Have any of you 'critics' actually LISTENED to this amplifier? I have one in my office, in my sound system. It is VERY GOOD, almost great, especially for its time of design.
Anyone know where Jan Lohstroh is? Met him once, interesting guy, but I don't think he did any more audio research, but I am not sure.
Anyone know where Jan Lohstroh is? Met him once, interesting guy, but I don't think he did any more audio research, but I am not sure.
The term 'TRANSIENT INTERMODULATION DISTORTION' was a technical term coined to separate it from 'STATIC INTERMODULATION DISTORTION' or what we liked to call it at the time SMPTE IM.
What is meant by this is that normal curvature of the transfer function creates IM distortion, however feedback often covers it up.
Transient intermodulation distortion or TIM is a separate IM source that is related to the rate of change of the signal. Usually this happened with transients, so this was added to IM. It is JUST A DEFINITION. It went well in Finnish I think.
SID or SLEW INDUCED DISTORTION was a parallel term used by Walt Jung, derived from an earlier IEEE (I think) paper. It too, has merit as a definition, but it sounds strange when translated into Finnish, so Matti was against it.
Doctor Cherry suggested 'SLOPE DISTORTION' just so he could get some sort of credit, I am not sure what Bob Cordell decided upon, we will have to ask him.
What is meant by this is that normal curvature of the transfer function creates IM distortion, however feedback often covers it up.
Transient intermodulation distortion or TIM is a separate IM source that is related to the rate of change of the signal. Usually this happened with transients, so this was added to IM. It is JUST A DEFINITION. It went well in Finnish I think.
SID or SLEW INDUCED DISTORTION was a parallel term used by Walt Jung, derived from an earlier IEEE (I think) paper. It too, has merit as a definition, but it sounds strange when translated into Finnish, so Matti was against it.
Doctor Cherry suggested 'SLOPE DISTORTION' just so he could get some sort of credit, I am not sure what Bob Cordell decided upon, we will have to ask him.
I might point out something, before people, as usual, go picking and poking at earlier work.
In any frontier, be it physical land, physics, or engineering, etc, there are explorers of seveal different types.
For example, let us go back 150-200 years in the exploration of the American West, onto California. In order to move people from the East to the West, paths had to be found that people could manage.
However, first there were the random excursions looking for something else, like gold, silver, furs, etc. These excursions created random discoveries of natural trails that might be passed on. Then, the pathfinders would be elected to find the BEST trail between points A and B. They would find a manageable trail, but it would not be necessarily perfect.
Then the road builders would come in, with all the previous knowledge and experience and adjust the route to be more optimum.
Now you might think of Matti Otala as a pathfinder. All of his early work is NOT without criticism AFTER THE FACT, but it was pathfinding when it was first presented.
The nitpicks that he as suffered (partially his own fault) are greater than any suffered on me, but then I am more careful, and seldom publish. Matti researched and published regularly until his inputs were rejected in the AES due to the usual politics, and the rest of us saw the light and stopped attempting to publish there as well.
In any frontier, be it physical land, physics, or engineering, etc, there are explorers of seveal different types.
For example, let us go back 150-200 years in the exploration of the American West, onto California. In order to move people from the East to the West, paths had to be found that people could manage.
However, first there were the random excursions looking for something else, like gold, silver, furs, etc. These excursions created random discoveries of natural trails that might be passed on. Then, the pathfinders would be elected to find the BEST trail between points A and B. They would find a manageable trail, but it would not be necessarily perfect.
Then the road builders would come in, with all the previous knowledge and experience and adjust the route to be more optimum.
Now you might think of Matti Otala as a pathfinder. All of his early work is NOT without criticism AFTER THE FACT, but it was pathfinding when it was first presented.
The nitpicks that he as suffered (partially his own fault) are greater than any suffered on me, but then I am more careful, and seldom publish. Matti researched and published regularly until his inputs were rejected in the AES due to the usual politics, and the rest of us saw the light and stopped attempting to publish there as well.
I found a small stash of some PowerGroove MOSFETS, FVN2/FVP2. I think these were in the original JC-80. They appear to be ultra-rare now and I can't find much of a datasheet. I though they still look pretty interesting because they are fast and just high power enough for a lineout stage. Are they worth looking at?
john curl said:
Hi John,
I don't really have a big problem with the term TIM or SID, as long as it is recognized that TIM is just slewing induced distortion brought on by inadequate slew rate. I often use the term TIM myself. I also agree that it was and is important to distinguish various forms of high-frequency distortion (e.g. TIM, CCIF) from "static" SMPTE IM.
I also don't have a problem with the DIM test that Matti came up with for TIM, and built a DIM tester myself. DIM measurement was just a big PIA to carry out, requiring a spectrum analyzer and lots of addition of distortion products. With the equipment at the time, measurement of TIM using DIM was really no more sensitive than THD-20. Indeed, I showed a strong correlation between the two in one of my papers. They largely just differ by a scale factor for most amps. ANY amplifier that has TIM will surely have very measurable THD-20.
It is also notable that if you run a DIM test on an amplifier, it will show up ANY HF nonlinearity, just like THD-20. You could have an amplifier that is not even close to slew rate limiting, but which creates HF distortion in the VAS or output stage that will show up stuff on the DIM test (just as with THD-20).
My big problem all along was that Matti wrongly blamed negative feedback for TIM, when the real culprit was inadequate slew rate and HF linearity. In particular, his assertion that high feedback and low open loop bandwidth caused TIM was plainly wrong. Many, many people have demonstrated that an amplifier can be built with an open-loop bandwidth of less than 100 Hz and NFB greater than 80 dB and have un-measurable TIM.
Many amplifiers of the time had high levels of TIM and THD-20, and were poorly designed in regard to slew rate and how NFB was applied. Matti made a good contribution by shining the spotlight on that. Still, it was not NFB that caused these problems. Perhaps you could argue that NFB ALLOWED designers to be sloppy and complacent about high-level HF lineaity. Just like any powerful tool, it can be mis-used and you can get your finger cut off.
A bad designer can make a bad amplifier whether or not it uses NFB.
Cheers,
Bob
Bob, let us not confuse the issues, settled in the past.
First TIM is a subset of DIM which includes ALL high frequency related distortions, NOT just slew rate.
Second, Matti wanted a test that was not especially effected by the DUT's actual frequency response. OF COURSE 20KHz full power, with a bandwidth of 200K is perfectly acceptable, but how often is that realized? This is why he preferred to sum the IM products that occurred WITHIN THE 20KHZ BANDWIDTH. Is this so surprising? Many here, have argued the same thing on similar measurements. For example, WHY you prefer 19-20KHz IM in your measurements.
Most of the time it is NEGATIVE FEEDBACK that forces the circuit to slew-rate limit in the first place, is it not? Take away the negative feedback, and you take away the slew rate limiting, and then there is no TIM! Look at Greiner's article in 1966, for this.
In those early days, NO significant mention of slew rate and its implications were mentioned in print, INCLUDING textbooks. Look at Cherry and Hoopers book from the late '60's. Any rate limited distortion in that nearly 1000 page book on amp design?
It is important that the slings and arrows passing back and forth are there for a good reason. I find Otala often ridiculed by people who took his work and are still running with it.
First TIM is a subset of DIM which includes ALL high frequency related distortions, NOT just slew rate.
Second, Matti wanted a test that was not especially effected by the DUT's actual frequency response. OF COURSE 20KHz full power, with a bandwidth of 200K is perfectly acceptable, but how often is that realized? This is why he preferred to sum the IM products that occurred WITHIN THE 20KHZ BANDWIDTH. Is this so surprising? Many here, have argued the same thing on similar measurements. For example, WHY you prefer 19-20KHz IM in your measurements.
Most of the time it is NEGATIVE FEEDBACK that forces the circuit to slew-rate limit in the first place, is it not? Take away the negative feedback, and you take away the slew rate limiting, and then there is no TIM! Look at Greiner's article in 1966, for this.
In those early days, NO significant mention of slew rate and its implications were mentioned in print, INCLUDING textbooks. Look at Cherry and Hoopers book from the late '60's. Any rate limited distortion in that nearly 1000 page book on amp design?
It is important that the slings and arrows passing back and forth are there for a good reason. I find Otala often ridiculed by people who took his work and are still running with it.
Scott, good find. Hard to find parts, better than IR or Harris. I know people who would buy them from you to fix their JC-80's.
Yes, high NFB makes inherently slow circuit seem to look wideband enough, though SR limited. And the limited SR is often a result of frequency compensation method used, like limited Ic = C x dv/dt
PMA said:
This argument is a bit chicken and egg. The problems arose from using op-amp concepts for power amps without a full grasp of the issues. Saying the NFB caused the problem is backwards.
john curl said:
Hi John,
First of all, I agree completely that DIM is a testing method and TIM is a an HF distortion mechanism. Although they are often linked, I agree that we need to be careful with our semantics.
I agree, I like HF IM tests because they create products that land in the audio band. That's why I like CCIF 19+20. I also agree that THD-20 measured in an analyzer BW of only 80 kHz is of limited usefulness. No argument there. HOWEVER, it is worth noting that the TIM mechanism, both soft and hard TIM, creates quite a bit of third-order harmonic distortion. So, an amplifier that shows DIM resulting from the TIM mechanism will show up third-order THD-20 at 60 kHz.
I also agree, and have pointed out, that there was completely inadequate understanding of slew-rate-related HF distortions back in those old days.
In fact, I think there is a good chance that Matti himself originally did not know about slew rate and slew rate limiting distortion, and actually stumbled onto it and re-invented it. It took Walt Jung to point out to him that TIM was nothing more than the ultimate consequence of slew rate limiting.
In any case, as I have said over and over, Matti deserves credit for putting the spotlight on slew-related distortions. He also deserves credit for coming up with several useful and novel tests, including DIM, IIM and PIM.
The thing that I stronly disagree with is the generalized blame wrongly laid on NFB for TIM. JUST BECAUSE YOU CAN'T HAVE TIM WITHOUT NEGATIVE FEEDBACK DOES NOT MEAN THAT YOU GET TIM WITH PROPERLY DESIGNED FEEDBACK. This is what you completely miss.
Moreover, TIM is usually the result of an input stage with inadequate signal-handling ability. Guess what? An amplifier designed without NFB MUST have an input stage with large signal-handling capability because it MUST handle the full input signal. I guess you could say that building an amplifier without negative feedback FORCES you to do some things right that you should have done right in the first place (like input stage signal-handling and open-loop linearity).
Cheers,
Bob
scott wurcer said:
I hope everyone understands what I mean. Usually it is necessary to apply a kind of dv/dt restriction somewhere inside when using high loopgain to remain stable, especially for complex load.
Bob Cordell said:
I don't even think that TIM requires feedback at all. I would call the result of a current-starved VAS stage that just cannot feed the output FETs Cin TIM, too.
The usual Cdom compensation usually happens to be limited by the input pair tail current, but in this context that's more of a symptom than the cause itself.
regards, Gerhard
- Status
- Not open for further replies.