I want to make at least a small attempt to support having high slew rates in audio systems. I look at this from a slightly different perspective. I come from the ultrafast optical laser community where we make 10 femtosecond or shorter pulses routinely. To do this we need two important ingredients: Bandwidth (spectral width) is required to make a short pulse in the fourier limit (eg 30 nm coherent bandwidth at 800 nm makes a 35 or so femtosecond pulse). This is easy in a gain media like Ti:Sapphire that can support 100 nm of bandwidth. However, if the pulse is chirped meaning the colors are separated in time (dispersed), the optical pulse can be any length in time in theory. This chirp is used in chirped pulse amplification (that won a Nobel prize recently but was invented for radar decades earlier) with so we can reduce the peak intensity of laser pulse by making it longer in time as it gets more energy and then compressing the pulse by de-chirping it to put the colors back to the same phase or time. The phase of the various colors in the pulse needs to be flat (all colors come at the same time) ideally to make the shortest laser pulse. Thus, to make a short pulse at need control of bandwidth and phase.
A system that has a high slew rate in general can produce and preserve an impulse which similarly means it should preserve the phase of the spectral components of a signal. In other words, if properly designed, a high slew rate system will preserve the spectrum vs time of a signal as it propagates. Why does this matter for audio? Dispersion causes the sizzle of sharp transient peaks like a snare drum or cymbal crash or the initial strike of a piano wire to smear out in time. Does this say anything about distortion? No. Does it means is has to sound better? No. I just think this is not really appreciated widely and I do know from the optical world that dispersion can lead to a loss of fidelity and information. There may well be some Faustian bargains to make a circuit with a crazy slew rate and bandwidths but given the choice between two systems I know nothing about I would be more tempted to try the one with the wider bandwidth and higher slew rate. My $0.02.
This is also incidentally why I am horrified to design speaker crossovers. Putting poles on purpose near frequencies that have signal power will naturally cause dispersion if left uncompensated.
A system that has a high slew rate in general can produce and preserve an impulse which similarly means it should preserve the phase of the spectral components of a signal. In other words, if properly designed, a high slew rate system will preserve the spectrum vs time of a signal as it propagates. Why does this matter for audio? Dispersion causes the sizzle of sharp transient peaks like a snare drum or cymbal crash or the initial strike of a piano wire to smear out in time. Does this say anything about distortion? No. Does it means is has to sound better? No. I just think this is not really appreciated widely and I do know from the optical world that dispersion can lead to a loss of fidelity and information. There may well be some Faustian bargains to make a circuit with a crazy slew rate and bandwidths but given the choice between two systems I know nothing about I would be more tempted to try the one with the wider bandwidth and higher slew rate. My $0.02.
This is also incidentally why I am horrified to design speaker crossovers. Putting poles on purpose near frequencies that have signal power will naturally cause dispersion if left uncompensated.
Last edited:
This bandwidth triggered phase error is negligible compared with what crossovers and drivers will do
I just think this is not really appreciated widely and I do know from the optical world that dispersion can lead to a loss of fidelity and information.
Fair enough, but then you have to design for a flat frequency response and a flat group delay, both measures of the small-signal behaviour. The slew rate limit just needs to be high enough to process the steepest signals without internal clipping or excessive distortion.
Slew rate is an indicator of the response time of the system to changes in input.
Actually, even at insufficient slew rates, the system starts responding immediately to changes in the input, albeit in an unexpected manner (tri-wave vs sine for example).
This is the correct wayI designed my amplifier such that it can handle full-power square waves without slew-rate limiting.
The active range of the amplifying stage (current limit(s) and voltage limit(s)) does not depend on rise/fall time
Slew rate determines how fast the amp responds to nonlinearities and then correct them. It is why FETs are better outputs.
Not saying that, but if one specifies 'a square wave' without further data, it's a meaningless statement.The active range of the amplifying stage (current limit(s) and voltage limit(s)) does not depend on rise/fall time
I can specify a 1kHz square wave source with 50us rise- and fall time and pretty much any amp will handle that.
Now specify 100ns rise/fall and you're talking.
Jan
Completely wrong. You guys still don't know what slew rate is?Slew rate determines how fast the amp responds to nonlinearities and then correct them. It is why FETs are better outputs.
Google is your friend.
Edit: actually, even if you know nothing about electronics, just the English words should give you a clue.
Slew rate - the rate at which something slews in amplitude per time unit. Plain English.
If you talk about an amp output, obviously slew rate is how fast the output voltage slews per time unit. Like volts/second, or in better manageable units, volts/usec.
Or you want to talk air defence guns? Slew rate is the rate at which the gun moves from one postion to another in degrees per second or in better manageable units, mils per second where 6400 mils is 360 degrees. The US Navy found out the importance of high slew rate the hard way at Pearl Harbor.
Jeez.
Jan
Last edited:
We don't know the square wave rise/fall times so this statement is empty.
Jan
You can design amplifiers such that they.don't go into slew rate limiting, even with a zero risetime square wave at the input. You then either have to design all stages before the dominant pole for a peak input voltage equal to the peak-peak value of the square wave (so twice its peak value for a symmetrical square wave), or you have to apply some passive low-pass filtering between the input and the first stage.
- Home
- Amplifiers
- Chip Amps
- Slew rate myths floating around the internet