In one of the earlier references a test was documented showing that uncorrelated tones were inaudible below a certain (relatively high) level. I have seen other references suggesting that certain im products were audible at levels way lower. I don't think we have adequate knowledge of what distortions are audible to define standards. At least those that are accepted in other arenas (e.g. car audio) may not be appropriate here. Further the test equipment was built that way because it matches the test needs for a linear class A device. The fact that its not too sensitive to crossover distortion that many find objectionable most indicates that its not the best way to test for crossover distortion. Some analyzers reject harmonics above the 5th and its plausible that those harmonics are the important ones. And possibly there is some other parameter, independent but with a correlation to feedback that low feedback, or no feedback, optimizes for listening.
Hello Damien
Could you describe what your reference monitoring speaker is, that is the one that you use for evaluation and a little bit about the room they are in.
Regards
Arthur
With regards to distortion and feedback, we are about where we left off about 30 years ago, in this discussion. We know that negative feedback will reduce MEASURABLE distortion, and direct coupling allows a fair amount of negative feedback to be used. Marketing forces and convenience have prodded us to reduce heat dissipation, and this usually means more WEIGHTED open loop distortion that is then hidden from measurement by feedback. Is this the right direction? From an engineer's point-of-view, it is. From a listeners point of view, it seems a step backward.
As Tico says: IF you listen to a selected direct disc recording where everything in the recording chain is carefully chosen, perhaps with NO IC op amps, quality wire, good microphones, and a good disc cutter, as you might find with a Sheffield or Crystal Clear recording of the '70's, you will find a glimpse of sonic nirvana. No digital, no analog magnetic tape, just an amplified microphone connected to a disc cutter, made of discrete tube or solid state amps. This is like having a meal at Chez Panisse. It was once fairly easy, if expensive, now almost impossible to get that essential sound quality.
Why this is true, I will go into shortly.
As Tico says: IF you listen to a selected direct disc recording where everything in the recording chain is carefully chosen, perhaps with NO IC op amps, quality wire, good microphones, and a good disc cutter, as you might find with a Sheffield or Crystal Clear recording of the '70's, you will find a glimpse of sonic nirvana. No digital, no analog magnetic tape, just an amplified microphone connected to a disc cutter, made of discrete tube or solid state amps. This is like having a meal at Chez Panisse. It was once fairly easy, if expensive, now almost impossible to get that essential sound quality.
Why this is true, I will go into shortly.
Last edited:
Yes I was thinking of the Crystal Clear LP where you designed the console electronics for.....
Arthur Fiedler and the Boston Pops CCS-7003
In one word: awesome
http://cgi.ebay.com.au/Arthur-FIEDL...dZViewItemQQptZAU_Records?hash=item35a7987dad
Get it!
Last edited:
Attempting to sort out, 'what works' vs 'what should work' we find a few points. First, the actual amount of distortion, below a few percent, IF it is low order, 2, 3rd, 4th, mostly, in fact almost exclusively, is relatively OK, and even if it changes the original character of the live performance slightly, it does not significantly modify the listening experience.
This is why phono cartridges work pretty well, analog tape recording was OK, and even open loop single ended tube amps appear to sound OK, even when the actual measured distortion is significant.
On the other hand, many audio designs from the late '60's till today, that measure very very good, sometimes less that .001% at typical listening levels, can sound strained, fatiguing, lifeless, etc. Why this is so, appears on how and why we use negative feedback. Do we use it to make 'good specs' from just OK, designs? Is the amount of feedback flat over the entire listening range? Are there hidden distortions generated, that our normal test equipment does not directly measure, in fact, tends to ignore?
This is why phono cartridges work pretty well, analog tape recording was OK, and even open loop single ended tube amps appear to sound OK, even when the actual measured distortion is significant.
On the other hand, many audio designs from the late '60's till today, that measure very very good, sometimes less that .001% at typical listening levels, can sound strained, fatiguing, lifeless, etc. Why this is so, appears on how and why we use negative feedback. Do we use it to make 'good specs' from just OK, designs? Is the amount of feedback flat over the entire listening range? Are there hidden distortions generated, that our normal test equipment does not directly measure, in fact, tends to ignore?
Back, almost 40 years ago, Bascom King, (an audio designer, even today, and formerly a reviewer for 'Audio' magazine), another designer, Geoff Cook, and I made a pilgrimage to see Richard Heyser, and ask him WHY, good specs sometimes sounded lousy, and why we could hear differences between audio equipment with vastly better measurements than the phono cartridges or the loudspeakers we were using. While he, like the rest of us, even today, don't have all the answers, Richard Heyser pinned it to the use of negative feedback in active amplifiers. He had done comparative tests, himself, and found this so. Still, his forte, then, was loudspeaker measurement, and he became even more famous in the next few years for this, rather than amplifier distortion measurement. However, he did point out to us, the low 'dimensionality' of the test signals that we used, and how the test equipment developed to evaluate these test signals could be fooled by REAL sound. He, in fact, got us to 'think outside the box' so to speak, driven into us from years of college study and standard textbooks.
I went the way of looking into TIM. Bascom went into extensive measurement, and Geoff stayed with tube design where he made amazing designs. Still, while we knew what we heard, we could not be absolutely sure that EVERY design we made was virtually perfect, with regards to listening response.
I went the way of looking into TIM. Bascom went into extensive measurement, and Geoff stayed with tube design where he made amazing designs. Still, while we knew what we heard, we could not be absolutely sure that EVERY design we made was virtually perfect, with regards to listening response.
In 1975, I heard the first 'breakthrough' in a commercial amp that I had not heard since the dawn of semiconductor amp products. This was the so called 'Otala' amp designed by Jan Lohstroh, and Matti Otala. I compared this to SAE, Marantz and virtually everything out there, and this little amp beat them all. What was 'magic' about this amp? Well, distortion was just OK. Power was marginal, but it still sounded great. Since the design of this amp was published in the AES, several things were noted. The amp was VERY fast, perhaps 5 times higher in slew rate than typical commercial amps. Second, the open loop bandwidth was 20KHz or so. Third, the only protection on the output were power supply fuses, this was important, because we found many amps strongly compromised by 'elegant' protection circuits. Fourth, the output devices had significant quiescent current, more than 150ma ea, even 250 ma, if practical. This meant that the class A power, at low listening levels, was really usable.
Was this the direction for new amp design? Maybe so, but compromises have been made along the way, for more power, adequate speaker protection, lower measured distortion, etc. over the years. Generally, to the detriment of overall high quality design.
Was this the direction for new amp design? Maybe so, but compromises have been made along the way, for more power, adequate speaker protection, lower measured distortion, etc. over the years. Generally, to the detriment of overall high quality design.
It might be pointed out that the 'topology' of the 'Otala' amp was pretty good, we had designed better topologies, in the USA, but where we went astray was in the AMOUNT of feedback used. 20dB once again seemed to be the magic number. More, and you got lower open loop bandwidth, lower slew rate, even if you got better IM and harmonic distortion specs. At first, we thought that TIM was the only culprit, but we found out later, that even amps with 100V/us, could still sound marginal, IF the open loop bandwidth criterion was severely restricted. You see, it is relatively easy just to remove load resistors from the first stage and substitute current sources to get 100-10,000 times more open loop gain. However the open loop bandwidth will diminish correspondingly. This is where we get today's IC op amps. It seems to change the sound, even if other parameters are improved, and slew rate is maintained. This is the puzzle at hand.
No puzzle, just wrong. There no basis other than anecdotal listening to sustantiate this claim. Your pontificating is just a reset to years ago, obviously nothing is going to change even the slightest detail of what you believe.
Does listening impression have no value in assessing the sound quality?
Well, we come to this: There seems to be something wrong that is aggravated by low open loop bandwidth, even though standard measurements don't seem to show much. Over the decades, we have dropped measured distortion from -90 to -140 in some cases. Yet, it does not seem to help much. What we seem to be missing is a phase modulation or 'jitter' type of distortion that normal test equipment completely ignores. In fact, normal test equipment ignores this sort of distortion essentially at the same magnitude that an oscilloscope ignores AM distortion, the normal sort of distortion that we measure. It is because the test equipment is not thought to need this sort of test. I think we will find this important in future.
Is there anything in the type of testing explained here do you think?
http://www.nordost.com/downloads/New Approaches To Audio Measurement.pdf
The article is marketing and has inconsistencies but the type of testing reported may have some merit I suspect.
Last edited:
"Traditional audio measurement approaches are based in the frequency
domain, using FFT technology and the steady state test tones it relies
on."
As soon as you see that, you can be well prepared for what's ahead.
I was particularly taken by the comparative spectrographs on the top of page 4, clearly showing the "tweak" setup to have more HF noise and hash. That, of course, is not mentioned....
- Status
- Not open for further replies.