You guys amaze me, but then we did the research in the 1970's, almost 40 years ago. Have things changed since then?
In some ways, yes.
We found that playing analog vinyl records could create problems with slew rate limiting, then, just as it potentially could have caused problems yesterday when I played some vinyl records on my audio system. Of course, my audio playback does not have any realistic potential for slew rate related problems, but it was not always that way.
In my opinion, to ignore the work of people who invested a great deal of time studying the problem, is to recommend the mediocre, rather than strive for audio quality.
When Matti Otala and I, separately made successful amps that measured 100V/us slew rate, 40 years ago or more, we HAD to do it with power transistors with 4MHz F(t). Today you can get 10 or more times faster, equivalent output devices, and it is easy to get good slew rates, if you know what you are doing.
Julian Vereker, formerly of Naim, told me, almost 40 years ago that I should design at least 30A of output current into the power amp I was building at the time. Why? By the way, did you know that Julian's choice of an NPN power device (he used quasi complementary) had an F(t) of 10MHz or so? Trade secret for years. And so on...
In some ways, yes.
We found that playing analog vinyl records could create problems with slew rate limiting, then, just as it potentially could have caused problems yesterday when I played some vinyl records on my audio system. Of course, my audio playback does not have any realistic potential for slew rate related problems, but it was not always that way.
In my opinion, to ignore the work of people who invested a great deal of time studying the problem, is to recommend the mediocre, rather than strive for audio quality.
When Matti Otala and I, separately made successful amps that measured 100V/us slew rate, 40 years ago or more, we HAD to do it with power transistors with 4MHz F(t). Today you can get 10 or more times faster, equivalent output devices, and it is easy to get good slew rates, if you know what you are doing.
Julian Vereker, formerly of Naim, told me, almost 40 years ago that I should design at least 30A of output current into the power amp I was building at the time. Why? By the way, did you know that Julian's choice of an NPN power device (he used quasi complementary) had an F(t) of 10MHz or so? Trade secret for years. And so on...
Last edited:
Anyone who wants to get a very good idea as to how we got 50V/us for amps, and 5V/us for preamps as a TIM standard, should visit Walt Jung's website and read his EXTENSIVE articles on SID,( which is TIM.) If only measurements were made with that amount of care today, we would not be slipping backward.
We all came around to the conclusion that you could still avoid TIM, even if you have a relatively low open loop bandwidth. However, the problem of PIM is aggravated by low open loop bandwidth. PIM is what we think is as equally important as TIM. This seems to separate the subjective differences when comparing fast amps, all else being equal.
Yeah - so? Is that an outlandish requirement?
These days when it's DAMN hard to find any newer power device with an Ft of less than 20 MHz, and when most are above 30 MHz, and some do 60 MHz?
Otala and Lohstroh got 100 V/uS for 25W/8 Ohms and 3 MHz BD devices.
Last edited:
In the CFA vs VFA DIY section... designer(s) in both camps quickly learned how to make higher and higher slew rates.... just to see its effects... both measured and listening results. It seems IMO that those who built some of the designs thought the faster SR was better up to a certain point. That point though was higher than 50V/usec.
Note also that the 50v/usec was mostly based upon the input signal... and not on how to make an amplifier be extreamly linear and what that requires. Todays designs which are SOTA have higher SRates than 50. Go over to that section and read it and come back with your own 'IMO' about todays SOTA numbers and listening results.
THx-RNMarsh
Note also that the 50v/usec was mostly based upon the input signal... and not on how to make an amplifier be extreamly linear and what that requires. Todays designs which are SOTA have higher SRates than 50. Go over to that section and read it and come back with your own 'IMO' about todays SOTA numbers and listening results.
THx-RNMarsh
Last edited:
John, some subject have a way of reinventing themselves. Slew rate is one of them ever since the Japanese industry stopped making a big fuss about it, as they did in the late 70ies, when the race was on between Sansui and Kenwood, mostly, but thers were not too far behind.
Capacitance in the PSU is another such subject, and I expect it to soon turn up here yet again.
Several thousand messages ago, you and I mostly discussed it and agreed that the bare minimum we should never go below is 1 V/uS per peak volt of output, but we also agreed that more won't hurt. You mentioned then, as you did a while ago in another thread, that designing for 100 V/uS was not too much of a problem, if one put his mind into it straight away from the initial design stage.
It's the "bad students" who need to have it repeated every now and then. Flunk once, try again.
I am NOT a slew rate freak, but I agree with your initial designs approach. Demian told something along the lines that nobody really knows how and why, but wide bandwidth (and hence high slew rate) amps somehow seem to sound better. That statement literally perfectly reflects my own experience, so obviously I agree.
Mark my words, the next topic in the near future will be something along the lines of what is the minimum capacitance we need and can too much capacitance be a bad thing for our amps.
Slew rate limiting comes in when you have not enough current in the input stage to charge the cap load (normally the compensation cap). So, the cure is a no-brainer...
Jan
Of course you want a safety factor - that's part of the audio folklore
A 20 kHz signal has a slew rate of around 0.15V/us/V; so aiming for 1V/us/V implies a safety factor of 6 to 7 times.
Of course, at the frequencies and levels where most of the music is, the safety factor is very much higher. At 1 kHz full power (which is relatively rare in music) the safety factor is more than a 100...
Jan
flat line to sine? what you see is finite slew rate == that of the continuous sine - its only the even higher order derivatives that become larger than the continuous case
no physical system can do ideal square wave - air can't transmit it, microphone diaphragms won't accelerate at "infinite" rate
its common for most recording mics to have <20 kHz corner frequency - a few used for drum kits may be 25 kHz - Earthworks fastest "recording" microphone is 50 kHz - long way from electronic amplifier's limits
no physical system can do ideal square wave - air can't transmit it, microphone diaphragms won't accelerate at "infinite" rate
its common for most recording mics to have <20 kHz corner frequency - a few used for drum kits may be 25 kHz - Earthworks fastest "recording" microphone is 50 kHz - long way from electronic amplifier's limits
Last edited:
Personally, I see the whole discussion between what's better, VFB or CFB, as totally ridiculous. Possibly because I have heard some great amps from both camps and many more bland amps from both camps.
Just as I have heard some good and many bad mediocre amps using tubes, MOSFETs, BJTs and whatnot.
I believe it's not what you use, but how you use it within a complete design. Do it thoughtfully and consistently, with good knowledge of the parts you are using in any specific place, and no matter what you use you will likely get some good vibes. Slap it together using the world's best ever devices and you'll end up with a mediocre sound at best.
Nobody really needs anybody, John, you, Demian or Jan to tell them what can be had on the market. What you can do is help those who want to learn how to make the best use of what they have or can relatively easily have (buy). That kind of advice is literally priceless.
In a private correspondence, Demian told me, among other things, a few pointers which short circuited me to the point that he sent me back to the first grade. But I understood him perfectly, partly because he was both crystal clear about it told me the whole thing, and I spent the last month and a half experimenting on the many variants of what he sent me out to look for. And I am pleased to say that I learnt more in that short time than I did in the last two or more years.
Thorsten Loesch always does that to me, been doing it for the 11+ years. If there's any justice in this life, he and I will actually meet in person some day, and I will be able to at least treat him to a decent lunch, or dinner.
THAT'S what you "old guys" from the "Old School" can help with as no-one else can. THAT'S why a few members here get on my wick with word picking everything John says, they can't argue with his logic so they pick his words and waste a lot of our time for essentially nothing.
So it was with slew rate. Joh mentioned 100 V/uS as a healthy limit, and immediately people jump and start nitpicking, few understood the general point that was made.
It becomes very tiresome sometimes, not to metion unfiar.
Not disputing that, but look hard enough and you'll find an exception to everything (except taxes, even God fears the IRS and its international equivalents).
Talking about general principles only, not precise data for precise models.
And while perhaps not strongly related, not completely unrelated either.
Of course you want a safety factor - that's part of the audio folklore
A 20 kHz signal has a slew rate of around 0.15V/us/V; so aiming for 1V/us/V implies a safety factor of 6 to 7 times.
Of course, at the frequencies and levels where most of the music is, the safety factor is very much higher. At 1 kHz full power (which is relatively rare in music) the safety factor is more than a 100...
Jan
"...part of audio folklore"?
Engineers like to shave just below the skin then, I take it? Given that they are not misled by folklore?
Do you shave your chin differently than your upper cheek, since it is higher and nearer to yours, perhaps it needs a bigger safety factor?
You know, perhaps you should propose that kind of reasoning to a mechanical engineer, see what he'll tell you. If they built cars that way, most of us would be very much dead by now.
Pretty much the same question could be put regarding amplifier power output. At times, I have the feeling that some amps were made for lab use rather than music.
no I "pick" at John's statement's that appear at odds with well established engineering principles
he has amply demonstrated unwillingness to engage in actual dialog - he just wants to make pronouncements - refuses to actually go into technical depth in justifying, reconciling his statements with real EE, Circuit, Signal Theory, or Psychoacoustics where conflicting results are pointed out
in actual dialog he would correctly restate his detractors objections showing he actually understands and present his reasons for differing - not call them personal attacks
he has amply demonstrated unwillingness to engage in actual dialog - he just wants to make pronouncements - refuses to actually go into technical depth in justifying, reconciling his statements with real EE, Circuit, Signal Theory, or Psychoacoustics where conflicting results are pointed out
in actual dialog he would correctly restate his detractors objections showing he actually understands and present his reasons for differing - not call them personal attacks
- Status
- Not open for further replies.