Do all amplifiers sound the same? Level matched listening test

[MaxwellsEq]

Indeed you can measure TD+N with WN using FSFA measurement.

It roughly looks like this, a quick, dirty one:

View attachment 473757

View attachment 473758

Interesting. Can you supply Total Harmonic Distortion alone - in other words separate out the noise from the above graph? All power amplifier specifications are relative to a nominal THD usually at 1 kHz.

 

[Sokel]

Interesting. Can you supply Total Harmonic Distortion alone - in other words separate out the noise from the above graph? All power amplifier specifications are relative to a nominal THD usually at 1 kHz.

I edited it adding a WN one as well.
By this method we can't separate noise or HD, it only gives total TD+N results.
It's explained here:

FSAF Measurement

 

[Sokel]

Interesting. Can you supply Total Harmonic Distortion alone - in other words separate out the noise from the above graph? All power amplifier specifications are relative to a nominal THD usually at 1 kHz.

Partly scrap the above, you can also see noise but if I have studied the paper well enough, it's the residual noise prior to the measurement.*
I'll have to look at this one.

But it looks like this:

*Edit: nope, it's the actual noise.

 

[pma]

I listened a few times before running the Foobar ABX test and was convinced I could hear a very, very slight difference. I felt amp_Y had a slightly different presentation of vocal edges and decays. To be clear, they sound so very similar I wasn't sure, so ran the Foobar ABX test. I deliberately chose to use the entire samples including the long introduction rather than shortening the test to the vocal bit. In the end I got 10/16 - so possibly a subtle audible difference but statistically not quite significant. BTW, I'm not going to say what I think the "tell" is. I used TRUTHEAR x Crinacle Zero BLUE, no EQ and the room is very quiet.

foo_abx 2.2.1 report
foobar2000 v2.24.6
2025-09-02 10:54:26

File A: PMA - Amplifier Comparison - 1.01 - amp_X .wav
SHA1: ef3b0eb1670f5332234d52e12c17eb0d7f2dd80a
File B: PMA - Amplifier Comparison - 1.02 - amp_Y.wav
SHA1: 609806867a3232fbb56fe54e795dddddc6c71280

Great, thank you! I do not intend to belittle your test result, however 10/16 is still not enough as a confirmation, with its p-value of 22.7%. We need 5% or less, thus 12/16 for a single test. I know, from my hundreds of ABX sessions, that 10/16 can be still achieved by chance, even if you do not listen and make a random choice of X and Y. Again, not to belittle your result, but we need more. And, cherry picking of test results is not possible. In case of multiple ABX tests, the results are summed. If you get 10/16 three times in a row, the total is 30/48 and this would make 5.57% p-value, probably good enough to be a valid check of audibility. The tests may be done with pauses, not necessarily in a single sequence.

 

[MaxwellsEq]

Great, thank you! I do not intend to belittle your test result, however 10/16 is still not enough as a confirmation, with its p-value of 22.7%. We need 5% or less, thus 12/16 for a single test. I know, from my hundreds of ABX sessions, that 10/16 can be still achieved by chance, even if you do not listen and make a random choice of X and Y. Again, not to belittle your result, but we need more. And, cherry picking of test results is not possible. In case of multiple ABX tests, the results are summed. If you get 10/16 three times in a row, the total is 30/48 and this would make 5.57% p-value, probably good enough to be a valid check of audibility. The tests may be done with pauses, not necessarily in a single sequence.

Thanks Pavel - I agree , 10/16 is simply not statistically significant. I actually wrote that in post #119!

The music is sufficiently listenable to possibly repeat the test, but I don't think I could do three sets of tests.

 

[peng]

Don't ask me - I'm not an amplifier designer.

So you believe an amplifier designer would know the answer?

Simple question, Anthem said the following in the FAQ section:

None of our components are designed with a "sonic flavor" other than playing exactly what's in a recording. Unfortunately with pop CD mastering, pushing levels way into overload regardless of how much distortion this adds is all too common. Recordings of acoustic instruments with minimal or no processing during mastering sound more natural, therefore they are a much better test of how natural-sounding the playback equipment is.

Sound reproduction equipment doesn't know the difference between a music signal and a movie signal, or for that matter the musical score within a movie soundtrack. Accurate for one means accurate for the other.

Given what they said, would you say their amps would sound different than the ones you like, and/or you amps not designed to be accurate/or transparent? I have to ask because I am curious to know how others pick their amps, I thought most would pick those designed for transparency/accuracy, and would then do whatever they like to EQ the output, or use harmonic generators to add harmonics to their liking, if they prefer certain kind of flavour. If I want to pick an amp that has the flavour I like, then I would have to audition many amps in my setup, that would not be practical, whereas, to pick one that is transparent, I could at least rely on measurements like ASR, or Stereophile's or like @pma and others on ASR, who did their own measurements (I intend to do that myself soon).

and a follow up question, another amp designer, also an EE and who was quite well known because of his successful and popular design of the so called current dumping (sort of class A amplifier based on the "current dumping" design concept) and speakers apparently did not do listening tests in his design process for the purpose of designing for a certain sound quality that is "better", but would do so in the end to check something that actually matters.

Given that, would you say QUAD amps are not going to be on your short list, because the designer apparently did not use the "trust your ears" approach to guide his design?

https://theartofsound.net/forum/showthread.php?13291-Interview-with-Quad-s-Peter-Walker-in-1978

So I really would like to know what your logical thinking is on this topic, thank you in advance for sharing.

Don't tell me that an amp designed doesn't ask people to listen to and comment on its sound before they release it on the general public. Beta testing, etc

No problem, I wouldn't tell you unless I have supporting facts.

If you can't tell a great sounding amp from a good sounding one, get your hearing fixed, rather than look at measurement to offer the answer.

Hearing, unlike seeing, cannot be fixed at such, once one lost the ability to hear perfectly in the audio band, such as 20-20,000 Hz, that's it. And, I guess you are just joking, but still, the fact is, aside from temporary hearing issues due to certain sickness, or just too much wax etc., the popular use of hearing aids can compensate the loss to some extent, but then "amplifiers, that such devices use" will get in the way again, complicating matter much more. Hate to repeat this, but we both know measurements, if done well, are not much less subjective than going by ears.

To me, logically speaking, the biggest issue with going by ears is, by whose ears? Amp designers, or the so called "Soundmasters" that Denon and Marantz claimed they rely on, or trust our own ears but then are we going to randomly purchase/return, borrow a great number of devices so that we can try them all on our setups? Or you would still use measurements to make up a short list of say a dozen of devices, to make it more practical in terms of time and money in the process?

 

[pma]

@MaxwellsEq You can do it in three days, in a week, as you like … one test per day. It is still valid.

 

[jEDGEc0m]

Did you test it? I don't think it was that hard. :
View attachment 473655View attachment 473657

If I can do it ( I never get it right on other similar blind tests) , then you should be able to do it too.
(up 1 dB compared to down 1 dB and flat was the easiest to spot)

Thought I'd give that a go. certainly is way easier when it is +1dB, but still, could hear.

0.5dB was much more challenging. felt like some luck in this result to be honest. I don't agree that it "feels like" i successfully passed.
This was on speakers. i think it might have helped to make sure the volume was fairly low. sine tones do weird things in a room. better not move your head around.

Moderately fun. Thanks

 

[peng]

Try foobar2000

foobar2000 for Mac

foobar2000

www.foobar2000.org

They have an ABX comparator plugin:

https://www.foobar2000.org/downloads/foo_abx-mac.fb2k-component

Good info, thanks. I stopped using Foobar after I purchased JRiver because I thought if I spent the money I should use it. I wonder if JRiver has something similar, if not I would use Foobar again if it is still free lol...

 

[pma]

It is free and probably the best protocol.

 

[pma]

@peng : never believe company advertisements, white papers and people who are involved in a product with some kind of money profit.

 

[peng]

That's a test I keep asking for, preferably one that doesn't last a few ms.
Spectrum is additive and we keep forgetting that.

I don't enjoy repeating the same spiel of mine on the "power" term, but I would repeat this for convenience. So, I guess the test your keep asking for might actually be: What's the maximum Voltage and current the amp can deliver, with pink noise input, and for what duration?

I don't think power should be use in this case, because people (as they always popped up and saying something...) would say it is not the "watts", but current that we need to worry about blablabla..). The fact is, if the max power is say 500 W into 4 ohms with test duration lasting 2000 ms, it would seem like a good number but we would not know how much of that 500 W was dissipated in the amp's output stages. One way to get around that would be to do something like the "power cube" (imo outdated..) tests, or Amir's reactor load test (I prefer that) and I assume you probably referred to that test, but one that last longer than a few ms, and with pink noise instead a 1 kHz, am I right about that?

 

[pma]

So, I guess the test your keep asking for might actually be: What's the maximum Voltage and current the amp can deliver, with pink noise input, and for what duration?

IMO such request is pointless in case of amplifiers. It makes sense with speakers. Music cannot be substituted by a pink noise. Only short passages might be considered such, especially when DR is high. Such pink noise power parameter for amplifiers would be misleading. Noise is a stationary random signal. Music is a non-stationary transient signal. Quite a big difference.

 

[peng]

IMO such request is pointless in case of amplifiers. It makes sense with speakers. Music cannot be substituted by a pink noise. Only short passages might be considered such, especially when DR is high. Such pink noise power parameter for amplifiers would be misleading. Noise is a stationary random signal. Music is a non-stationary transient signal. Quite a big difference.

Fair enough, my point is more about V+I instead of just P, as to input pink noise vs sinus, I get your point, though it wouldn't hurt to have both if time consuming is not the issue.

 

[pma]

Yes, Vrms into specified load.

 

[Adis]

I don't enjoy repeating the same spiel of mine on the "power" term, but I would repeat this for convenience. So, I guess the test your keep asking for might actually be: What's the maximum Voltage and current the amp can deliver, with pink noise input, and for what duration?

I don't think power should be use in this case, because people (as they always popped up and saying something...) would say it is not the "watts", but current that we need to worry about blablabla..). The fact is, if the max power is say 500 W into 4 ohms with test duration lasting 2000 ms, it would seem like a good number but we would not know how much of that 500 W was dissipated in the amp's output stages. One way to get around that would be to do something like the "power cube" (imo outdated..) tests, or Amir's reactor load test (I prefer that) and I assume you probably referred to that test, but one that last longer than a few ms, and with pink noise instead a 1 kHz, am I right about that?

Are you looking for something like this:

 

[peng]

Are you looking for something like this:

Not really, those are typically available already, published by manufacturers such as Benchmark, McIntosh, Bryston, Anthem, Hypex, Purifi, and many other reputable manufacturers.

I would be looking for something more like the following:

20-20,000 Hz, 40 Vrms, 7A rms

then we don't have to rely much on the load impedance and/or phase angle, because we are given the maximum voltage and current.

Just a quick example though not a very good one, if your load impedance is 2 ohms minimum, you can calculate the Voltage using the simple Ohm's law, that is V = IR = 2X7 = 14 Vrms and that is within the 40 Vrms rated voltage so you know the amp can drive your speaker as long as it does not demand more than 7 A rms.

Now if you still want to know the less useful so called power, or watts, then you use the simple power formula, that is P = I^2XR, or V^2/R, or in this example, P = 7x7x2 = 98 W.

But again, P is less useful because what does 98 W really mean (aside from the obvious lack of phase angle/reactive power consideration) practically speaking anyway, why would you be interested in it as long as you know whether the amp can deliver the current the speaker load demands under specific conditions (such as duration, peak etc.) and the amp's rated output voltage? It is still not going to be a perfect spec, but more useful than just to say 110 W into 8 ohms.

 

[Adis]

Not really, those are typically available already, published by manufacturers such as Benchmark, McIntosh, Bryston, Anthem, Hypex, Purifi, and many other reputable manufacturers.

I would be looking for something more like the following:

20-20,000 Hz, 40 Vrms, 7A rms

then we don't have to rely much on the load impedance and/or phase angle, because we are given the maximum voltage and current.

Just a quick example though not a very good one, if your load impedance is 2 ohms minimum, you can calculate the Voltage using the simple Ohm's law, that is V = IR = 2X7 = 14 Vrms and that is within the 40 Vrms rated voltage so you know the amp can drive your speaker as long as it does not demand more than 7 A rms.

Now if you still want to know the less useful so called power, or watts, then you use the simple power formula, that is P = I^2XR, or V^2/R, or in this example, P = 7x7x2 = 98 W.

But again, P is less useful because what does 98 W really mean (aside from the obvious lack of phase angle/reactive power consideration) practically speaking anyway, why would you be interested in it as long as you know whether the amp can deliver the current the speaker load demands under specific conditions (such as duration, peak etc.) and the amp's rated output voltage? It is still not going to be a perfect spec, but more useful than just to say 110 W into 8 ohms.

Yeah, ok. I misunderstood, I just skimmed what you were looking for. I thought that in addition to a simple sine measurement, you were looking for a wide spectrum measurement, which was probably made with pink noise...

 

[pma]

Sorry, it is impossible to measure distortion with pink noise directly. It is possible to do so with MLSSA, which is a different signal and a different method.

 

[NTK]

Sorry, it is impossible to measure distortion with pink noise directly. It is possible to do so with MLSSA, which is a different signal and a different method.

I came across this FSAF (Fast Subband Adaptive Filtering) measurement in REW, which is a pretty new technique. The earliest reference John gave was from 2020. It looked interesting. I have not tried or looked further into it myself.

FSAF Measurement

John's discussion thread:

FSAF (Fast subband adaptive filtering) measurement

This thread is for discussion of the FSAF measurement process. The FSAF process was developed by by Michael Tsiroulnikov aka Michael Zrull. See See Michael Zrull (2020) Fast Subband Adaptive Filtering (FSAF) MATLAB Central File Exchange and Michael Tsiroulnikov (2022) Loudspeakers for AEC...

www.avnirvana.com