Chris,
If I understood Bob's earlier posts correctly, he showed that with a forward path with no PIM, and a linear feedback path, you can generate PIM.
Is that correct? Bob?
jd
Demian,
Actually, the artifacts were explained in the paper, its that they just were not labeled in the graph. There's a lot of sense in your post but I don't think that trying to focus on several issues at the same time instead of step-by-step getting to grips with the one that seems stubborn right now is a good career move.
BTW Funny that you mention non-harmonic tuning. When I visited Jan Lohstroh last week, he explained that concept to me (he has a beautifully restored 1942 Steinway model C, and I had no knowledge of these things). It does of course NOT mean that any non-harmonic frequencies are generated by IM or from a single source. It does mean that two snares that should generate harmonic frequencies are de-tuned so that the two tones are slightly non-harmonic. The way I understood it (I haven't looked at the wikipedia entry) this is necessary to fit the twelfth-root-of-2 steps into an octave.
jd
Last edited:
Demian,
Mind you, it's not the same issue. Creating a non-harmonic signal is not the same with processing a harmonic signal.
Inharmonic oscillations in mechanical resonators (as strings or, as a matter of fact, quartz crystals) is a well understood phenomena. As you invoked math, the reason for this inharmonic behaviour is the violation of another essential condition in the Fourier theorem: the signal periodicity. And physically, mechanical resonators are generating aperiodic signals because of the nonlinearities in their mechanical properties.
You would agree that generating an aperiodic signal is different from transforming a continuous and differentiable periodic signal in an aperiodic signal (as JC's extraordinary claim implies). Such a transformation would ultimately violate the causality principle.
Demian, it is true that perfect tuning doesn't really exist, but humans have had thousands of years to perfect what works pretty well.
The real point is: Lower order harmonics, i.e. 2'nd, 3'rd, and 4th, at the very least, can be tolerated in very high amounts or else both vinyl records and analag magnetic tape would have been found unacceptable, from the very beginning. This also includes direct radiator loudspeakers, as well.
Many designers have wondered about this for multiple decades, and studies going back to the 1930's. at minimum, have given tolerance levels of distortion which can be very high, with lower order harmonic/IM products.
However, the dilemma exists that negative feedback, especially with op amp type designs, can lower MEASURABLE harmonic/IM distortion to almost nothing at all, perhaps 1 part in 1 million, YET the sound quality of the electronics can still sound compromised, even almost annoying.
The reasons why this is so, is much of my quest for a better approach to making audio electronics. I say AUDIO electronics and not test equipment, etc electronics, because it appears to be something that the human ear detects, that most engineers and scientists don't yet completely know about, that is more important than very low typical measured distortion. This is what people have been working at for decades, either by making sonically acceptable products, just by brute force, or in trying to create a test that is outside the boundary of normal test methods, that will give a visual indication of what we hear. This is the point of this discussion.
The real point is: Lower order harmonics, i.e. 2'nd, 3'rd, and 4th, at the very least, can be tolerated in very high amounts or else both vinyl records and analag magnetic tape would have been found unacceptable, from the very beginning. This also includes direct radiator loudspeakers, as well.
Many designers have wondered about this for multiple decades, and studies going back to the 1930's. at minimum, have given tolerance levels of distortion which can be very high, with lower order harmonic/IM products.
However, the dilemma exists that negative feedback, especially with op amp type designs, can lower MEASURABLE harmonic/IM distortion to almost nothing at all, perhaps 1 part in 1 million, YET the sound quality of the electronics can still sound compromised, even almost annoying.
The reasons why this is so, is much of my quest for a better approach to making audio electronics. I say AUDIO electronics and not test equipment, etc electronics, because it appears to be something that the human ear detects, that most engineers and scientists don't yet completely know about, that is more important than very low typical measured distortion. This is what people have been working at for decades, either by making sonically acceptable products, just by brute force, or in trying to create a test that is outside the boundary of normal test methods, that will give a visual indication of what we hear. This is the point of this discussion.
The ultimate tactic: when everything else fails, move the discussion in the muddy waters of subjective experiences.
Now waiting for the GEB to jump in and "clarify" the situation.
I do not think there is any tactic. It is the same experience of many of us - circuits that measure extremely low THD and IMD, have wide CL bandwidth and reasonably high slew rate still sound very different. It is up to you, syn 08, to find an appropriate method that would explain anything. I take it as a fact, experiment is above model to me.
Not really. Its up to whoever makes the claim to prove that the claim holds water.
This is the claim:
"circuits that measure extremely low THD and IMD, have wide CL bandwidth and reasonably high slew rate still sound very different" due to the use of neg feedback (or due to the use of ICs; take your pick).
To be proven:
a) this is a fact;
b) if it is a fact, it is due to neg feedback (or due to the use of ICs; take your pick).
After that, we will worry about why it could be so.
jd
is this back to front?
Don't experiments often lead to unexpected results and lead to further experiments to try to find what is happening?
Then the experimenter starts to formulate a hypothesis to explain the phenomena.
The theory and the proof often follow afterwards.
Well, sure, if you have unexpected results you do further experiments to make sure that what you unexpectedly found, is in fact correct. That is the 'proof' that your initial unexpected result is in fact repeatable and reliable.
What we have so far is a lot of anecdotes, uncontrolled, unrepeatable in a controlled way. So naturally, it wouldn't make a lot of sense to start to speculate how it happened if we're not even sure it did happen.
jd
That depends on the fact; if a research group would experiment and claim that a khaliba root concoct cures liver cancer, that I am sure scientists will jump in to reproduce the experiments according to the latest medical and pharmaceutical protocols (including blind tests) and, if confirmed, then dig for the ultimate scientific explanation.
However, if I am telling you that a flying saucer landed in my backyard yesterday and the LGM asked for my 8702A network analyzer to fix their hyperspace engine, I'm pretty sure you won't jump in the first plane to watch them.
There is another example: there are well known facts proven by psychotherapists and their patients that can't be tested blindly, but great improvements happened despite all scientifically proved tests denied such a possibility. Still, some insurance companies cover psychotherapy in cancer healing.
It is about human perception and what happens when it is taken in consideration.
Similarly, when I design and build audio electronics that don't exhibit known to me errors in principle that add unnatural sound alterations that inevitably present in common topologies the electronics as if don't exist between the source of a music and a listener.
But with substraction tests (Baxandall, Hafler..) with real music material, we hear (with same circuits) nothing. Very strange.
No. Only after application of HP/LP filters that equalize the tested channels (and add their own complex transfer functions). Without that, there are great differences audible, and we can argue if to separate linear and non-linear issues, though the both IMO must be considered.
Well, here we are. Ignoring the sonic differences and demanding the measurements. As audio designers, such as myself, have noted, some amp and preamp designs sound better than others, yet our measurements do not show why.
Now, the difference between many here, and me, seems to be that I actually have listened for the past 40 years to people who know more and have more experience in audio design. Yes, once I came to know certain people, I seriously listened to their 'subjective' experiences as well as to their help in difficult concepts, such as 'delay' group, phase, or real, and other areas where I did not have a complete understanding.
In fact, it was almost 40 years ago, when Richard Heyser (me, name dropping again) told me about his experience with his no feedback power amp that he had designed to go to the Moon, or some such, and he had taken it home to listen to it on his Klipschorn. He was accidently surprised that it sounded better than his existing amp. He told me that it was the amount of NEGATIVE FEEDBACK that was the difference. Wow! It hit me, that I used 40db feedback in my amp design, and the Radiocraftsman (co-designed by Dick S, Mitch C, and Sid Smith in the early '50's) only used 20 dB of feedback and probably had a higher open loop bandwidth, to boot. Gee, I didn't need PROOF that this might well be the case. I took it for a direction to explore.
Why I have to PROVE everything to a bunch of skeptics, who are so insulting as to call me 'intellectually dishonest' or not forthright in my explanations of how to design an amp, is beyond me.
Now, the difference between many here, and me, seems to be that I actually have listened for the past 40 years to people who know more and have more experience in audio design. Yes, once I came to know certain people, I seriously listened to their 'subjective' experiences as well as to their help in difficult concepts, such as 'delay' group, phase, or real, and other areas where I did not have a complete understanding.
In fact, it was almost 40 years ago, when Richard Heyser (me, name dropping again) told me about his experience with his no feedback power amp that he had designed to go to the Moon, or some such, and he had taken it home to listen to it on his Klipschorn. He was accidently surprised that it sounded better than his existing amp. He told me that it was the amount of NEGATIVE FEEDBACK that was the difference. Wow! It hit me, that I used 40db feedback in my amp design, and the Radiocraftsman (co-designed by Dick S, Mitch C, and Sid Smith in the early '50's) only used 20 dB of feedback and probably had a higher open loop bandwidth, to boot. Gee, I didn't need PROOF that this might well be the case. I took it for a direction to explore.
Why I have to PROVE everything to a bunch of skeptics, who are so insulting as to call me 'intellectually dishonest' or not forthright in my explanations of how to design an amp, is beyond me.
Last edited:
PMA, I hope that you keep on track with your efforts.
Demian has completed a test board that we might use together to do some additional measurements, such as the original TIM test. Demian and I, both, have the test equipment to do the job. I am sidetracked until I can hire Demian's tech. part time, to move forward.
Demian has completed a test board that we might use together to do some additional measurements, such as the original TIM test. Demian and I, both, have the test equipment to do the job. I am sidetracked until I can hire Demian's tech. part time, to move forward.
We had practical results, that with simple linear RC filters we can in substraction test practicaly cancel "distortion" produced by "enough" linear (acording ordinary test) circuit. Resulting products are at levels obtained with obvious THD and IMD tests. If You consider noninfinity bandwidth as distortion, than yes, in direct substraction from non limited signal it is hearable. But can you hear this "distortion" after passing signal thrue simple bandpass RC filters with corner frequency e.g 5Hz and 200kHz??
- Status
- Not open for further replies.