 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
In Reply to: Re: what determines the "speed" of an amp? posted by Duke on February 12, 2000 at 12:01:16:
Duke wrote:"Back in the early 80's high slew rate was all the rage, but then along came "perfect sound forever", and suddenly everything above about 10 kHz was hash anyway. Suddenly slew rate made no sonic difference, except to the vinylist. Now with SACD and upsampling DAC's available and becoming more affordable, slew rate is going to matter again."
Uh, why?
At 48kHz (the high frequency limit of 96kHz sampling), if you were to swing it +/- 60 volts (or the equivalent of about 225 watts RMS into an 8 ohm load) its maximum rate of change would be a little over 12 volts per microsecond. Of course this scenario would never exist in any realworld home audio system. Under realworld conditions, anything anywhere near 48kHz would be so low in level I doubt you'd have anything with a rate of change greater than a volt per microsecond at typical power levels. So most any amplifier with a slew rate of a few volts per microsecond wouldn't be in any danger of slew rate limiting even with the increased bandwidth of the new formats.
se
just picked up copy of the march 2000 hi fi world, page 55 by noel
keywood has some spectral analysis of cd and dvd-a.
the john basile quartet on dvd-a at 24/96 (no other info on which
segment, length, etc) shows peak at -5db(300hz?) , 20khz at -40db and
-46db at 30khz. there is stuff at -70db at 40khz. do not know the
(octave,third, sixth?) of the analyser but this shows there is
energy up into the supersonic. might be also why there are tweeters
made for the japanese market up to 120khz.leelock
leelock wrote:"just picked up copy of the march 2000 hi fi world, page 55 by noel
keywood has some spectral analysis of cd and dvd-a.the john basile quartet on dvd-a at 24/96 (no other info on which
segment, length, etc) shows peak at -5db(300hz?) , 20khz at -40db and
-46db at 30khz. there is stuff at -70db at 40khz. do not know the
(octave,third, sixth?) of the analyser but this shows there is
energy up into the supersonic. might be also why there are tweeters
made for the japanese market up to 120khz."Ok. Let's take 40kHz and assume a full-scale swing and set full scale at +/- 2 volts peak. That's a rate of change of 0.5V/us. Run that into an amplifier with a gain of 20 (26dB), and we get 10V/us. Add John's recommended 5x guardband, and we get 50V/us. Piece of cake. :)
se
just saw in another mag that the chief engineer of tannoy says that
cymbals go up to 100khz. when we get the 192khz stuff in place does
that mean the 50 v/us goes away and we trade up?i seem to remember some old articles that says marantz japan has
tried sampling rates up to 500khz and in each case sound quality
improvement was audible. is this where we are headed after the 192khz?of course by that time i will need hearing aids to get it--hmmmmm-- do
they make some lightweight single ended tubed triodes with bone
conduction------leelock
The microphones will have to improve if we want to get 100K response, except
for special, very noisy, instrumentation mikes. The standard is about 40KHz for the best practical mikes today, and has been for the last 30 years.
it is true the popular and mainly colored(musical?) mics in wide use
do not go up in response as the capsule needs to be small and the
electronics very quiet(expensive?). there is also no choice in polar
patterns--only omni for the really high end specs.however earthworks(founded by ex-dbx guys) makes instrument and
music mics with outstanding characteristics. their instrument mic
is omni pattern and get out to 55khz at -3db and with self noise of
26dba with max spl of 150db. the low end is something like 5hz at
-3db.i guess this is a chicken and egg thing where there was no demand for
the response possible with the likes of the earthworks (i think the
small diameter b&k mics also went up to very high khz) since the
domestic playback systems could not take it. but just like the super
tweeters now coming out in japan with 100khz response i think the
industry will respond with 50khz and higher mics once sacd and dvd-a
starts hitting the market in volume. imagine a digital mic with smarts
to average the noise out and microfabrication in silicon....
leelock
Thanks for the input. I have not followed this manufacturer, but I have helped design mikes over the decades and know the tradeoffs. This is the problem: It requires a small area diaphragm to have a high resonant frequency and to have a flat response at high frequencies. The electronics has to handle VERY high Z and still be very quiet. This is a difficult tradeoff, added to the fact that small diaphragms have low output as well. B&K made very small instrumentation mikes, but they were very noisy and could only be used in very high ambient environments or measurement purposes.
Cymbals may, indeed, have harmonics out to 100kHz. That is pretty much meaningless, though, since 16/44 digital systems prevent any signal past 22kHz.Trevor
Trevor Wilson wrote:"Cymbals may, indeed, have harmonics out to 100kHz. That is pretty much meaningless, though, since 16/44 digital systems prevent any signal past 22kHz."
Yes. But what got this sub-thread started was commentary on the new digital audio formats and their extended bandwidths as it relates to slew rate.
se
CD's, I don't listen to no stinking CD's! ;-)
.
Your statement makes sense, but once upon a time so did the assertion that a 44.1 kHz sampling rate would accurately replicate signals up to 22 kHz. Why does a higher sampling rate sound better to me, when I can't even hear 22 kHz? Okay maybe this is apples and oranges, but it seems to me sampling rate is in some ways analogous to slew rate.Now that I have an upsampling DAC I hear a greater difference between amps, specifically on piano and cymbals. Over the past two weeks I have experimented with six different amps - solid state high power, solid state low power Class A, single-ended 845 triode, and three OTL tube amps. I hear a greater nuance and sweetness in both the crash and shimmer of cymbals, as well as the impact (yes, impact) of piano, with the high slew rate OTL gear. Is this cause and effect? I don't know, but for now I believe so.
Slew rate matters because of transient modulation distortion (TID).You want your amp to be relatively immune to the effects of RF noise.
RF noise of sufficient level will cause amplification stages to
hit a "brick wall" of sorts if you do not have sufficient slew-rate
(and/or power supply rejection), causing audible distortion.A lot of high quality amplifiers have relatively large bandwidths,
over 200 kHz. This is a way to ensure that you still have a fair amount
of useful gain at 20 kHz. This gain, if taken linearly, will decrease
distortion, phase, and output impedance (improve damping factor).But a high bandwidth also means higher frequency noise. To properly
support this "noise bandwidth" without TID you necessarily need a high
slew-rate.Matt
Steve, I have to correct a wrong assumption on your part as to what is the optimum slew rate for an amplifier. It is important that the slew rate limit is NOT approached, as this represents VERY HIGH distortion. Also, only small amounts of hi frequency need be present to create a VERY HIGH RATE-OF-CHANGE of the audio signal. Think about a 100HZ sq wave with a 1us rise time. The info at 100KHZ will only be 1/1000 the amplitude at 100HZ, but if it is not there, then the risetime will be compromised. This is an extreme case, but it shows that sine waves can be a poor judge of slew rate potential. We have found for audio that a minimum slew rate of .5V/us for every volt of output either + or -, is necessary. This gives 50V/us for a 100W amp. Most serious designs, even Parasound, has more than twice this number. This info can be shown in Eero Leinonen, Matti Otala, and my paper: 'A Method for Measuring Transient Intermodulation Distortion' in the AES Journal in Apr.1977.
john curl wrote:"Steve, I have to correct a wrong assumption on your part as to what is the optimum slew rate for an amplifier. It is important that the slew rate limit is NOT approached, as this represents VERY HIGH distortion."
Howdy, John.
I wasn't attempting to ascribe any optimum slew rates for amplifiers. Just giving it a sense of scale. And a signal with a rate of change less than the slew rate of the amplifier will not slew limit the amplifier or else the slew rate of the amplifier has obviously been mis-specified.
"Also, only small amounts of hi frequency need be present to create a VERY HIGH RATE-OF-CHANGE of the audio signal. Think about a 100HZ sq wave with a 1us rise time. The info at 100KHZ will only be 1/1000 the amplitude at 100HZ, but if it is not there, then the risetime will be compromised."
Well technically if you have a square wave with a fundamental frequency of 100Hz, the magnitude at 100kHz should be zero since that's an even multiple of the fundamental. But I see what you're trying to say. So let's take a closer look.
Let's say you've got a 1 volt square wave with a fundamental of 100Hz and you're carrying that out to the 999th harmonic. The rate of change of a 99.9kHz sine wave (and therefore the maximum rate of change of a 100Hz square wave carried out to the 999th harmonic) is 0.6276902 V/us. If you remove that harmonic (leaving you with the 997th), the rate of change would be 0.6264336 V/us. A difference of 0.0012566 V/us or 0.002% by my reckoning. Not very significant I don't think.
If you were trying to make some other point, please clarify.
"This is an extreme case, but it shows that sine waves can be a poor judge of slew rate potential."
Depends how you use the sine wave. The rate of change of a square wave can be no greater than that of a sine wave equal to the highest frequency component of that square wave.
"We have found for audio that a minimum slew rate of .5V/us for every volt of output either + or -, is necessary. This gives 50V/us for a 100W amp."
That last part doesn't seem to add up. An amplifier delivering 100W RMS into an 8 ohm load will be swinging +/- 40 volts, no? And if you're recommending 0.5V/us for every volt of output, that'd come to 20V/us instead of 50.
se
Steve,I have to agree with John on this one, I followed the whole TIM thing from the beginning, and read the papers and literature. Some tried to dismiss it as 'good design practice' and unnecessary to be specifically addressed, others tried to play the slew rate numbers game and trivialize what was needed.
The slew rate, is by definition, the point at which the amp is literally pushing the signal out as fast at it will go. The feedback loop has been broken, the amp is racing to catch-up. The actual input signal has dissappeared, and all that is being output is an error signal of the amp racing to catch up. This is indeed 100% distortion and then some. Many amps that were specified for the best numbers use a huge amount of overdrive to maximize the slew rate, pushing the amp very hard into non-linear operation. Others may remove a front end filter or some of the protection circuitry to allow the amp to even reach the specified slew rate on the test bench. For most, to assure some semblance of linear operation (not necessarily low distortion), the slew rate of the signal must be kept to a fraction of the amps rated slew rate.
CD is not the only source, neither is 48 kHz DAT tapes. As a long time owner of a MC cartridge front-end TT system, and quite a few direct-to-disk or half speed vinyl recordings, I am regularly faced with significant information out past 20 kHz.
Jon Risch
John Risch wrote:"I have to agree with John on this one, I followed the whole TIM thing from the beginning, and read the papers and literature. Some tried to dismiss it as 'good design practice' and unnecessary to be specifically addressed, others tried to play the slew rate numbers game and trivialize what was needed."
I wasn't attempting to dismiss or trivialize anything. I was simply trying to give some sense of scale to the issue and express my opinion that I don't believe the new digital audio formats necessarily represent a call to arms with regard to slew rate (and it was the new digital audio formats that was the context of the message I originally responded to).
se
Of course, you are correct in this, Steve. It doesn't change my calculations, however. Most amps, today, are fast enough. This was not true perhaps 20 years ago. Both Crown and Marantz then made large amps with slew rates of 5V/us. This could be a problem, but today amps are generally more than 50V/us. It has been my experience that 10-20V/us is a bit too low for best performance, but it is mostly because of other factors related to slew rate, not the slew rate itself that is responsible.
Small point of correction/clarification, here, John. My old Marantz Model 500 (ca. 1974) has a slew rate of 11V/uS. It was a very impressive sounding beast, in it's time (when it didn't blow up). I must rebuild it sometime.....Trevor
Did you measure it, or did you read the spec sheet?
Both. A really long time ago. I acquired the amp and used in my system, whilst I was the Australian service manager, for Marantz. I vividly recall the sonic improvement, over the Japanese product, I had been using, up until that time. On the bench, both amps measured almost identically. The sound was something else. It was the amp, which started me on my journey. It was the amp which taught me to view spec sheets with some degree of scepticism. It was the amp, which showed that all amps do not sound the same.Trevor
I still maintain that the specified slew rate of the Marantz 500 is 5V/us. We used the Marantz 500 in the '70's for a low frequency amp. I can't find a direct record of that spec. and will consider any proof that it is greater, but I stand on my memory. For the record, the slew rate of the Phase Linear 700 is 11V/us.
I've just checked my service manual, for the numbers. It states:"Slewing rate: Faster than 11 Volts per microsecond."
The owner's handbook says the same thing. Yeah, before you ask, I've retained the original packing box, too. Whasat? Did someone mention 'Anal Retentive'?
How long did your 500 last? As service manager, I repaired every one of the things that made it to Australia (only 3 did), at least 5 times, before I got a hold of some decent Japanese output devices, to replace the horrible old Motorola things. Legend has it, that Marantz built 300 Model 500 amps. From initial design, to the end of the warranty, the whole episode was a US$3 million loss, for the company. Legend has it that Jim Bongiorno had something to do with the design (I'm not suggesting that the two facts are related, BTW). Marantz replaced the 500, with the 510/M. Possibly the foulest sounding amps, ever released by ANY hi fi company. It was reliable, though.
Trevor
... when he identified AA as the best audio forum on the Web.John and Trevor,
I once watched a frined pop the electrolitics in his 500, not once, but twice! Both times he was trying to test at full power on a McAdams tester, but the caps blew like firecrackers before clipping set in. The second time it happened, he was trying to prove to everyone that the first time was a fluke. Both times he was showing off the new gear to a crowd, and both times Marantz repaired the amp. After the second incident, the amp was returned with a note requesting that he not do any more testing on the McAdams. I guess those oil-cooled resistors just presented too complex a load for the poor thing. It sure was an embarassment to Dennis.
I actually sold one of the things to a guy once, but only at his insistance. Not that it sounded bad, but after the confetti came out of two of the things I found it hard to recommend them.
Interesting conversation guys.
Thanks,
Charles
The McAdam, I used, had air cooled, aluminium-clad Dale, 250 Watt resistors, mounted on a large heatsink. I melted the solder, on the cables, connected to the resistors, under full power tests, with my Marantz 500, once. The 500's most serious problem, was the fragility of the output devices. There are two electros, used on the driver supply, which were prone to failure. I wouldn't like to see the main electros fail, though. Those puppies were BIG. The mess would be terrible.Trevor
Steve, we have been at this amp business for a long time. We've also done the research and published it decades ago. We have heard it all before. Now, the .5V/us per volt "rule of thumb" is related to the volts peak to peak of the audio signal, since a square wave can start at the - and end at the + extreme of the amp. I won't quibble over 40 vs 50 V/us. Most of us do over 100V/us anyway because we can do it easily today with modern transistors/FET's and it's probably better to be on the safe side. Slew rate limiting is essentially 100% distortion of part of the audio signal. To keep this distortion below 1% or so on this rapidly changing part of the audio signal, it is important that we not approach clipping the input or second stage of the amp. It is interesting that low slew rate amps usually have very nonlinear input stages, because emitter degeneration is typically omitted, and this makes the situation worse. This is why we suggest at least a guardband of approximately 5 times, in order to not generate significant TIM below the slew rate limit. We have found transients that approach .1V/us pp and this would lead to .5V/us perV pp for an amplifier. I have measured it myself with a fast storage scope and mistracking MC cartridges after RIAA EQ is applied. I hope this clarifies the situation.
john curl wrote:"Steve, we have been at this amp business for a long time. We've also done the research and published it decades ago. We have heard it all before."
I'm sure you have.
"Now, the .5V/us per volt "rule of thumb" is related to the volts peak to peak of the audio signal, since a square wave can start at the - and end at the + extreme of the amp. I won't quibble over 40 vs 50 V/us. Most of us do over 100V/us anyway because we can do it easily today with modern transistors/FET's and it's probably better to be on the safe side."
I was responding to what you said in your initial message, that there should be 0.5V/us for every volt of output "either + or -." Apparently that's not what you meant to say, but that's what you did say and so I took you at your literal word (it's been perhaps better than a dozen years since I read your's and Otala's papers so none of this is still on the tip of my brain). + OR - would indicate peak voltage, as opposed to peak-to-peak which would be + AND -. And for 40 volts peak, that would have equated to a slew rate of 20V/us instead of the 50V/us you specified and that's why I said it didn't appear to add up.
se
Steve, typing on the fly makes for bad grammar, or at least confusing statements at times. I agree that my statement was ambiguous, after reading it later.
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
