Usually the maximum full power 20kHz signal slope (at the zero crossings)
is considered to be the maximum slew rate needed, plus a safety factor.
For 100W into 8R, the output voltage is 40V peak, so the output waveform
at 20kHz at full power is: 40V x sin(2Pi x 20kHz x t).
The maximum signal slope is then the derivative wrt time, with the magnitude
of the resulting cos function set to unity, the maximum value (for example at t=0).
d/dt {40V x sin(2Pi x 20kHz x t)} = 40V x 2pi x 20kHz x cos(2Pi x 20kHz x t)/sec
At t=0, then 40V x 2pi x 20kHz/sec = 5V/uS. For a safety factor, perhaps use 10VuS.
Slew rate limiting is somewhat similar to clipping, except that it happens on the fastest
changing part of the waveform, instead of on the peaks.
Last edited:
Jon, it's a bit more subtle. Slew rate limits the high frequency response for large signals only.
For small signals, the slew rate usually doesn't come into play so doesn't limit the high frequency response.
It is not unusual for an amp to have a bandwidth spec of say 1MHz, and that usually implies small signal. Then if you drive it with a large signal you find that practically it is limited to just a few 100 kHz.
Some honest opamp manufacturers for instance report both on their data sheets. Others count on the lack of knowledge in their customers and only report 'bandwidth'. ;-)
Jan
Up to a point. If you calculate the required slew rate for full power at 20kHz, and double or triple it to be safe, you're at the point of diminishing returns. You can double it again and again for bragging rights, but your dollars are probably better spend elsewhere. Like buying music ;-)
Jan
Jan
Once your limit is 10% above the max signal slew rate you are 100% happy, its a limit that kicks in, without ill-effect until it kicks in. And its set by passives so isn't affected by device variation. Its just like clipping, 10% below clipping and its a non-issue, 1% above and its bad news.
What's then important is switching distortion, which can be major contributer to cross-over distortion at HF (especially with CFP output stage)
What's then important is switching distortion, which can be major contributer to cross-over distortion at HF (especially with CFP output stage)
SR simply tells you what the maximum rate of change of the amplifier output is assuming that the input signal rise time is >> faster i.e. it is not the limiting factor for the maximum rate of change.
On a VFA, this will be related to how much current is available from the LTP current source to charge/discharge the 2nd stage comp cap - See Bruno Putzey's 'The F word' for a good summary and explanation (although the main thrust of the article is to talk about why feedback is not a bad thing). On a CFA, the mechanism is a bit different because of the current-on-demand behaviour.
A good rule of thumb (Robert Cordell) is that the slew rate should be 1V/us per peak output volt. So if your amp's peak output voltage is 10V, a SR of 10V/us is about right. As Jan says, you can double or triple it (it costs almost nothing to do BTW unless you go over the top) but you will add nothing to the perceived performance.
If the SR too low (from SR = 2*Pi*f*Vpk) you will get SID and any input signal above the maximum frequency shown in the formula would triangulate. On modern VFA amps, you won't see SR limiting below 100 or 200 kHz at full power. So for a 100 Watt amp delivering 100 kHz at full power you get a minimum SR requirement of 25V/us - our rule of thumb of 1V/us per peak output voltage (40V peak for 100 W amp) gives us plenty of safety margin since we would only start to slew at 160kHz.
SR is not the same as small signal (here I talk about small voltage i.e. 1-2 V peak-peak) rise/fall time. Usually this is much faster.
On a VFA, this will be related to how much current is available from the LTP current source to charge/discharge the 2nd stage comp cap - See Bruno Putzey's 'The F word' for a good summary and explanation (although the main thrust of the article is to talk about why feedback is not a bad thing). On a CFA, the mechanism is a bit different because of the current-on-demand behaviour.
A good rule of thumb (Robert Cordell) is that the slew rate should be 1V/us per peak output volt. So if your amp's peak output voltage is 10V, a SR of 10V/us is about right. As Jan says, you can double or triple it (it costs almost nothing to do BTW unless you go over the top) but you will add nothing to the perceived performance.
If the SR too low (from SR = 2*Pi*f*Vpk) you will get SID and any input signal above the maximum frequency shown in the formula would triangulate. On modern VFA amps, you won't see SR limiting below 100 or 200 kHz at full power. So for a 100 Watt amp delivering 100 kHz at full power you get a minimum SR requirement of 25V/us - our rule of thumb of 1V/us per peak output voltage (40V peak for 100 W amp) gives us plenty of safety margin since we would only start to slew at 160kHz.
SR is not the same as small signal (here I talk about small voltage i.e. 1-2 V peak-peak) rise/fall time. Usually this is much faster.
Yes. The slew rate of a 1V signal is 1/40th of the signal at 40V. So at low levels, slew rate doesn't come into the game so your amp has a very wide freq band.
Bandwidth is generally understood to be a small signal parameter. Often you see both of them specified in a data sheet, small-signal and large-signal bandwidth. The difference can be significant.
Jan
Bandwidth is generally understood to be a small signal parameter. Often you see both of them specified in a data sheet, small-signal and large-signal bandwidth. The difference can be significant.
Jan
At lower frequencies, the signal has corresponding lower slew rate of course. Also at lower levels it has lower slew rate. So when you get closer to slew rate limiting to the point that it starts to cause distortions, that will be when the frequency or level or both goes up.
What does slew rate really mean? It is the change of voltage in time. That is why it is measured in Volts per Second, or Volts per Microsecond. Look at a sine wave going through zero. When is that steepest (= highest slew rate, most voltage per second change)? It gets steeper when freq goes up, and it gets steeper when level goes up.
Jan
What does slew rate really mean? It is the change of voltage in time. That is why it is measured in Volts per Second, or Volts per Microsecond. Look at a sine wave going through zero. When is that steepest (= highest slew rate, most voltage per second change)? It gets steeper when freq goes up, and it gets steeper when level goes up.
Jan
Last edited:
I suspect that you are not looking at this in the best way!
During normal signal conditions (amp not overdriven or bandwidth exceeded) the slew rate limit is not experienced so has no meaning, so no effect.
If you want an amplifier not to slew rate limit on any input signal that's smaller than the low-frequency clipping limit, you need to consider what happens with square waves. As it turns out, you then need about twice as high a slew rate limit as you would expect based on the amplifier's bandwidth and peak output voltage.
Suppose an amplifier has a peak output voltage Vpeak, a small-signal gain A and a small-signal bandwidth fBW. Assuming first-order behaviour, the time constant of its response is tau = 1/(2 pi fBW).
When everything remains linear, immediately after a step from -Vpeak/A to +Vpeak/A at the input, the output voltage will start increasing with a rate of 2 Vpeak/tau = 2 Vpeak 2 pi fBW = 4 pi fBW Vpeak, so twice what you would have with a sine wave with frequency fBW and amplitude Vpeak at the output.
Whether this is at all relevant for music is another matter. A long time ago Finish researchers measured the rate of change of signals on records and found that all of them could be reproduced at full volume without slewing by any amplifier that could handle an 8 kHz sine wave. The music with the highest rate of change was called Deutsche Marschmuzik: Einzug der Gladiatoren, which I certainly wouldn't want to hear at full volume.
By the way, whether distortion increases substantially just below the slew rate limit depends a lot on the design of the amplifier. With some local feedback in the stages before the dominant pole (that is, the input stage in a traditional design), you can get quite close to the limit without any significant distortion degradation. With a soft clipping input stage, it's a different story.
Suppose an amplifier has a peak output voltage Vpeak, a small-signal gain A and a small-signal bandwidth fBW. Assuming first-order behaviour, the time constant of its response is tau = 1/(2 pi fBW).
When everything remains linear, immediately after a step from -Vpeak/A to +Vpeak/A at the input, the output voltage will start increasing with a rate of 2 Vpeak/tau = 2 Vpeak 2 pi fBW = 4 pi fBW Vpeak, so twice what you would have with a sine wave with frequency fBW and amplitude Vpeak at the output.
Whether this is at all relevant for music is another matter. A long time ago Finish researchers measured the rate of change of signals on records and found that all of them could be reproduced at full volume without slewing by any amplifier that could handle an 8 kHz sine wave. The music with the highest rate of change was called Deutsche Marschmuzik: Einzug der Gladiatoren, which I certainly wouldn't want to hear at full volume.
By the way, whether distortion increases substantially just below the slew rate limit depends a lot on the design of the amplifier. With some local feedback in the stages before the dominant pole (that is, the input stage in a traditional design), you can get quite close to the limit without any significant distortion degradation. With a soft clipping input stage, it's a different story.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.