Slew Rate Question.

[Mr Funk]

Hi there

I read a lot about pre amps on these boards and many people seem to be getting very caught up in importance of the slew rate. I understand that the slew needs to be fast enough if it wants to keep up with strong, fast transient detail, but exactly what would be considered slow enough to be of 'detriment' to the sound and what would be considered fast enough that any faster response is of no benefit, due to the fact our ears don't work any better?

20V/uS seems to be the average for a mic pre amp, but from what I have explained to me 15V/uS (and therefore 20V/uS) is MORE than enough?

Please read explaination below and tech-heads tell me your opinions. This was given to me by a manufacturer of a certain pre amp. I know the guy knows his stuff, but do you agree that 15V/uS is more than adequate?

"Even if a transient were to go from 0V to 15V (that's to the operating rails of our power supply and equivalent to an output level of +21.5dBu) that means that our pre amp can track from 0V to 15V in 1uS or in frequency terms it could track a 1MHz input signal (the frequency equivalent of 1uS). Even allowing for the debate about tracking audio signals above human hearing threshold (20kHz for those few of us with perfect hearing), this means that our mic amp can track an input waveform with a rise time 50x the limit of human hearing, or in digital terms ONE TENTH of a sample running at 96kHz sampling rate.

So I can assure you that a 15V/uS slew rate is (much) more than enough to capture the finest audio details which any real source can deliver. I have nothing against very high slew rate amplifiers but to say they have a dominent effect on the capture of sound is well.......... no comment"!

[Ethan Winer]

Funk,

I agree with that explanation. Too many people get caught up in esoterica that has no relevance to the performance that's actually needed. Witness the enthusiasm some people have for 192 KHz sample rates!

However, an amplifier's slew rate has to be fast enough to track any input signal presented to it. Even though nobody can hear much past 20 KHz, some microphones can capture content much higher than that with percussive sources like tambourines and snare drums. If the amp can't keep up it will generate distortion within the audible range. But I agree that any amplifier that can swing from one power supply rail to the other in one microsecond is surely fast enough.

--Ethan

[McQ]

Mr. Funk:

FWIW, in general, if you wanna find a first-order approximation for the maximum allowable slew-rate, use:

• SR = kAf

where:

• k = 6.28 x 10^-3
  A = Maximum peak voltage
  f = Maximum frequency of interest
  SR = Slew-rate in Volts per microsecond

for example, what slew-rate is required for a 15Volt (+22.7dBu) signal at 50kHz?

• A = 15Volts
  f = 50 (...kHz)

then,

• SR = (6.28 x 10^-3)*(15)*(50)
  
  SR = 4.71 Volts/microsecond

Make sense?

Cheers,
McQ

[Dan Kennedy]

I'm in agreement with both McQ and Ethan with one caveat...

A high RF level can cause problems with some amplifiers if they don't have wide enough bandwidth. This of course is simply dealt with by bandlimiting the input signal. Which then reduces dramatically the slew rate requirement of the amplifier.

Keep the crap out up front, and it's amazing how well things sound, fast or not.

As an edit:

It is important to bandlimit the output signal too, because crap sneaks in that way as well...

[Flatpicker]

Originally posted by Dan Kennedy
...This of course is simply dealt with by bandlimiting the input signal. Which then reduces dramatically the slew rate requirement of the amplifier...

At what frequency do you "bandlimit"?

[chessrock]

And just how would one go about band-limiting at the source? I always thought that was generally something reserved for further on down the line.

[Mr Funk]

I think wahat Dan means the bandwidth limiting is done in the design stage of the pre amp?

Here is a quote from a certain pre amp designer.

'15V/uS is many times what is required to capture even the fastest audio transient. Normally 5 or 6V/uS is fine but again it depends upon the gain you are asking the stage to provide.

There's nothing wrong with very high slew rate amps and I understand the desire for them, BUT when you design an amplifier stage with a very high gain bandwidth using a very high slew rate design, you'll never guess what else it can pick-up?

Radio 1, Radio 2, Radio 3 , and yes Pentium 4

So there is a danger that it becomes very attractive to lots of real world air borne stuff. Yes you can filter it out at the front end using RF filters (oops there goes all the point of a super high slew rate amp) but really it's always a question of the best compromise'.


I wonder if RFI etc is ever a problem for the Buzz Audio MA1.1, which has a super fast slew rate and massive bandwidth?

[chessrock]

Originally posted by Mr Funk
I wonder if RFI etc is ever a problem for the Buzz Audio MA1.1, which has a super fast slew rate and massive bandwidth?

Or how about this one?:

http://www.audioupgrades.com/products.htm

3,000V/uS -- that's pretty insane.

[Dan Kennedy]

Actually the bandlimiting is mostly a matter of the transformers in the recording gear we build. In the neighborhood of 150 Kc.

Keep in mind that GR also builds industrial electronics, from flow meters, humidity control systems, elevator control systems, weigh scale electronics, etc.

Everything we do is subject to fairly stringent regulatory testing, as well as just the requirements for competent design. So the desirable signals are determined, the crap defined, and excluded as much as possible with the smallest impact on the signal.

It's done with inductors, ferrites, capacitors, common mode chokes, resistors, board layout, mechanical enclosure design, all sorts of tricks.

As I've said elsewhere, I'm a big fan of keeping the bad out, letting the amplifiers work at a real good clip, then protect them from the real world again at the output.

But super high speed devices bring their own issues to the interior of a product, including the worst offence, instability.

Slew rate by itself is only one indicator of the quality of a circuit, and one that has gotten latched onto by the public as a major factor, when in reality, to me at least, it's only one of several, and not the most important.

A lot of the blame goes towards early mixing boards and effects units built with LM741's and derivitives like the RC4558, which not only had lousy slew rate, but couldn't get out of their own way when it comes to driving a real feedback network, much less a load. It isn't only the slew rate that makes those amps suck.

Oh, yeah, offset too, which meant all those unbiased coupling caps...

[Flatpicker]

Originally posted by Dan Kennedy
Actually the bandlimiting is mostly a matter of the transformers in the recording gear we build. In the neighborhood of 150 Kc...

Thanks for the info, Dan. Transformers sure solve a lot of problems and I agree it's the way to go...