Audio Asylum Thread Printer
Get a view of an entire thread on one page
|For Sale Ads|
Yes I know all about M. Otala and his papers. He had an agenda obviously to promote his name. His technical findings are true and accurate. But that doesn't mean they show a PRACTICAL problem. TIM is a classic text book exercise when it comes to audio amplifiers.
Now when Elon Musk populates Mars and in a few ten thousands years when our bodies adjust to the thinner atmosphere, Perhaps TIM will be a legitimate factor for audio amplifiers on Mars where sound can travel faster!
TIM is not necessarily about slew rate. It was promoted as an attack against feedback. The old Einstein time paradox where you can't experience something before it happens. And it's and a serious problem in some areas of electronics design. But not for audio. The path delay of a typical audio amplifier is too short to have any detrimental effect on an audio signal coming from a natural source.
I have no doubt there are a plethora of papers online about the horrors of TIM. But do these papers have traceable credibility? Hint, Stereophile is not an accredited reference!
As for MEANINGFUL measurements, don't use a square wave generator like Mr. Otala did. Take a modern digital waveform snap shot of a musical session. Measure the fastest rise time. Multiply by 5 using the classic engineering 5x bandwidth rule. Then see what the required rise time of the total amplifier circuit is. Now in fairness perhaps some early solid state gear with slow power transistors did fall short, But no amplifier designed past 1980 transistor technology should have a TIM problem.
Edits: 07/07/17Follow Ups:
So apparently Otala's work ushered in a golden age in transistor amp design, until Joe Roberts showed up with Sound Practices in the 90's and ruined everything.
Otala's work ushered in nothing. He documented a phenomena that while technically accurate and easily demonstrable with simple test equipment, it has no practical influence in natural music and sound reproduction.
Now I did say that early 1960s slow power transistors may have caused this effect to be audible, but that was fixed by the progression of transistor technology in the 1970s onward.
As for Joe Roberts, he is the national sales rep for Silbatone as well as an accredited Anthropologist. What does that have to do with this discussion?
The original post here by Tre' was a request for an explanation from Dennis Fraker concerning a statement Dennis made about common mode effects in amps. The subsequent post by Cris I responded to addressed the request by Chris for information explaining the SET phenomena, and I made a reference to frequency modulation distortion in speakers among other things in my post. As you have stated your comments in the present discussion were in reference to FM distortion in amps. The two things are related but they are not the same thing of course. Otala claimed that FM distortion was audible in transistor amps and he provided data to support this. You seem to be claiming that FMD in modern transistor amps is not audible and you have not supplied any supporting evidence or links for this. When I did a web search to respond to your first "hogwash" post I found some controversy concerning Otala's procedure for minimizing FMD by limiting negative feedback to a certain level in amp stages Otala specified, but I didn't find anything agreeing with your position that FMD has become inaudible in transistor amps. BTW with the post above here you seem to have moved the goal post from the 80's into the 70's, and Otala's first manuscript was submitted in 1976.
Denis is a manufacturer of SET amps and has promoted their use over other amp types. Once again the SET amp craze was introduced to the US through Sound Practices magazine by editor Joe Roberts who was a self confessed solid state Krell high end salesman who was listening to Western Electric speakers (sourced from the ceiling of a shoe store) and powered by a Western Electric 300B tube amp, and he thought this sounded better than what he listened to at work.
I would be interested to read any references you provide concerning the non audibility of FMD in transistor amps with a beginning reference point some time in the 70's as of now. My comments in the present discussion were in regards to Modulation Distortion In Loudspeakers which was the title of a JAES paper by P.W. Klipsch from 1968.
I never spoke of "FM distortion" It is called TIM or Transient Inter-modulation Distortion" And you have provided no evidence TIM is audible. Just because you can simulate the problem with test equipment doesn't mean you can hear it with normal program materiel. The subjective camp is always saying how they can hear things no test equipment can measure. But here's the truth with TIM, even legacy test equipment can measure distortions that cannot be normally heard.
How about my test example. Let me refine it further:
Capture a 15khz square wave from a hobbyist quality signal generator on an audio workstation. Then record any random sampling of music or acoustically generated sound effects, gunshots etc, and compare the rise-times. Now tell me if any of those natural rise-times are a steep as the 15khz square wave? Further analyze and measure the rise-times. Then tell me if the path delay through a typical DC coupled amplifier is anywhere near the rise time of the test material? One popular claim is that dues to the path delay of the amplifier, the feedback fails to "catch the leading edge" and that is TIM. That sounds quite plausible to a non-engineer. But enter mathematics and we see how minuscule the problem is. The leading edge is so slow in relation to the amplifiers path delay that all that is missed is a minuscule fraction of the leading waveform edge. Now in light of all the other larger distortions, mainly from the speaker system, that tiny fraction of a waveform edge distortion is audible?
This proves TIM is a theory exercise. It can easily be simulated with even cheap test gear. But it never happens in the real world of audio reproduction. Also note by specifying hobbyist grade test gear, I am putting the test in the TIM camp's favor. A pro signal generator would have much steeper rise times.
My professional background is the design of broadcast television equipment. Today it's mostly digital but in the late 1970s when I started it was mostly analog. I have designed many analog video amplifiers during that era and we used negative feedback extensively. We never worried about TIM and note that TIM should be a much larger problem with video frequencies, eg 6mhz. So if a video amplifier circuit which is very similar to an audio version is fast enough for TIM to be irrelevant, how can it possibly be a problem with audio amplifiers?
I am not interested in articles or AES paper paraphrases from Strereophile, Absolute Sound, or similar consumer product sales magazines. Lets look at this from a solid engineering perspective and analyze the problem. When you do it becomes very clear.
Again I'm not saying Mr. Otala's results are wrong. They are quite correct from a pure technical perspective. But what is wrong is the conclusion that this phenomena is detrimental in modern DC coupled amplifiers.
Thank you for a plausible explanation of your theory that TIM is a non issue in modern (post some time in the 70's) transistor amp design. As to evidence contra to this, Otala (and others) published two JAES papers addressing this: 1. Threshold of Audibility of Transient Intermodulation Distortion (in 1978); and 2. Psychoacoustic Detection Threshold of Transient Intermodulation Distortion (in 1980). In the first paper a group of six of the most sensitive subjects (from a previous test of 68 listeners) were tested and it was found that "The results show that in certain passages of music, 0.003% of distortion is clearly audible". This was followed by ""Low distortion values were perceived only as changes in the sound character, and not as distortion" which could possibly provide somewhat of an opening for your position. The second paper described further testing with improved test equipment and conditions. Otala has provided test results and data to support his assertion that TIM is audible. Another google search by me with the subject "transient intermodulation distortion+audibility" didn't turn up anything saying that TIM is not audible in modern transistor amps, but to be honest I stopped reading after 30 references. There is ample evidence that TIM is audible, but anyone is welcome to make a case for whether or not it's it's an important a concern in modern solid state amp design.
But rather than scrape references to papers online, why don't you address my theory as to why TIM is not a factor on real audio signals?
Those papers you recently referenced are for IEEE and AES members. Otherwise you have to pay for them. So again did you read them or just cite the synopsis. Because buried within them may be the answer I am providing.
P.S. I hold membership in IEEE, SMPTE. SBE
As Otala has performed tests demonstrating that TIM in transistor amps is audible and that it can be detected down to a level of 0.003% by 6 out of the 68 listeners in his test group, then it would seem that there would need to be tests with a group of at least that size to seriously contradict Otala's findings. Neither you or I have been able to find any references to tests like this. Let's try it another way, can you supply the name of a transistor amp from the 80's with TIM specs lower than the 0.003% which was clearly audible to the 6 golden ears? Just what number would you buy as being below the level of audibility, and what amps have specs matching this? Feel free to add in any TIM specs for SET's too as we are on the Tube DIY Asylum
I quoted directly from sentences in both of the abstracts of the JAES papers I referenced, and I assumed that this would be obvious to anyone who has read the papers as you claim to have As to digging occult information out of these papers, well I would have to go down to the library, I would need help with the microfiche machine, and my knee is acting up causing me to drag my foot at times. Well you get the idea. If you were to write a rebuttal to the Otala papers to the JAES, then I would drop the Hogwash stuff, and you'll have to use your real name as you'll be subject to peer review. I've been to a number of JAES meetings and nobody wore a mask or used a pseudonym.
> As to evidence contra to this, Otala (and others) published two JAES
> papers ... In the first paper a group of six of the most sensitive
> subjects (from a previous test of 68 listeners) were tested and it
> was found that "The results show that in certain passages of music,
> 0.003% of distortion is clearly audible".
> As Otala has performed tests demonstrating that TIM in transistor amps
> is audible and that it can be detected down to a level of 0.003% by 6
> out of the 68 listeners in his test group ...
6 out of 68 is a really poor sampling, on the level of a statistical error. While TIM theory looks profound, real audibility is an open question. Its a common behavior in human society to raise a problem (often miniscule or even non-existent), in order to develop and sell "unique solution". BTW, TIM issue was widely used in advertising material back in 198x.
I own two amps engineered to minimize TIM (Sansui G Pure DC series). They sound great. I heard, had (and built myself) others, solid state and vacuum tube, which design don't take into account TIM at all. They also work great.
Thus, I assume, TIM made more noise by itself rather to contribute anything useful.
It would be nice if there was an easily accessible example of TIM distortion on the web or disc, demonstrated with a short musical performance repeated with more and more levels of TIMD until the effect became obvious. Unfortunately none of the Stereophile Test Discs have this, or anything else I have. To be honest I'm not really sure what it would sound like. The 6 golden ears of the 68 testees really couldn't detect it as distortion at 0.003% in Otalas test, they could only tell there was a difference. It would certainly be nice to have another test following Otala's with a larger group which would provide a better base rate, but how likely this is to happen I can't say. My original comments in the present discussion were directed at intermodulation distortion in loudspeakers, which is related to FIM in amps but really quite a bit off topic from the original post by Tre' up page.
As to how important FIM distortion in amps in the entire audio scheme of things is I really can't say. The subject is somewhat interesting to me but it's largely overshadowed by issues in speakers. I built a Hafler DH200 in the 80's but the specs show only IM distortion @ 0.005% and TIM is not mentioned. I though it sounded great when I built it, but it got replaced by what became a highly modified Dyna ST70 in my horn rig, and after that it served in our HT until a Yamaha 5.1 amp bumped it out several years ago. The Hafler sounded good in the HT with Pioneer CS80's until one time the Dyna was awaiting parts and I brought the Hafler back down into the horn rig in the basement and it sounded terrible! Very dull and slow, typical of what has been described as the "mosfet haze". Turning the volume up changed things, but you don't really want to listen to a string quartet for instance at heavy metal levels. This is likely due to the capacitance at the gate of the mosfets changing with the drive level as per the late Alan Wright's observations. So I really don't know what to make of the TIM thing and how it fits into the audio hierarchy, but until we get something better I'm going with Otala's test results. BTW my ST70 is currently triode wired with the ultra-linear tap disconnected, and I wouldn't expect it to do well in THD tests, but it's going to stay this way until I get something better.
> It would be nice if there was an easily accessible example of TIM
> distortion on the web or disc, demonstrated with a short musical
> performance repeated with more and more levels of TIMD until the
> effect became obvious. ...
> To be honest I'm not really sure what it would sound like.
> The 6 golden ears of the 68 testees really couldn't detect
> it as distortion at 0.003% in Otalas test, they could only
> tell there was a difference.
It would require recording from test amplifier dummy load, and playing recorded data again on benchmark unit, thus, subjecting test data again to distortions of benchmark amp (no matter how miniscule they are), speakers, and room acoustics.
As Gusser stated before, TIM issue doesn't make practical sense, and IMHO, it is entirely correct. Someone wanted to make a party, they had a drink and show, now its over.
High stability of the amplifier is much more important factor, and it something that is time to time neglected with aftermath from HF spurious oscillation and loss of clarity to burned components and speaker coils.
But you still don't acknowledge the engineering problems I raised that clearly contradict Otala's claim that this is audible with program material. Ahh, but did he even say that? I do agree it might be audible with test signals but then that's not real program material we listen to.
Just more "appeal to authority". Typical audiophile response when presented with real engineering problems that dispel their beliefs..
I think I see where you're going with this, you are intending to write a JAES paper and you are testing some of the ideas out here. Here's some suggestions to clean up the argument:
To say something is hogwash is to say it's useless and should be thrown away, so drop the hogwash and just say that. You would be expected to provide at least some rudimentary test results, so just have some buddies over for beers and test them. If you add some TIM to a musical selection you are adding energy to the signal and thus effectively raising gain, and that's an old Hi fi salesman trick to turn the volume up a bit to make the thing you are trying to sell sound better. With this you can probably get some people to actually prefer the TIM'd examples, and Otala did note that some of the test subjects did. It's also fairly easy to prove that people are guessing during audio tests if you make them guess, so do rapid switching among orchestral recordings with both loud and quiet passages. After awhile they won't even know what ocean they're in.
You have criticized Dennis and Jeff for promoting speculative and unsubstantiated theories, and you also leave yourself open for criticism in this regard if you don't provide something a bit more concrete. Lisa Randall once said to me when I asked her a question about Loop Quantum Gravity that "It's just a theory", though she did admit she respected the author of LQG. So far you just have an interesting theory, but the ear is the final arbiter in all things audio to paraphrase Olson. And it wouldn't hurt for you to make some appeals to authority too.
BTW Otals did say that the 0.003% TIM was clearly audible to the six golden ears on musical selections, but he noted that they could only hear a difference, but not identify which version had the TIM added. That alone would seem to open an area to study. Also Otala did use square waves as you stated, but he also used triangle waves in some of the tests too.
We're pull'n for ya!
"You have criticized Dennis and Jeff for promoting speculative and unsubstantiated theories, and you also leave yourself open for criticism in this regard if you don't provide something a bit more concrete. Lisa Randall once said to me when I asked her a question about Loop Quantum Gravity that "It's just a theory", though she did admit she respected the author of LQG. So far you just have an interesting theory, but the ear is the final arbiter in all things audio to paraphrase Olson. And it wouldn't hurt for you to make some appeals to authority too."
Stupid analogy! I have provided evidence based on simple physics why TIM is not an issue with acoustically produced sound. You have yet to comment on the theory I have presented. All we hear is "Otala says, Otala says".
As for listening tests, first of all 6 out of 83 is hardly a convincing sampling. We also don't know the controls present or how these tests were conducted. ABT, DBT? "I'm going to play some TIM for you, tell me if you hear it?" Yes some classic audiophile tests were that unprofessional.
You also fail to separate the technical facts of TIM versus the final effects on the systems as a whole. Sure it can be simulated, but does it present an issue in real world audio reproduction? Basic atmospheric and mechanical physics says it can't.
Ok so you don't like TIM being called hogwash. Perhaps you are right in that TIM is real and can be demonstrated - with test equipment. But what is hogwash is that we can hear TIM with modern audio amplifiers designed with late 70s / early 80s technologies.
I'm tired of hearing your monotonous references to 40 year old research papers - OK, 39 years old in fact. Show me how an acoustically produced square wave rise time within the human hearing range can exceed the path delay of a typical DC coupled audio amplifier. Show me a microphone that can capture such an event even if out atmosphere would allow such a pressure wave to propagate in the fist place?
Again show me the math that supports TIM audibility! I know a test generator can do it. But what about real music, sound effects, and speech?
P.S. "If you add some TIM to a musical selection you are adding energy to the signal and thus effectively raising gain"
This is not totally correct either. You are thinking solely of harmonic distortion. IM distortion does not follow the same rules. Look up Bessel functions.
do please explain how slew rate does not translate into bandwidth.
Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
Post a Message!
This post is made possible by the generous support of people like you and our sponsors: