Another big opamp listening test

Which of the files you prefer by listening?

  • aa

    Votes: 6 22.2%
  • bb

    Votes: 13 48.1%
  • cc

    Votes: 8 29.6%
  • dd

    Votes: 7 25.9%

  • Total voters
    27
  • Poll closed .
Status
Not open for further replies.
Hi,

just loaded the files on my server and did a first listening.
The differences are certainly small and I´m quite sure the source material isn´t anywhere suitable to make a proper decision.
One-guitar, a HiHat and one-female voice ... typical highend plingplong ... and what a annoyingly high pitched sqeeking mumbling girlie voice that is 🙄
It´s missing on bass, lower mids, dynamics, staging.
What made You choose such a meaningless piece of music for a test where the expected results would be small anyway? :scratch:

jauu
Calvin
 
Just for fun, below is the link to the file which is an amplified audiodiffmaker result between the sample that went through the "best" opamp in the test and the "worst" opamp in the test. To get just anything audible, the diffmaker result was then amplified by 26dB (20x). One can evaluate the "audibility" of the amplified difference between the best and worst opamp in this test. Best and worst meant strictly in technical terms, according to datasheets and measurements performed.

http://pmacura.cz/diff1_26dBamplified.zip
 
What? It sounded to me like the source for the listening test was a LP, and it has at least one loud pop in it.

Seeing as you mentioned it...

I hinted in my previous posts that there was more to this track than meets the eye (or ear). The click/s (there is more than one) and something that sounds suspiciously like 'groove noise) in the RH channel at one point. Yet it is way to good to be unprocessed vinyl, or could it be digital with something added.

You can also hear the background noise (hiss) fade in/out as the mix changes.

One thing is for sure as Pavel is finding out, it doesn't matter how much care you put into choosing a track and preparing files, nobody ever seems happy to take them at face value.
 
Mooly,
Yes, exactly.

Except I'm not sure what you mean about taking it at face value. If one listens carefully, it is as you describe. So in a way that is at face value because it is a statement of fact, not subjective judgement about whether you like it or not.

But, I would agree that you will always get some complaints from people about something or other. In some cases, criticism may be more well founded than in other cases. However, in this case its hard to criticize the selection of source material since the question is only about which file you prefer. I guess maybe there could be criticism for not having a button to select a preference for none of the files.
 
Karl, I think the origin of those non-musical sounds is an analog master tape.

If so, it still leaves the question of how they got on the master tape. The pops shouldn't be there for sure. I don't know why a mix engineer would leave some of those artifacts there for a mastering engineer to fix. It's got to be easier to fix those things at an earlier stage in the production process.

In addition to what Mooly described, I would add that all the wav files are distorted everywhere. For example, in close listening the vocals sound like distortion everywhere in all the tracks. However, some of that may be due to the test setup data converters rather than the source, I don't know. Again, this is just intended to be a statement of factual observation. I'm not trying to say I like it or don't like it.
 
Last edited:
Not kidding, I think 30 second slices of (obviously same original file) pink noise played through the 4 competing OpAmps would be much easier to compare.

Aural memory is very very short (unless difference is *gross* of course) and the song sample chosen has "too much variation" so, by definition, it´s impossible to focus on 10/12 different elements at the same time for comparison.

Remember we are looking for *minute* differences here.

If anything, we should listen to short segments of just one instrument at a time, in sequence. Human voice is also an instrument in this case.

As in, for example:
* the first 8 bars, almost 20 seconds, of the bass line.
That alone, passing through all 4 preamps sequentially, many times, say 8 times, so you can deeply focus on it, pick small possible differences and confirm/discard on next pass.
* then same bass line, end to end, again 8 times through 4 Op Amps.
Write down whatever impressions.

* repeat with the hihat, which plays all the segment, some 25 or 26 times, plus the splash played 3 times around second 30 , the "triangle" (don´t know how you call it) around second 46, it might also be a tiny bell.
Again write down impressions.

Repeat with all instruments, separated, or in a few cases together IF they belong to the same family as in the percussion elements I mentioned above.

Voice by itself is incredibly complex and revealing, when listened at alone will reveal lots of nuisances.

I must disclose that I am *heavily* involved in Music, both live and recorded, since forever (think 1969 and do the Math), have spent countless hours in Recording Studios, probably around 200 records, I live from listening to sound nuances and getting customers the sound they want.
Of course have thousands of hours listening to *separated* sound tracks such as I mention above, time and time again trying to improve them or make them match other instruments in the mix.
I listen to the sound sample offered in this test and clearly "see" how each layer was added , how each instrument was miked and equalized ..... and thinking how would *I* have recorded it 😉
As in that dreaded hihat which sounds too dull and gated too short, sounds as fake as a rubber coin.

Have also operated live sound for ages, although it´s completely different , there loudness is King, anything else takes second seat.

So sorry for finding some of the sound opinions expressed here (I mean all over the Forum) ..... how to call them? ... naïve, innocent, in the best case .... and thinking "how can they claim to actually hear this or that?"

And also being very skeptic about "virtual listening tests", for obvious reasons.

usually confirmation bias,ofcourse.

With due respect, in this case having 4 identifiable segments kills the blind test element, psychologycally once listener thinks he heard something in one track, will be expecting it and "hear it again" if you know what I mean.

Not my thread or test, but personally I would have built a long file, say each segment played 25 times, switched and mixed at random (for the listener) but of course known and recorded to poster, numbered some way, say each one lasts exact 1 minute or 30 seconds, so what player displays becomes the segment ID, and let listener write down whatever he thinks about each one.
As in: "I found 1, 12, 23, 42 dull and uninspiring, 5, 6, 15, 34 lively and transparent .... " etc.
No need to post results here until test is ended, they might be sent directly to OP for analysis, and statistically analyzed and disclosed at the end.

I BET many will be surprised by results, which will be quite eye opening.

Not a Politically Correct phrase here , but I guess many results , in this truly blind test, will be as accurate as flipping coins.
 
I will say more after the poll is over. However, the music choice has not made much (any) difference when I was preparing the test.

Let's face it, someone will always complain about the music choice. I remember SY posting a recording of a singer and guitar recorded in his living room with simple stereo mike and someone complained about the production values!
 
Getting back to differences between the opamps in the particular test circuit used, to the extent differences may be discernible, what I don't get is the point in asking which I prefer. What would it mean if there were some preference? Probably not much for most practical purposes.

What seems to be making the thread interesting is some of the discussion it has spurred, not interest in whatever the test outcome might end up being.
 
Bill, I asked some kind of a question. In fact I would like to know if anyone is able to tell the audible difference from the files where technically no audible difference exists. That is the point. And last, I am looking forward to see the reactions when the opamps will be disclosed.

In case that the thread has started further valuable discussion, I think it is a nice outcome.
 
Not kidding, I think 30 second slices of (obviously same original file) pink noise played through the 4 competing OpAmps would be much easier to compare.

Very true! In fact, I am doing this kind of test for myself. But the guys might find it boring 😉

There is just one more revealing test. Let it play nothing, connect to very low noise power amplifier and put your ear directly to the tweeter. Then you are able to distinguish all of the 4 opamps used 🙂.

The only question is how it corresponds to listening of music. Technically, the best is to make a choice by measurements.
 
There are differences between all the amps, and it may mostly be noise, but hard to tell exactly what is different in such terms for me. I would say some sound a little "smeary," but the effect is very small. And I don't know how that impression would translate to a measurement. Also don't know if it would show up in a different circuit.

But, I think you erred in your survey because some people would like to vote against an amp, rather for for one. You might assume in that case all the amps other than the disliked one would receive equal votes, but not necessarily so. If people didn't want to vote for the first one in the list, they might just pick the second one out of convenience. You would need to run the test multiple times with the amps in different orders to see if there is such an effect, otherwise you can't rule things like that out.

Probably would have been better if people could have rated each amp on a scale of 1 to 5, with a description of how to interpret the different numbers in the scale, like does 3 mean fine, but not exceptional, or does it mean I can hear some distortion/noise, but it's not a problem.
 
Last edited:
Bill, I asked some kind of a question. In fact I would like to know if anyone is able to tell the audible difference from the files where technically no audible difference exists. That is the point. And last, I am looking forward to see the reactions when the opamps will be disclosed.
.

I was just noting with a smile that some people prefer to complain than try. I have great admiration for you for organising this, although the usual spoilsports are not very active these days so you might not get too many spoilers. I must request to be excused from participating in this whilst the poll is open as a current domestic issue is preventing serious listening.
 
Also an interesting observation:

When I listened to the first sample, I said: wow!
At the second sample I said to myself: the first was better.
Then at each subsequent sample, they sounded alike.
Back to the first: it also sounded like all others...

Like with women: the first impression is the deepest 🙄
 
Yes I know Samuel Groner's work, great job on opamps measurements and finding differences.

...

I think we are saying the same just in other words.

Yes, Pavel, I think we are. I was just providing context to others about the measurements commentary at high noise gain. And want to redouble my thanks for your efforts here!
 
I'm trying to be positive here with constructive feedback.

Why are you wasting time with op-amps?

High open loop gain feedback devices are a waste of time, not to mention anything at the micro geometry scale like that of IC op-amp is never going to yield good sonic results, compared to large scale silicon, even if the design doesn't use feedback. Or vacuum tube for that matter.

Next, from the look of the circuit layout it has quite few problems. The power supply pin decoupling capacitors have ceramics close and electrolytic far away. Also they are in parallel which creates a high Q tank circuit which is not good for damping any impulse response and subsequent feed or re-feed back into the supply pins. PSRR and what you think you will not hear in the noise floor is unbelievable.

Also the ceramic you chose was a ferroelectric type and not a NPO or COG type? Is this correct? Big no no for audio to use ferroelectric ceramic decoupling anywhere in audio.

Next is there a star pattern or solid ground? Perhaps not an issue so much with audio, but in mixed signal with noisy digital stuff around one needs to think about layout and noise contamination.

Last best op-amp I heard by a country mile was Burr Brown OPA627. Quite a big difference to everything else out there with a price to match. But even it could be improved by biasing the o/p stage into Class A single ended operation with a current sink or source and pulling 5-10mA through it.

Also common mode impedance matching at the op-amps inputs yields surprising sonic gains, but is not always easily possible if the source impedance is dynamic. I think you would have trouble measuring a lot of this, but sonically it is easily recognisable particularly in a mixed signal digital/analog environment such as D-A converters where there is nasty RF and supersonic noise mixed in with the audio base-band signal.

Once you move to discrete and vacuum tube designs the op-amp becomes a poor substitute for quality sound reproduction.
 
Status
Not open for further replies.