Slew rate calculations - how much do I need?

Status
Not open for further replies.
30V/uS is enough for a 100W RMS/8R amp, that is about 2 uS
raising time at full power...
Values higher than 50V/uS are quite useless, as it s always a trade off
with stability behaviour, so what is gained at a end is losed at the other
one, resulting in an umbalanced amp that will be often worse...
 
I've seen it suggested, more than a few times, that to avoid the early onset of distortion due to slew rate inadequacies that the preferred Slew Rate should be ~10times that calculated as required for maximum signal swing at maximum operating frequency (~Power bandwidth?).
If we apply this to a 28.3Vac signal @ 20kHz then we need ~5V/us from the amplifier and should design for a preferred 50V/us to avoid that distortion.
 
Yes, its common to hear that 10:1 'safety margin' yet I've not heard its derivation so I for one remain something of a 'slew rate skeptic'. Fact is, different input topologies have differing distortions (Bob Cordell uses the term 'Soft TIM' for the kinds of distortions which result from high slew rate signals). The slew rate as a single number tells us almost nothing about the high frequency linearity of the input stage, which is the quality of greater importance.
 
"If we apply this to a 28.3Vac signal @ 20kHz then we need ~5V/us from the amplifier and should design for a preferred 50V/us to avoid that distortion. "

So Jung may have been correct when he suggested 0.5V~2V per µS per peak volt of output (20V~80V per µS for a 100W/8R amplifier).
 
Well, I have done a large set of real life measurements on many, many opamps under different conditions, i.e. with gain (not +1 or -1 gain only). The CCIF IMD 19+20kHz test is VERY sensitive to soft slew rate induced distortion, especially 3rd and further odd components. It is not enough to use 10x margin. Opamps are good for audio if they have SR higher or equal 50V/us. Still, there is a BIG difference between the completely same 2 opamps of the same family, one with 55V/us and second (decompensateed) with 135V/us. It is audible, believe me or not. Forget 5V/us for opamps.
 
Whatever the marketing department thinks will sell more amps. Many amps today are rated at 10% THD (common in digital/switching amps). Actually full power bandwidth is not often specified. Frequency response at 1 Watt is the usual figure quoted if any. THX has dominated the specs world for amps in the broader marketplace, which is why many many amps have 29 dB of gain.
 
Here is some logic around full power bandwidth specs-
The best available distribution media for audio today are based on 176.4 and 192 KHz sampling rates. These are capable of 100 KHz full output response. As such its seems quite valid to expect 100 KHz full power bandwidth. More may sound better.

Very few microphones have response above 20 KHz. Almost nothing besides special measurement microphones gets past 50 KHz. However I strongly believe that anything more than a Gaussian rolloff at band limits will have a sonic impact so I would leave the system limits to be the input transducers.
 
Hmm, this was a really old thread started by me.🙂
I think it still is a really old thread started by you. 😛

Check back the next time the moon's in first quarter.
OK, I think I've figured this out. While doing some other testing I noticed abnormally high THD on the 49990 channel (-86dB 😉) and eventually tracked it down to cracked solder joint on the connector pin for the plus input. (The proto board I'm using not's plated and hence the connectors don't have a good mechanical install since I need to solder underneath them. Also, no chassis yet so no strain relief.) After rewetting the joint I can no longer hear a difference between the 49990 and 49713 and a repeat of the impulse measurements that found a difference previously now shows identical responses.

Opamps are good for audio if they have SR higher or equal 50V/us. Still, there is a BIG difference between the completely same 2 opamps of the same family, one with 55V/us and second (decompensateed) with 135V/us. It is audible, believe me or not.
Have a link to the details? This contradicts the result I just got (no audible difference between 22V/us and 100+V/us) so I'd like to understand what you're looking at which I'm not.
 
Have a link to the details? This contradicts the result I just got (no audible difference between 22V/us and 100+V/us) so I'd like to understand what you're looking at which I'm not.

PMA I believe mentioned this on the JC Blowtorch thread where I asked him for more details. However, they were not forthcoming. The parts in question are the OPA627 and OPA637 - when listening to them it would be prudent to know what gain they were running at (OPA637 isn't stable below noise gains of 5).

If >50V/uS is 'good' then obviously the OPA627 (55V/uS) is not 'good enough'.
 
From a quick look at the 627 and 637 specs my bet would be the differentiator's GBP and not slew rate. Full power bandwidth is below 100Hz for both so the 637's 10dB of additional excess loop gain relative to the 627 should be relatively noticeable. I haven't worked with the 6x7, but that's in line with the results I've gotten with OPAx134 and LME497x0 parts as well as the 49713 and 49990.
 
Yes, that was my tentative conclusion too. The difference in distortion would be fairly easily measurable - the OPA627 needing to run at +14dB for an apples-apples comparison. Pity Samuel Groner didn't test the 637, his plots for the 627 make interesting reading. He did test the 827 though and that shows slightly better HF distortion than the 627 even though its slew rate is around half that of the 627. Better GBP though.
 
Here is some logic around full power bandwidth specs-
The best available distribution media for audio today are based on 176.4 and 192 KHz sampling rates. These are capable of 100 KHz full output response. As such its seems quite valid to expect 100 KHz full power bandwidth.

I believe Parasound's HALO A21 is quite an acclaimed amplifier but I'm confused over its specs.

HALO%20A21%20SPECS.jpg


Power Bandwidth:
5Hz - 100kHz, +0/-3dB at 1W
Would this not be generally termed Frequency Response?

Slew Rate:
130V/microsec

My understanding of Slew Rate is that it is a large signal parameter. With a SR of 130V/us, does it mean Full Power Bandwidth of 100kHz at 400W RMS into 4 ohms? Or in this instance, it only applies to 5Hz-100kHz at 1W.
 
Would this not be generally termed Frequency Response?
Thats what I'd call it, but marketing guys and engineers rarely agree on terminology. 😉

With a SR of 130V/us, does it mean Full Power Bandwidth of 100kHz at 400W RMS into 4 ohms?
Parasound claims 130V/us for my A23 too. It's a good amp---I like the A51 and A52 as well, though the performance delta relative to the A23 is small and the bang for the buck lower in my opinion---but the Audio Precision says its slew rate is more like 20V/us. In the admittedly somewhat unlikely chance John's still paying attention to this thread perhaps he can illuminate us on the conditions under which 130V/us occur.
 
Many of these amps will fry their output Zobel network at 100 KHz full power. The amp can deliver but the 1W resistor will turn to smoke (I'm good at this . . ) meaning they can't really spec it that way. But the 1W measurement is not a power bandwidth measurement.
 
Is that what they are using, a 1W for the Zobel? I understand that under normal use, it won't smoke since there's very little musical power at 100kHz but if an amp is rated to produce full power at 100kHz, the resistor should be rated for power testing at that frequency without smoking.

Parasound claims 130V/us for my A23 too. It's a good amp---I like the A51 and A52 as well, though the performance delta relative to the A23 is small and the bang for the buck lower in my opinion---but the Audio Precision says its slew rate is more like 20V/us.

I have not heard the Halos but John, being a highly accomplished designer, would know exactly what he doing. Judging from the specs, it appears that he went for sonics first instead of setting out to achieve some targetted specs. Nothing wrong with that approach. That's the way I design my amps too.
 
So, I have a pair of Crown D-75A power amps. They are rated for 40w into 8 ohms. Crown says they limit the slew rate to 6V/us on purpose to limit RFI. The math says that minimum slew rate for this amp should be 6 for Vrms @ 50kHz, but about 8 for Vpp (17.8 vs. about 25 volts). The inputs have an RC filter set at 723Khz, and there's extensive RF bypassing in the circuit. The opamp in place is a MC33079 (7V/us). I was wondering if I should replace the op amp with something hotter, like the LME49740 (20V/us). If so, what other precautions should I take to insure stability? I'm operating these amps in a semi-rural home environment with not much RFI as far as I know.
 
Status
Not open for further replies.