What I do not understand is: I cited a figure (3) in a paper that helped to author 33 years ago that had distortion artifacts that I had overlooked at the time, and until these artifacts were pointed out to me by an expert in FM transmission.
In my excitement of the find, and its implications, I brought it up on this thread. I am old, and 1/2 blind, and have not used my lab in any serious way, for months, yet I am commanded to reproduce the test, myself, or I am in some way, 'intellectually dishonest'. Now, in truth, I could reproduce the test with similar or even better test equipment with a little outside help, as I am getting from Demian, even now. Yet, I still need to hire Demian's tech to put it all together, and the minute details of the test set-up are vague in my memory, since I last performed a parallel test at the Finnish Government lab in 1976. However, the most important details are in the paper itself, since the point of the paper was to INTRODUCE the new test procedure the the world, at large, and creating an international standard for measuring TIM.
If an individual cannot get the test procedure from Fig 1, and any related text, it shows a deficit in understanding engineering procedures, or something equivalent. Again, insisting that I provide exact details of the test, is a waste of my time, and I don't have a perfect memory, so I would, in any case, have to guess about any minor details that were left out of the paper, myself.
Anyone who feels out of sorts with this, should not try the test at all, since it would imply that some serious change in the test was likely, and every conclusion made thus, will probably fall away into meaningless chatter.
For example, some people might think that the test procedure is inconvenient, and they might change it. Others will not use a UA741 op amp, and a substitute might not show much of anything, just like Bob Cordell's testing from PIM showed very little. This can be shown in fig 4 of our paper where a simple switch position removed BOTH TIM and PIM from the graph, still using a similar peak to peak output voltage and a similar spectrum, to the untrained eye.
In fact, this was the second part of the EUREKA moment for me, as to see why Bob got such limited results in his testing, years ago.
Look everyone, is it not obvious, once it has been pointed out to you?
In my excitement of the find, and its implications, I brought it up on this thread. I am old, and 1/2 blind, and have not used my lab in any serious way, for months, yet I am commanded to reproduce the test, myself, or I am in some way, 'intellectually dishonest'. Now, in truth, I could reproduce the test with similar or even better test equipment with a little outside help, as I am getting from Demian, even now. Yet, I still need to hire Demian's tech to put it all together, and the minute details of the test set-up are vague in my memory, since I last performed a parallel test at the Finnish Government lab in 1976. However, the most important details are in the paper itself, since the point of the paper was to INTRODUCE the new test procedure the the world, at large, and creating an international standard for measuring TIM.
If an individual cannot get the test procedure from Fig 1, and any related text, it shows a deficit in understanding engineering procedures, or something equivalent. Again, insisting that I provide exact details of the test, is a waste of my time, and I don't have a perfect memory, so I would, in any case, have to guess about any minor details that were left out of the paper, myself.
Anyone who feels out of sorts with this, should not try the test at all, since it would imply that some serious change in the test was likely, and every conclusion made thus, will probably fall away into meaningless chatter.
For example, some people might think that the test procedure is inconvenient, and they might change it. Others will not use a UA741 op amp, and a substitute might not show much of anything, just like Bob Cordell's testing from PIM showed very little. This can be shown in fig 4 of our paper where a simple switch position removed BOTH TIM and PIM from the graph, still using a similar peak to peak output voltage and a similar spectrum, to the untrained eye.
In fact, this was the second part of the EUREKA moment for me, as to see why Bob got such limited results in his testing, years ago.
Look everyone, is it not obvious, once it has been pointed out to you?
Hi John,
The cause also seems to come from my analysis, which took place many, many years before Barrie's. The point I am making here is that I am not denying the existence of PIM or the mechanism that created it, nor am I at variance with Barrie's stuff.
I think perhaps the problem here is the speculation on both "sides" as to the meaning and origin of those blips in Figure 3. Without reproducing the experiment and measuring those blip frequencies with high precision, I'm not sure we will get anywhere with this. I think the blips can be explained by ordinary TIM and are harmonically related; you seem to think they are symptomatic of PIM and are inharmonic. Either one of us could be right, based on the very limited detail of the data we have been looking at. There is no denial or politics going on here, just a purely technical disagreement of what the blips might be.
Given that the poor little 741 was being hammered so hard, I really don't care whether the blips were TIM or PIM. As Jan pointed out, when something is getting close to slew rate limiting, all sorts of things are possible. Here, on this forum, we are presumably talking about the subtleties of high-end audio, where we are well away from abusing a 40-year-old garbage op amp, and where we are well away from slew rate limiting.
I thought that we had agreed to get off this pesky PIM thing and move on to more interesting aspects of the negative feedback debate.
Cheers,
Bob
John,
I agree. There are numerous people on this forum who can reproduce the test. All they need is a 3.18 kHz squarewave generator, a 15 kHz sinewave oscillator, a low-distortion mixer to add them in the ratio of 4:1, and a 30 kHz or 100 Khz single-pole LPF on the output of this source. Then they just need to look at the results on a semi-decent PC sound card-based spectrum analyzer when they hit the 741, connected as described in Figure 3, with this signal so that it puts out 5 V p-p. The resulting distortion is so high that a spectrum analyzer with high dynamic range is not required. As far as I know, that's about all there is to it.
The only thing I'm not sure of is whether the Figure 3 was with DIM-30 or DIM-100. Do you know, John?
Cheers,
Bob
Frankly speaking, I do not see any reason why to repeat that test with the bloody slow old uA741. I believe what John has measured. Those who do not may either shut up or repeat the experiment (if they are not lazy). These verbal wars are leading to nowhere and bring nothing new. I am sure we have more important issues to discuss and solve.
Personally, PMA, I would start with a 741, BUT I would think that a 4558 or its equivalent should be the next on the list. It is slightly faster, and we might have to go to 10V pp to get interesting and easily measurable results, but it IS representative of what is in most mid-fi audio equipment that we use today, at home, in our autos, etc. Perhaps it is not shown on a schematic of the audio equipment used, but it is a component in the processors that our audio signal flows through. Trust me, 'they' are not going to use something better, until we prove to them, that they are comprised in some way, more than traditional measurements show.
Gerhard did a conclusive experiment:
http://www.diyaudio.com/forums/showpost.php?p=1917377&postcount=1027
Even so, JC disagrees:
http://www.diyaudio.com/forums/showpost.php?p=1917799&postcount=1043
Why on earth would anybody bother to power up a signal generator when it's obvious that no experimental or theoretical argument is going to be accepted by JC and his gang.
I couldn't care less about the 741 as well; to me, the issue is that JC and his crew are attempting to use those 40 years wrong conclusions as a ultimate argument against opamps and NFB. Speaking about intellectual honesty...
Speaking of an intellectual honesty, I've asked you how good are openloop parameters of a thingy containing voltage to current converter loaded on current to voltage converter. However, using the latest available devices you get.
Me? I certainly missed your question. Would you please point me to your original query and I'll certainly adress it, if I know the answer.
It is essential that the old measurements are repeated, so we can see that
the new test setup gives the same results as the old. Everything else would
be a cargo cult.
(But then most of the hi end market has cargo cult as a business model.)
Read Richard Feynman's article on solid scientific work.
Let me google some of it for you:
<http://www.lhup.edu/~dsimanek/cargocul.htm>
I'll do another batch of measurements over the weekend.
Frequency and chip proposals welcome.
regards, Gerhard
BTW the 741 as in Fig. 3 was sandbagged with +-5V supplies only.
Kinda unfair for 5Vpp output.
Just to be fair, I think that Gerard's measurements are interesting, but they don't show me what I found on my own graph, as easily as I would like. It might just be our individual interpretation of the different artifacts. For example, symmetry around the initial test tones is nearly impossible, so far as I know, with just addition, multiplication and subtraction to create IM products. It just doesn't fit, yet I am told that I am ignoring Gerard's input.
I must admit that Gerard made a pretty good test, although I would have left out the second IC at the output. Less is best, I would think.
I really don't know or understand his conclusions, if any.
I must admit that Gerard made a pretty good test, although I would have left out the second IC at the output. Less is best, I would think.
I really don't know or understand his conclusions, if any.
Last edited:
It was in my previous answer to you, about power transfer. It seems to me you like to load current sources on opamp inputs in inverting modes. How linear is coupling between such thingies, and how good are overall parameters of such a combination?
I don't have the foggiest idea how you concluded I like that particular configuration. However, I'll look into the OLG when I'll have a spare moment (perhaps tonight). Otherwise, if you enjoy matching to source and load impedances to provide a maximum power transfer in audio, it's your call. I disagree it makes any sense unless you are using 1 mile long cables.
Hi John,
A couple of other mid-fi candidates to test would be a TL071 and a 5534. Heaven only knows how many 5534's are used in studio equipment.
Cheers,
Bob
The post amplifier was necessary because your tests had been done with a
5K load including feedback and I would have lost 40 to 50 dB feeding the
50 Ohm RF spectrum analyzer, so noise would have been an issue.
AF spectrum analyzers may have 1 Meg inputs, but then AF is RF without
most of the problems. I first tried a TEK P6201 FET probe, but at this level
it produced ridiculous IMD by itself. (probably open loop, but the step response is great.)
The active probes from my Infiniium scope have a funny Coax+9Pin
connector, so they cannot be abused for anything else.
The post amplifier is not expected to correct anything. If at all, it will
make things worse. And then, the LT1028 was a lucky choice, since
LT1028 in first AND second stage were not really challenged by this
abusive level.
I was under the impression that I had successfully verified your measurements
from 30 years ago. The form of the spectrum is the same, and I won't
fight for minor differences like -60 vs. -66 dB. Given the generation mechanism,
i.e. mixing half a dozen of times, even very small amplitude variations on the
input side give vastly different results.
I have abstained from any conclusions, just injected some facts as input to the discussion.
Gerhard
If you want zooms around interesting frequencies, just tell the numbers.
Last edited:
- Status
- Not open for further replies.