Hi Pavel,
The fruit test is quite revealing for me. Thank you.
My system is not high end by any means, however I do have a pair of old 5W Sony Alnico 4.5" Full range. I tried several DHT preamps with ~0.05% THD on it and found a notable difference in the 3D spatial illusion. Nelson Pass mentioned about the efect of the phase of second harmonic on the perception of spatial information of sonic presentation. In case you are also interested in phase related audibility of distortion profile, I mentioned in post #264 that inversion of signal prior to the tube stage should produce a different effect. I think 2H phase related effect would be more difficult to perceive on a headphone or on a multiway that reproduce a signal on one driver and its second harmonic on another, especially around 1kHz where human spatial perception is most sensitive to phase.
That might be a kind of point, however I assume it is quite difficult to distinguish between influence on sound of the "pure" nonlinear distortion and other possible issues in the amplifier, like level-only related instability, hysteresis, cross-over distortion artifacts, frequency-dependent non-linearity etc. The distortion is often just an indicator of another issue and the sources of distortion may be very complex. The distortion would not be then defined as a polynomial distortion a
*X^n + a(n-1)*X^(n-1) + ..... + a(0), and would not be easy to simulate, especially in case of frequency dependent distortion. Maybe in LTSpice, which takes ages on a music sample.
My goal was to exclude frequency dependent distortion and to exclude cross-over and hysteresis like distortions and to approach to a simple polynomial non-linearity. That's why open-loop tube gain stage, that's why class A output stage with a negligible distortion and fairly low output impedance. Of course the more complex distortion would result in not only 90° multiples of phase of the distortion components.
*X^n + a(n-1)*X^(n-1) + ..... + a(0), and would not be easy to simulate, especially in case of frequency dependent distortion. Maybe in LTSpice, which takes ages on a music sample.My goal was to exclude frequency dependent distortion and to exclude cross-over and hysteresis like distortions and to approach to a simple polynomial non-linearity. That's why open-loop tube gain stage, that's why class A output stage with a negligible distortion and fairly low output impedance. Of course the more complex distortion would result in not only 90° multiples of phase of the distortion components.
Last edited:
Hi. Can you point me towards that? I can imagine it creating a pseudo random delayed reflection effect?
Ok, let me try this another way as it is a difficult area to get your head around - as Jakob posted some quotes from a well known authors & book in this area - "Perceptual Audio Evaluation–Theory, Method and Application"
I've no objection to blind tests, I use them all the time - they are listening from another angle, just as listening to different music formats is another angle as is sighted listening. We all draw our personal conclusions from these types of listening but that's what they are - personal related to our playback system, our experience & our perception.
Now when multiple blind listening results are reported, I have no problem with that - just as I have no problem with multiple sighted listening impressions being reported. I will read sighted listening reports & evaluate them based on the amount of information & prior knowledge of the person reporting. If detail reporting is given of exactly what was heard - music, specific sounds in the music that are notable, etc, this gives me some specifics to test for myself. But the sighted listening impressions are anecdotal & either interesting enough for me to check myself or not i.e. I don't conclude anything from them unless I've done my own listening.
When people extrapolate their SIGHTED listening personal impressions into a generalization for all then there usually is big criticism of this over-reaching.
Now I don't see the same objections happening with blind listening impressions when people extrapolate their BLIND listening personal impressions into a generalization.
Blind listening is assumed to be of a higher quality (closer to what's true) than sighted listening - the reason given is that a bias is removed. When it comes to forum blind tests the most often suggested (sometimes demanded) blind test is an ABX test. In fact I've seen people reject the preference test protocol as not relevant until a 'difference' test is first performed. This attitude has elevated ABX 'test' to a blind unquestioning acceptance of the results. The only time I've seen results being examined are when the null hypothesis is NOT supported i.e. a difference has been shown to be audible.
What I & Jakob are trying to do is introduce people to the difficulties involved in perceptual testing, particularly the pitfalls of ABX testing which is the most requested or demanded blind test protocol when 'proof' of listening results are called for
What I would like to see is a more balanced & critical approach to such blind tests but I understand that it is difficult to overcome years of a particular message that "ABX is truth"
Both Jakob & I have presented links to research to back up what we post - it's not just our biased opinion, it's recognised that blind testing is difficult. We are not talking about tests that are of publishable quality, just some basic checks on such testing, in order to qualify the quality of the results & not blindly accept the null hypothesis.
From that book, in case people missed it before:
I've no objection to blind tests, I use them all the time - they are listening from another angle, just as listening to different music formats is another angle as is sighted listening. We all draw our personal conclusions from these types of listening but that's what they are - personal related to our playback system, our experience & our perception.
Now when multiple blind listening results are reported, I have no problem with that - just as I have no problem with multiple sighted listening impressions being reported. I will read sighted listening reports & evaluate them based on the amount of information & prior knowledge of the person reporting. If detail reporting is given of exactly what was heard - music, specific sounds in the music that are notable, etc, this gives me some specifics to test for myself. But the sighted listening impressions are anecdotal & either interesting enough for me to check myself or not i.e. I don't conclude anything from them unless I've done my own listening.
When people extrapolate their SIGHTED listening personal impressions into a generalization for all then there usually is big criticism of this over-reaching.
Now I don't see the same objections happening with blind listening impressions when people extrapolate their BLIND listening personal impressions into a generalization.
Blind listening is assumed to be of a higher quality (closer to what's true) than sighted listening - the reason given is that a bias is removed. When it comes to forum blind tests the most often suggested (sometimes demanded) blind test is an ABX test. In fact I've seen people reject the preference test protocol as not relevant until a 'difference' test is first performed. This attitude has elevated ABX 'test' to a blind
unquestioning acceptance of the results. The only time I've seen results being examined are when the null hypothesis is NOT supported i.e. a difference has been shown to be audible.What I & Jakob are trying to do is introduce people to the difficulties involved in perceptual testing, particularly the pitfalls of ABX testing which is the most requested or demanded blind test protocol when 'proof' of listening results are called for
What I would like to see is a more balanced & critical approach to such blind tests but I understand that it is difficult to overcome years of a particular message that "ABX is truth"
Both Jakob & I have presented links to research to back up what we post - it's not just our biased opinion, it's recognised that blind testing is difficult. We are not talking about tests that are of publishable quality, just some basic checks on such testing, in order to qualify the quality of the results & not blindly accept the null hypothesis.
From that book, in case people missed it before:
BTW, the most interesting part of this was the subsequent testing of the distorted sine wav signals - the results of blind testing this signal has been the real learning that is happening here, I believe
As this sine wav testing is a form of post-test evaluation, I don't understand why people would have an objection to introducing some control like this in the blind test itself?
As this sine wav testing is a form of post-test evaluation, I don't understand why people would have an objection to introducing some control like this in the blind test itself?
This possibility is clearly stated in the thread title. It says "Can you tell . . .". It doesn't say 'can anyone tell'; it doesn't say 'can someone else tell'. So the test was can some individual using his own equipment tell. The result appears to be 'No' for most people using their normal equipment. You keep complaining that the test is not what you would like it to be (so what?). You keep complaining that we have misunderstood the result (we have not). You keep complaining that the test did something different from what it said it would do (it did not). Sorry, the universe is as it is and in this universe Pavel ran a test which gave a particular result. You need to accept this and move on.mmerrill99 said:
Like others here, I am tiring of this seemingly endless and fruitless argument between a bunch of DIYers on one side, and the (suspected) salesman and the pedant on the other side. Like all such threads, we are now going round in circles.
If what you say is correct then why would Pavel state after the sine wav test
And then:
So again, this shows that Pavel assumed users & their systems were capable enough to participate in the test & later discovered to the contrary & he regards their results as not meaningful.
You're tilting at windmills here & denying what the test originator intended & posted through his journey on this thread
Yes, so please stop denying what is obvious for all to see & I won;t have to correct your mis-information.
I've stated my position, Pavel & others have stated theirs - it would be good to now move on
Thanks, I'm interested in speculation as to why
A good question, ScottJ - I'll ignore your attempts at being insulting, yet again
Anybody got links to how this "H2 generator" is perceived in listening?. I did a quick search but haven't found listening impressions on this specific "2nd harmonic generator" but I believe it is based on the same idea implemented using a Nuvistor tube, the 8056 but the H2 generator is using jfets. However, I'm not sure this is a generator of anti-phase H2 which Nelson seems to emphasize is the 'secret sauce'
Here's where I picked up some detailed listening impressions of this Katz's Corner Episode 25: Adventures in Distortion | InnerFidelity
Admittedly it's just one guy's listening impressions but the guy is the well known Bob Katz so I respect his impressions.
Last edited:
Any link to the listening tests?
indra1 said:
Some questions, please? - NP seems to put the effect down to anti-phase H2 but I'm not sure your DHT is anti-phase H2 nor is Bob Katz' Nuvistor
So my very speculative speculation
is that perhaps it is just an effect of low signal level H2, not necessarily anti-phase?How would low level H2 distortion change our perception of the sonic presentation in the ways described by both you & Bob Katz - he states "the 3 dimensional space has greatly expanded" "all the detail and clarity that anyone would desire" I'm not sure I agree with his conclusions, however "My conclusion is that since distortion can only be added, not subtracted in this chain, the psychoacoustic mechanism is distortion masking. I believe that this approach to mask harsh distortion by using euphonic distortion works, very well."
Let me explain why. There is a little known phenomena in psychoacoustics call "comodulated masking release " CMR, for short. It is little understood being a phenomena that can be observed but the exact mechanism can't be fully explained yet. The phenomena occurs in one example when a separate signal amplitude modulates in correlation with a signal that is being masked. What I think might be happening is that low level detail is being masked by some distortion & by introducing a low level H2 which is modulating in correlation to the signal, it is unmasking this low level detail so that it is now audibly perceived.
You can try a live demo of one example of CMR here (there are many different phenomena which seem to be explained by CMR - this demo is just one usuch example)
Comodulation Masking Release | Auditory Neuroscience
As I said, it's my speculation but worth thinking about
+1
I quoted Nelson Pass' H2 article, ask him.
A triode can produce negative or positive phase H2 depending on operating point, mine was operated negative phase.
Last edited:
OK, yours is anti-phase then - I'm not arguing with you just trying to tease out what I think is an interesting point you raised. I was wondering if there are any reports from the freebie H2 generator given out by NP at RMAF?
Maybe Bob Katz Nuvistor H2 generator is actually procusing anti-phase H2, I don't think it's mentioned in his article? His listening impressions are very similar to what NP states & you seem to concur with?
BTW, CMR could also work with anti-phase version of H2 harmonic
Basically what Pavel has done here is to ask "Can you hear this piece of gear when placed into your signal chain?" Some can, some can't.
The Foobar ABX may have its limitations, but it is used here on the forum for its great merits. It's easy to verify that the test was run correctly and the results are as reported. If you have another test that is better, and is easy to verify, please start a thread and perhaps ask for participants.
Clearly you just don't like the ABX test. The ABX has been repeatedly criticized in many threads over the years. It's old hat. But when the critics are asked to use some other type of blind test, they never will. Many agree to do it, but I've never seen one completed. If anyone has seen it on this forum, please point me to it.
The Foobar ABX may have its limitations, but it is used here on the forum for its great merits. It's easy to verify that the test was run correctly and the results are as reported. If you have another test that is better, and is easy to verify, please start a thread and perhaps ask for participants.
In fact the complete audio chain (electrical path) vs. original CD rip. I think this is quite interesting, because the data were recorded from the amplifier speaker terminals and the amplifier was loaded by my speakers. So it is an original vs. complete electrical audio chain + ADC. Isn't it interesting?
I simply suggested an improvement to the ABX test - the use of controls - again you don't seem to understand what is being said or choose to ignore it.
The post-test sine wav test was the most interesting part of this exercise, I believe - it acted as a crude control
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.