Purifi Class D

[Matias]

13 hours ago, barrows said:

I have found there is often a correlation between getting good subjective performance, and some additional current headroom for the output stage driving the amplifier (whether we are talking a preamp or a DAC).  A theoretically overbuilt output stage often sounds better.  I tend to suspect that IC opamps do not perform at their best when pushing their envelope in terms of current-but I do not have the proof of this.

Indeed, and if you notice, the top dCS DACs also have incredibly low output impedance. Like on Stereophile review of the Vivaldi Apex "The balanced output impedance was an extremely low 1.3 ohms from 20Hz to 20kHz". It is very unusual for sources (and preamps) to have such low output impedance, which as far as I know, is related to higher output current. I think dCS agrees with you.

[Matias]

For reference, my Makua has 44 ohms output impedance on XLRs (same for the Tambaqui). The RME ADI-2 DAC has 200 ohms. The latest Topping D90SE has 100 ohms, 

[barrows]

25 minutes ago, Matias said:

For reference, my Makua has 44 ohms output impedance on XLRs (same for the Tambaqui). The RME ADI-2 DAC has 200 ohms. The latest Topping D90SE has 100 ohms, 

100 ohms is a very common output impedance for DACs.  This is usually due to a single 100 ohm resistor at the output, and is there to insure that the capacitance of any cable which might be connected to the DAC will not cause problems.  Of course the DAC designer does not know what crazy cable design might be connected, and therefore has to be conservative with that resistor value.  This is to keep the final output device stable.  Theoretically, one might be able to get away with reducing this resistor value if one uses short and low capacitance cables, the other way to reduce the value of the output resistor is to raise the drive capability of the output stage to make it stable driving any conceivable load.  Like what dCS does with their discrete output stages.

[audiobomber]

13 hours ago, mocenigo said:

Judging audio equipment in sighted way biases the experience so much as to be almost always worthless, in the sense that it provides no actual information. We can at most determine the characters of a piece of equipment, but even attempting to determine whether "like" the sound of component A over component B is disrupted by sight and the biases, often inconscious, that seeing stuff causes in our brain.

Typical blind A-B or ABX tests are worthless. Sighted A-B tests are worthless. Short term switching between options will not provide reliable results. I have confirmed this personally and through others' test results, for decades.

 

I listen to my main system for several hours daily, If I make a change, I will almost invariably hear a difference immediately. But I can only judge the final result after hearing it for a day, or even several days. A change that sounds impressive can show problems after extensive listening. Maybe it sounded more detailed, but fatiguing long term, I may hear a bass problem I hadn't noticed at first, or find I'm not as engaged in the music. And it's always best to reverse the change when possible. 

 

I know confirmation bias is real, but I'm confident that it doesn't affect my listening tests. Many times over the years I have preferred the cheaper or less impressive option, the opposite of what confirmation bias would indicate. 

 

The only way a blind listening test would be relevant to me is if I could do it over the long term, in my own system. That would be very effective, but It's just not practical.

[mocenigo]

49 minutes ago, audiobomber said:

Typical blind A-B or ABX tests are worthless. Sighted A-B tests are worthless. Short term switching between options will not provide reliable results. I have confirmed this personally and through others' test results, for decades.

 

If "Typical blind A-B or ABX tests are worthless" then all listening tests are worthless.

 

49 minutes ago, audiobomber said:

I listen to my main system for several hours daily, If I make a change, I will almost invariably hear a difference immediately. But I can only judge the final result after hearing it for a day, or even several days. A change that sounds impressive can show problems after extensive listening. Maybe it sounded more detailed, but fatiguing long term, I may hear a bass problem I hadn't noticed at first, or find I'm not as engaged in the music. And it's always best to reverse the change when possible. 

 

I know confirmation bias is real, but I'm confident that it doesn't affect my listening tests. Many times over the years I have preferred the cheaper or less impressive option, the opposite of what confirmation bias would indicate. 

 

Confirmation bias is a beast that lives in the subconscious, and it may lead us to prefer the cheaper option, because we fear to spend too much on the more expensive one.  Furthermore, listening to a piece of equipment long term comparing it in a fair way to other long term evaluations would require a nearly impossible discipline: in the diet, in order to control any blood pressure level change that will effect hearing, in the position, in the level of cleanliness of the ear canal, and listening to the same things, and probably a lot of note taking. Then it would work. Provided you can shut down your subconscious.  Not impossible, but, well if you really can make a subjective sighted long term listening test objectively, then you are really a monster (in a positive sense!). There may be other people like you in the human species, but I bet my hands suffice to count them. Again, I am not saying this is absolutely impossible.

 

 Roberto

 

49 minutes ago, audiobomber said:

The only way a blind listening test would be relevant to me is if I could do it over the long term, in my own system. That would be very effective, but It's just not practical.

 

[Iving]

13 minutes ago, mocenigo said:

If "Typical blind A-B or ABX tests are worthless" then all listening tests are worthless.

 

Confirmation bias is a beast that lives in the subconscious

 

Hi Roberto,

 

Forgive me butting in - I don't have much to say about Class D - just a brief remark on what you are saying here.

 

As I explained on ASR only recently (harking to ostensible "reverse" as an instance), Expectation Bias is a subjective phenomenon arising out of Cognitive Dissonance. The latter is psychological conflict which arises because of the reward anticipated after an Instrumental/Operant decision. Beliefs or understandings are shifted post hoc to fit the decision. The driver of all conditioning is the reward. Stereotypically we audiophiles think of cables and the like. But these mental processes happen in all domains of human existence. How much more beautiful a person becomes once we have decided upon marriage. The "reverse" type you envisage is perfectly plausible, but really would be an instance where the belief or understanding "no difference" was generated to justify a less obvious reward. Examples might be the money remaining in your bank account the result of not buying something. Or being right (an unfortunately pervasive human disposition) on an audio forum.

 

Now I can quite believe that Reward/Reinforcement hunger resides in "the subconscious" (whatever that may be - but we don't really need to get all Freudian) and, so, I can agree that "Confirmation bias is a beast that lives in the subconscious".

 

All I want to say here is that "Typical blind A-B or ABX tests" rely on conscious cognitive/mental processing. In fact, listening skills training - very deliberate focus on this and that/tells, and repeated practice - can result in better measurable performance in ABX tests. The point is - that if Biases emanate from the "subconscious", then listening tests which are sensitive to subconscious processing are necessary in order to reach your conclusion, If "Typical blind A-B or ABX tests are worthless" then all listening tests are worthless. What's more - there would have to be parity of results between conscious and subconscious-sensitive tests in order for your assertion to be valid.

 

Such subconscious tests don't exist as a matter of formality. But in theory they could. You cannot begin this kind of discussion on ASR. As @PYP hilariously pointed out recently, most ASR members seem to have IBS. ASR is a blue collar socio-cultural movement with a chief honcho who must be revered - and a mission to convert whom they regard as the foolish. ASR is much more "Scientism" than "Science". If it were a political movement we should fear for our lives.

 

Not looking for a lengthy debate or argument. Just food for thought. I don't have a model for subconscious-sensitive listening tests - nor the inclination to derive and publish any. Once maybe. Not now. Thank you

[charlesphoto]

The one thing ASR’ers types refuse to admit is that bias can work both ways. They are so adamant that nothing can make a difference unless measured, they wouldn’t admit it even if it was night and day, let alone the subtleties we’re usually referring to with cables etc. 

[botrytis]

9 minutes ago, charlesphoto said:

The one thing ASR’ers types refuse to admit is that bias can work both ways. They are so adamant that nothing can make a difference unless measured, they wouldn’t admit it even if it was night and day, let alone the subtleties we’re usually referring to with cables etc. 

 

Well, that IS the point and why double-blind studies take so long to setup properly and to do as to avoid this exact issue.

[PYP]

Has anyone listened to the T+A A 200 stereo amplifier based upon PURIFI Eigentakt module?

 

At about $5k, an investment, but T+A is known for excellent build quality and adding their own innovative technology.  Interesting that they provide a button to lower the damping factor, if desired.  

 

https://www.ta-hifi.de/en/audiosystems/series-200/a-200-power-amplifier/?doing_wp_cron=1687373168.4108328819274902343750

 

 

[Jud]

4 hours ago, mocenigo said:

If "Typical blind A-B or ABX tests are worthless" then all listening tests are worthless.

 

I don't think this is in accord with the scientific literature.

 

Typical blind A/B or A/B/X tests run into the huge problem that (as confirmed by numerous scientific experiments over decades) human echoic memory - that is, our memory for sounds - lasts about 4 seconds.  See for example https://deutsch.ucsd.edu/psychology/pages.php?i=209 .  As noted in that demonstration, the very worst situation for audio memory is to listen to more audio immediately after the passage you are trying to recall - in other words, exactly the typical blind A/B or A/B/X test.

 

Then layer on top of this the other problem I mentioned in this thread, the fact that we react subconsciously to negative stimuli long before becoming consciously aware of them.  (When setting up experiments, scientists account for the difference between a reaction to stimulus by some body system - for example, breaking out in a cold sweat, as in the Iowa Gambling Task experiment - and the conscious awareness that allows experimental subjects to respond verbally that they are pleased or unhappy.)

 

Then why aren't all listening tests worthless?

 

As I mentioned to you previously in this thread, pattern recognition is the key.  We are terrible at audio memory, but we are excellent at pattern recognition.  So how do we get from audio memory to pattern recognition?  Training.  This is the well-known "10,000 hours to become an expert."  I wouldn't for a moment claim objective correctness of my reaction to a blinded or sighted listening test of, for example, a digital filter.  But I think people who have worked with such filters nearly every day for years might be able to achieve that objective correctness in a listening test.

[fas42]

It's remarkably easy to hear whether any audio system is distorting. In the areas that matter, subjectively. Simply use a recording that you know from experience is 'difficult', and turn up to a decent volume; then move closer and closer to to a single speaker on one side, noting particularly the treble content. In any conventional setup, at some point in the distance from the drivers this will become obnoxious to experience; strident, harsh, thoroughly unlistenable, completely "unmusical" ... this is, very audible, playback distortion. Although many may not believe this, any half reasonable setup can be evolved to eliminate this distortion - allowing one to go as close to the speaker as you like, without that 'unpleasantness'.

 

Why worry about distortion, as opposed to "being different"? Because, if two system setups sound different, then at least one is distorting. What's on the source. Most likely, both are, distorting. In my book, you worry about eliminating distortion; not, whether "things sound different!".

[davide256]

So, if the input stage is that important, would this perform better than the OA GaN monos I had?

 

https://www.buckeyeamp.com/shop/p/purifi-1et400a-amplifier-2-channel

 

 

 

[barrows]

10 minutes ago, davide256 said:

So, if the input stage is that important, would this perform better than the OA GaN monos I had?

 

https://www.buckeyeamp.com/shop/p/purifi-1et400a-amplifier-2-channel

 

 

 

For commercial Purifi amp builds, my current favorite is the more powerful mono amps from Boxem.  They feature a more sophisticated input stage, actually designed by a very competent audio engineer.  The amp listed here looks like it has a very simple input stage with just a single OPA 1612 per channel.  The OPA 1612 has proven to be a bit less than ideal for driving the Purifi modules.

As far as comparisons to "OA GaN" amps, I have no idea, but so far, all of the GaN amps I have seen any measurements from performed rather poorly in comparison to Purifi or Ncore based amps.  so far GaN based amps have not proven themselves to have any advantage, despite using very expensive GaN FETs for the output stage.

[Matias]

Don't forget the premium Apollon amps based on Purifi 7040, with linear PS for the modulator and Weiss discrete opamps. It is a cost no compromise build.

 

https://apollonaudio.com/product/apollon-pet-950-purifi-1et7040sa-premium-monoblock-amplifier/

[PYP]

@barrows and @Matias, are your recommendations based upon listening as well as specs and design?  Just curious.  

[barrows]

5 minutes ago, PYP said:

@barrows and @Matias, are your recommendations based upon listening as well as specs and design?  Just curious.  

I have not heard the amplifiers in question.  What I have heard is that both Ncore based, and Purifi based amplifiers can sound quite different depending on the design/implementation of the input stage.  The most simple basic designs, such as those using a single IC opamp, with limited current capability have never been the best sounding solutions.  Some (most) commercial amps available with Purifi modules use a very basic design, with just a single IC opamp in the input stage, often they just copy the Purifi Eval module (which Bruno Putzeys has admitted is a basic design and not entirely optimal).  On the other hand, look at the Boxem amp with the more powerful Purifi modules, it uses a much more complex design, and clearly more thought and individual development went into that.  Of course, I guess it could sound bad, but then why would the designer have bothered with the more complex and involved design...  Having listened to some of the reasoning behind their approach, I tend to trust that the engineer at Boxem knows what he is doing.

[leManu]

This post on another forum describe what I experienced with the Nilai.  

 

https://forums.stevehoffman.tv/threads/what-do-audiophiles-mean-when-they-talk-about-pace-rhythm-timing.701989/page-3#post-17350478

 

Exert: "When I plugged that amp into my system, the sound it produced was absolutely incredible. It was actually bordering on ludicrous. The band sounded like they were all hammered -- sounded like each band member had at least 10 shots of tequila! The bass was totally off, couldn't catch up with the drums, the cymbals were off the beat so much that it wasn't even funny. Basically, the entire band was seriously off with regards to timing. It sounded as if each band member was playing in a separate, isolated room.
Plugging my old amplifier back, everything returned to normal. So I returned that class D amp for a refund, never to attempt the same experiment again."

 

I don't know the technicalities about why this happens, but I sure know that I heard it loud and clear.  I know that the concept of PRaT has been laughed at by many, but I really don't mind using it now as I think it describe the coherence of the music.  And for me that's 10 times more important than distortion or other measurable aspects.

[mocenigo]

On 6/21/2023 at 4:45 PM, audiobomber said:

Typical blind A-B or ABX tests are worthless. Sighted A-B tests are worthless. Short term switching between options will not provide reliable results. I have confirmed this personally and through others' test results, for decades.

 

"Typical" is here the key.

Subjectivists claim that there are aspects that one must consider very carefully. For instance, I have a recording of Nielsen's 5th symphony (Sir Colin Davis, LSO, LSO's own label) where there is a descending clarinet introducing a movement. On DACs that are otherwise indistinguishable, the clapping of the keys has a different perceptibility, as if in some DACs that noise were louder (examples: more perceptible on a Topping D90, or on a Soekris DAC 1541, less on a Gustard X20 or or TEAC UD-501). So one should have a "gallery" of critical passages and listen to them in an ABX test. I posit that most passages sound exactly the same with 99% of the DACs commercially available now.

 

In other words, I support blind ABX testing, but not just with anything as the source.

 

I hope my though is clearer now (and I apologise for having replies only after two weeks, but I was extremely busy, work-wise).

 

 

On 6/21/2023 at 4:45 PM, audiobomber said:

I listen to my main system for several hours daily, If I make a change, I will almost invariably hear a difference immediately.

 

I guess you first balance the volume top to 0.1Db and then listen to specific passages. Then it is plausible.

 

 

 

On 6/21/2023 at 4:45 PM, audiobomber said:

But I can only judge the final result after hearing it for a day, or even several days. A change that sounds impressive can show problems after extensive listening. Maybe it sounded more detailed, but fatiguing long term, I may hear a bass problem I hadn't noticed at first, or find I'm not as engaged in the music. And it's always best to reverse the change when possible. 

 

I know confirmation bias is real, but I'm confident that it doesn't affect my listening tests. Many times over the years I have preferred the cheaper or less impressive option, the opposite of what confirmation bias would indicate. 

 

Well, I am sorry to have too spoil your opinion on this, but confirmation bias (whether positive or negative) is impossible to defeat, or let us say nearly impossible.

 

 best

  Roberto

 

On 6/21/2023 at 4:45 PM, audiobomber said:

 

The only way a blind listening test would be relevant to me is if I could do it over the long term, in my own system. That would be very effective, but It's just not practical.

 

[mocenigo]

On 6/21/2023 at 6:52 PM, charlesphoto said:

The one thing ASR’ers types refuse to admit is that bias can work both ways. They are so adamant that nothing can make a difference unless measured, they wouldn’t admit it even if it was night and day, let alone the subtleties we’re usually referring to with cables etc. 

 

Oh, of course it can work both ways. If one is absolutely persuaded that there cannot be any difference, they may be not hearing one even if it should be audible. And I am sure the most informed "ASR types" will agree on this. The point is that if a difference can be detected and in a reproducible way, then it exists. These tests can only prove the something exists (but they may still fail to), but cannot prove that something does not exist, and this is obvious.

[mocenigo]

On 6/27/2023 at 6:35 AM, leManu said:

This post on another forum describe what I experienced with the Nilai.  

 

https://forums.stevehoffman.tv/threads/what-do-audiophiles-mean-when-they-talk-about-pace-rhythm-timing.701989/page-3#post-17350478

 

Exert: "When I plugged that amp into my system, the sound it produced was absolutely incredible. It was actually bordering on ludicrous. The band sounded like they were all hammered -- sounded like each band member had at least 10 shots of tequila!  Plugging my old amplifier back, everything returned to normal. So I returned that class D amp for a refund, never to attempt the same experiment again."

 

I don't know the technicalities about why this happens, but I sure know that I heard it loud and clear.  I know that the concept of PRaT has been laughed at by many, but I really don't mind using it now as I think it describe the coherence of the music.  And for me that's 10 times more important than distortion or other measurable aspects.

 

I really wonder whether the listener was just used to more background noise. It is absolutely impossible that a modern, well designed amplifier may alter the signal such that "The bass was totally off, couldn't catch up with the drums, the cymbals were off the beat so much that it wasn't even funny. Basically, the entire band was seriously off with regards to timing. It sounded as if each band member was playing in a separate, isolated room." Something like that would require extremely difficult and, most importantly, intentional signal processing to be achieved.

 

Now, remember that people watching a movie at 48fps (I think it was The Hobbit) found it "wrong" because they were used to 24fps. Of course at 48fps a movie is more realistic, but if we are uses to the minute jerking motions at 24fps, too much realism may even disturb us. We may perceive the movements of the various objects on screen as more delineated, better defined, and their increased motional independence may even confuse us.

 

A very transparent amplifier will make you hear the various instruments (on a good recording) better separated, and one may perceive that as increased independence. You can better focus your attention on one instrument, then on another, and maybe perceive them as going each their own way. But what you are actually hearing is better resolution. The players are still playing in time as with the previous amp.

 

 

 

 

[Jud]

1 hour ago, mocenigo said:

 

In other words, I support blind ABX testing, but not just with anything as the source.

The problem, scientifically, is that no matter the source, the listeners really don’t have the ability to do blind ABX tests for anything that isn’t readily apparent in about 4 seconds, unless they are trained or the “right” answer is suggested.

 

I did an informal A/B test here once with a few dozen participants, regarding whether people could hear a difference between two quite different acoustic guitars given two second vs. 30 second samples (one a classic Martin, the other a 1930s Epiphone with a much smaller sound chamber and a couple of f holes). Those hearing the 30 second sample were able to discriminate better, with a p value of .06.  (In other words, one would have expected the same result by chance only 6% of the time.)

 

As you’ve said, one would hardly expect gross effects given the current state of electronics, so what’s left are relative subtleties. But some of these subtleties may be important to musical enjoyment. I doubt, based on the scientific work of the past decades, that ABX testing is a very good tool for teasing out any differences in equipment that may still exist.

[audiobomber]

3 hours ago, mocenigo said:

It is absolutely impossible that a modern, well designed amplifier may alter the signal such that "The bass was totally off, couldn't catch up with the drums, the cymbals were off the beat so much that it wasn't even funny. Basically, the entire band was seriously off with regards to timing. It sounded as if each band member was playing in a separate, isolated room." Something like that would require extremely difficult and, most importantly, intentional signal processing to be achieved.

That is called PRaT (Pace, Rhythm and Timing), and it's exactly why I was so disappointed with the Topping D50S. The band is playing, but the musicality and emotional involvement is AWOL. 

[Iving]

2 hours ago, Jud said:

The problem, scientifically, is that no matter the source, the listeners really don’t have the ability to do blind ABX tests for anything that isn’t readily apparent in about 4 seconds, unless they are trained or the “right” answer is suggested.

 

I did an informal A/B test here once with a few dozen participants, regarding whether people could hear a difference between two quite different acoustic guitars given two second vs. 30 second samples (one a classic Martin, the other a 1930s Epiphone with a much smaller sound chamber and a couple of f holes). Those hearing the 30 second sample were able to discriminate better, with a p value of .06.  (In other words, one would have expected the same result by chance only 6% of the time.)

 

As you’ve said, one would hardly expect gross effects given the current state of electronics, so what’s left are relative subtleties. But some of these subtleties may be important to musical enjoyment. I doubt, based on the scientific work of the past decades, that ABX testing is a very good tool for teasing out any differences in equipment that may still exist.

 

Hi Jud,

 

I spent a goodly while ... keen to understand this post as a coherent whole.

 

1. I get the point about echoic memory <4 secs. If empirically true, then blind test performance will be reduced for stimuli longer than 4 secs - as "interference" from the part of the stimulus >4 secs. will exacerbate what we might already expect from deleterious "forgetting" what is in excess of 4 secs.. Would you agree with this. In other words, keeping stimuli to <4 secs. is something we could do to enhance sensitivity in tests.

 

2. I tried and failed to get my head round the Design of your (GW/DR) guitar test. After making notes to myself I found your original 2015 thread. Looked like a great deal of social fun, but may I say that I agree with whomsoever intimated that the discrimination was easy. Not really a blind test as such. "Scientifically" may I say that the Design had no hope of isolating any independent variable such that you could say today "Those hearing the 30 second sample were able to discriminate better, with a p value of .06.  In any event, I find it impossible to map such an assertion to what you are saying about 4 sec. echoic memory. No doubt I lack enough detail or haven't drunk enough coffee.

 

3. Regards "electronics" (cf. speakers! and etc) I agree very much that "what’s left are relative subtleties. But some of these subtleties may be important to musical enjoyment". I don't feel strongly about you saying, "I doubt, based on the scientific work of the past decades, that ABX testing is a very good tool for teasing out any differences in equipment that may still exist." [something manufacturers could pursue, but don't pursue, and for understandable reasons, both good and bad] but imo I do think that testing could yet help us reconcile ourselves (as enthusiasts for enjoyment of music which we have in common) when it comes to the Subjectivist-Objectivist divide. Probably - there is only one "truth" - and we fail to find the common ground germane. I have had ideas in the past - some only vaguely formed - and began to formulate more (including ref. to "subconscious") provoked in the nicest way by @mocenigo. Trouble is - it takes a lot of time, and so on, and also goodwill amongst us to make mutual progress. That is what I find saddest about Forum participation.

 

Fortunately there is much joy to be had too! Wish I'd been around for your 2015 thread. I have been a member here since 2015. Perhaps - being a humble newbie - I ignored it!

[PYP]

6 hours ago, mocenigo said:

 

"Typical" is here the key.  Subjectivists claim that there are aspects that one must consider very carefully. For instance, I have a recording of Nielsen's 5th symphony (Sir Colin Davis, LSO, LSO's own label) where there is a descending clarinet introducing a movement. On DACs that are otherwise indistinguishable, the clapping of the keys has a different perceptibility...

best

  Roberto

I understand the intellectual interest in ABX as it pertains to our ability to differentiate sounds, but I truly don't understand how this applies to the enjoyment of listening to music at home or the task/hobby of assembling a home system for playback.  

 

Wouldn't any "test" also apply to listening to live music?  I have never been to a classical concert and heard someone whisper:  did you hear the clear clapping of the keys just then? 

[PYP]

6 hours ago, mocenigo said:

 

Oh, of course it can work both ways. If one is absolutely persuaded that there cannot be any difference, they may be not hearing one even if it should be audible. And I am sure the most informed "ASR types" will agree on this. The point is that if a difference can be detected and in a reproducible way, then it exists. These tests can only prove the something exists (but they may still fail to), but cannot prove that something does not exist, and this is obvious.

After extensive research, I was able to find this rare footage of ASR true believers.  Of course, they would be even more passionate talking about their hobby!