will be Pete Millett’s Engineer’s PP amplifier and new to me JBL speakers. I have the 2225 woofers, 2123 mids and a replacement 075 tweeters. Grand Funk Railroad will rock again at my house.
AB PP to me is about more power per pound / dollar than better distortion. Yes the sonic signature is different but that is ok with me.
All just for fun!

M Gregg

Disabled Account
2010-06-28 11:04 pm

The distortion will depend if the amps are working in class A or AB, I suppose you could argue that amps that run in Class A and have the ability to run AB will give more power (Watts).

I must admit many SET amps I have heard in HIFI shows do not seem to have the "drive" for some types of music. However I suppose it depends on the "Nut" you are trying to crack! I have heard Ongaku!

You could even argue that if an amp cannot reproduce the whole spectrum of frequencies "Bass comes to mind" within the music, it is distorting it!:)

Just for fun!

M. Gregg
Last edited:
My guess is that amp 3 will sound worse than amp 2, because it will have some crossover distortion. For the same 2% THD it is likely to have more higher-order distortion, which is generally agreed to sound worse than similar amounts of low-order distortion.

Which of amp 1 and amp 2 sounds best may be a matter of personal preference - it is clear that some people prefer their sound "warmed up" by 2nd order, which amp 1 will have. Those who prefer accuracy, or 3rd order, will prefer amp 2. To some extent it may depend on tastes in music too.

In all this I have assumed that no NFB is used, as this complicates the issue.
But won't the SE output distortion be dominated by 2nd harmonic and other even harmonics, whereas the PP amp will cancel a good bit of the 2nd harmonic distortion.

Even though the PP has 1/5 the distortion of the SE, it should have a higher percentage of odd harmonics than the SE amp.

Based on that, it should sound worse.

But won't the SE output distortion be dominated by 2nd harmonic and other even harmonics, whereas the PP amp will cancel a good bit of the 2nd harmonic distortion.

Even though the PP has 1/5 the distortion of the SE, it should have a higher percentage of odd harmonics than the SE amp.

Based on that, it should sound worse.


6L6 in SE generates wider specter than EL34 or EL84, especially with higher than usual plate voltage and higher load resistance. They look like were made for class A amps.
But won't the SE output distortion be dominated by 2nd harmonic and other even harmonics, whereas the PP amp will cancel a good bit of the 2nd harmonic distortion.

Even though the PP has 1/5 the distortion of the SE, it should have a higher percentage of odd harmonics than the SE amp.

Based on that, it should sound worse.


Where is it written that 2nd harmonic distortion sounds good?

Attached are IM distortion results that go with my plot above (same amp same power level). The big spurs just off the 7k are the result of the same bending of the transfer function that causes the 2nd harmonic distortion, the next spurs are caused by the 3rd, etc. Wouldn't it behoove me to get rid of those big spurs caused by the 2nd?

2nd harmonic distortion may sweeten a single note, but screws things up for real music. There is no reason to put 2nd harmonic distortion into your music if you can get rid of it. A well designed push-pull amp shouldn't have too much higher order distortion, either. That's what we should be shooting for, distortion as low as possible with special care not to generate even minuscule amounts of nasty harmonics. Then pair that with some efficient speakers so it won't be working very hard.


  • IM1WchanB.jpg
    250 KB · Views: 161
Where is it written that 2nd harmonic distortion sounds good?

It does not sound good, even though it had been written many times as if it sounds good. But the same percentage of higher order distortions sound much worse. There were many attempts to replace THD measurement by weighted measurement where instead of summing of harmonics caused by distortions they would be summed with coefficients of nastiness. Such measurement would correlate better with sound quality, but initiators could not come to consensus about coefficients of nastiness.
Also, plain sum even of weighted harmonics is not sufficient: lower order harmonics mask presence of harmonics of higher order that immediately follow by them.

What makes things more complicated, when distortions increase with decrease of loudness it again sounds dirty. Especially, when their specter goes wider with lowering of signal level, so static measurement is not sufficient.
And it is all about plain non-linear distortions! Add here modulations of phase shifts by the signal, and things become even more complicated...
But won't the SE output distortion be dominated by 2nd harmonic and other even harmonics, whereas the PP amp will cancel a good bit of the 2nd harmonic distortion.

Even though the PP has 1/5 the distortion of the SE, it should have a higher percentage of odd harmonics than the SE amp.

Based on that, it should sound worse.


Hello TheGimp,
Sound worse?
Perhaps it does not sound at all? The PP amplifier puts out more watts. You do not need to use them all. At reduced output perhaps the higher order distortion may be lost in the noise floor. If I remember Chrowherst correctly 2nd harmonic plus fundamental frequency equals 3rd harmonic. If the 2nd is reduced so is some of the 3rd.
What does the FFT look like?
Kicking thoughts and not being a troll, cool.
All just for fun!
Wow! That is one hell of a peak at 7Khz.

Not to mention the inter-modulation distortion around that 7KHz peak.

There are IM peaks greater than the 2nd Harmonic peak.

Well, it's an IMD test. Injected signal was dual tone 250Hz/7kHz, so 7kHz peak shouldn't surprise. The spurs are the IM caused by the nonlinear transfer function. And yes, they are about the same amplitude as the second harmonic of the 250Hz fundamental. Ignore the blip about two-thirds of the way between 1kHz and 2kHz as this seems to be an artifact of my test setup. I honestly didn't think it would work this well with the built-in motherboard soundcard.

Also notice that I have an ugly tumor on my spectrum at 14kHz due to the second harmonic distortion.

Remember, IMD can come from power supply noise and other places, but any curvature in the transfer function also causes IMD and IMD does not sound good.
It seems to me I see PIM there... Sideband strips are of different heights.

I'm only aware of the mathematics of FFTs on a high level. Could the unequal sidebands be errors due to this just being a single FFT, as opposed to multiple FFTs averaged like the earlier post? The amp in question doesn't work at the moment or I would just fire it up.

The only thing I can think of that would produce PIM would be the FQPF2N90 source followers, but at 1W output from this amp there are still always a couple of hundred volts Vds at all times.

Something to look at in the future.
DT, my concern about high order harmonics is that there is the possibility that even low levels may obscure ques that are necessary for the brain to properly process localization information. This would result in disruption of the spatial interpretation.

What was the scale on the FFT? Was 0db one watt?

The total power on the 8 Ohm load was 1W (measured with one of those Pete Millett interfaces). I probably should have cranked down the attenuation since these levels are pretty low but I wasn't really in precision measurement mode when I took these, I was just trying to figure out how to use the software. You may notice that I hadn't figured out how to get it to compute IMD% yet with the software, otherwise it would be displayed at the bottom. Then a regulator went out on the amp and I decided to redesign the PCB before fixing it and there it sits. I should pull out a commercial solid state amp out and run the same test and see how it does just for fun.

By the way, what do you consider a "high order harmonic?"