makers of "fast speakers" with specified rise time of less than 15 µsec specify - in case of active speaker- the driving amp should be as fast as 100 Volts/µsec.
Why that? Every acoustical musical instrument is a "lumped" spring-mass system. In terms of psychoacoustics "the brain" identifies these by the rise time of the respective harmonics . It also makes an assessment about the physical size of the instrument by these rise times. That is well researched. To "the brain" the rise time of the fundamental is less significant than that of the higher harmonics. S how fast are musical instruments? With the exception of particular percussion , and piccolo flute, very slow. Seconds ( organ) to some 50 µsec.
Typical however is 50- 300 msec ( strings). So what are "fast amps" good for ? Compact disc cut off at 20 kHz ( to avoid aliasing sampling theorem) 20 khz has cycle time 1/2*10^4= 50 µsec and the first quarter has 12.5 µsec.
Thus we have for ex. 100 Volts in 12.5 µsec but not any such fast musical instrument. What gives? How fast should an amp be?
Why that? Every acoustical musical instrument is a "lumped" spring-mass system. In terms of psychoacoustics "the brain" identifies these by the rise time of the respective harmonics . It also makes an assessment about the physical size of the instrument by these rise times. That is well researched. To "the brain" the rise time of the fundamental is less significant than that of the higher harmonics. S how fast are musical instruments? With the exception of particular percussion , and piccolo flute, very slow. Seconds ( organ) to some 50 µsec.
Typical however is 50- 300 msec ( strings). So what are "fast amps" good for ? Compact disc cut off at 20 kHz ( to avoid aliasing sampling theorem) 20 khz has cycle time 1/2*10^4= 50 µsec and the first quarter has 12.5 µsec.
Thus we have for ex. 100 Volts in 12.5 µsec but not any such fast musical instrument. What gives? How fast should an amp be?
I have the impression that you mix up the rise time of the envelope of the signal with the rise time of the signal itself, but maybe I simply don't understand you.
A long time ago some Finnish researchers measured what slew rate was needed to play back gramophone records at maximum volume without slewing. The worst signal they found had a similar slew rate requirement as an 8 kHz sine wave, it came from a record called "Deutsche Marschmuzik - Einzug der Gladiatoren" (which, judging by the title, I wouldn't want to hear at all, and certainly not at maximum volume).
Anyway, the slew rate needed for a sine wave is 2 pi f V, where f is the frequency and V is the peak output voltage, so for playing back "Deutsche Marschmuzik - Einzug der Gladiatoren" at 100 W into 8 ohm, you would need 2 pi * 8 kHz * sqrt(2) * sqrt(100 W * 8 ohm) ~= 2.01 V/us.
Taking a considerable safety margin because there may be music around that requires higher slew rates than the Finnish researchers found, and because many amplifiers have increased distortion before running into slew rate limiting, 20 V/us seems like a reasonable target for a 100 W amplifier.
A long time ago some Finnish researchers measured what slew rate was needed to play back gramophone records at maximum volume without slewing. The worst signal they found had a similar slew rate requirement as an 8 kHz sine wave, it came from a record called "Deutsche Marschmuzik - Einzug der Gladiatoren" (which, judging by the title, I wouldn't want to hear at all, and certainly not at maximum volume).
Anyway, the slew rate needed for a sine wave is 2 pi f V, where f is the frequency and V is the peak output voltage, so for playing back "Deutsche Marschmuzik - Einzug der Gladiatoren" at 100 W into 8 ohm, you would need 2 pi * 8 kHz * sqrt(2) * sqrt(100 W * 8 ohm) ~= 2.01 V/us.
Taking a considerable safety margin because there may be music around that requires higher slew rates than the Finnish researchers found, and because many amplifiers have increased distortion before running into slew rate limiting, 20 V/us seems like a reasonable target for a 100 W amplifier.
No the rise time. Because said fast speakers specify rise time. For a very simple reason. There is a standardized procedure how to measure the rise time of a mechanical device ( such as a robot arm ) but not a procedure for "slew rate". The rise time is derived from "pulse response".
There is no way to determine the "slew rate" of such a mechanical device and none for musical instrument, either. The unit of measurement is however the same in electronics for either rise time and slew rate.
Last edited:
hahfran, could you please explain what that standardized procedure is and how it gives a result of seconds for an organ, because right now I have absolutely no idea what you write about.
For an electronic device the rise time is usually defined as the time between crossing 10 % and 90 % of the final value of the step response. For the envelope of the signal of an organ the time needed to get from 10 % to 90 % would be something in the many milliseconds to seconds range. The signal itself varies much faster than its envelope.
By the way, according to Jan Didden, slew rate is originally a term related to electromechanical devices, namely to American anti-aircraft guns.
For an electronic device the rise time is usually defined as the time between crossing 10 % and 90 % of the final value of the step response. For the envelope of the signal of an organ the time needed to get from 10 % to 90 % would be something in the many milliseconds to seconds range. The signal itself varies much faster than its envelope.
By the way, according to Jan Didden, slew rate is originally a term related to electromechanical devices, namely to American anti-aircraft guns.
Maybe I misunderstand but This dosnt make sense. The rise time of a speaker just tells you it's HF cutoff. Rise time is not in seconds its in something per seconds. As in volts/second. ( degrees rotation/second for anti aircraft guns? ). For speakers Ild guess it's meters/second excursion. So how do these relate? You have to consider the speaker excursion and especially its efficiently. If it takes 5volts to drive a speaker to its full excursion and 1 volt for another (let's say for the same spl) the first speaker will need an amp with 5 times the rise time and 25 times the power which means it will have a faster rise time. ( more power means faster rise times, the volts part of v/us)
If the amps rise time ( not slew rate, which is not the same ) is fast enough to reproduce a full power 20k sine wave than its fast enough to drive any speaker. (Unless it's HF roll off is below 20k then a slower amp will do, but you won't find one, except maybe in a sub )
Rise time = freq. x power
Last edited:
It is the method of measurement "step response" . In technical acoustics the step response of e.g. a speaker is triggered with MLS signal and by integration the rise time is calculated. There is no method for slew rate. When a string is picked it is step. Same for drum beat etc. Always step response. Therefore same with amps to stay in the subject. I think the request for "fast amps" has no objective foundation. Except in recording where there is only the mic which can be very fast responding to steps.
I'm going to make some assumptions, please correct me when they are wrong.
You write something about maximum-length sequences. I'll assume that these are used only because directly measuring a loudspeaker's step response causes some practical problem, such as overdriving the loudspeaker and/or getting a too poor signal to noise ratio out of the measurements. If that's correct, then we can forget about maximum-length sequences for the moment and assume that the step response is measured by some appropriate method, whatever it may be.
Second, I'll assume that rise time is defined the same as in electronics, that is, the time needed for the step response to get from 10 % to 90 %.
For a linear time-invariant system, the step response holds the same information as the magnitude and phase versus frequency characteristics, although some things may be easier to see in a plot of the step response and others in plots of the magnitude and phase versus frequency characteristics.
A well-known rule of thumb says that for a well-damped step response (without non-linear effects such as slewing), the rise time is approximately 0.35 divided by the bandwidth in Hz. That means that 15 us rise time is an indirect way to specify a bandwidth of approximately 23.33... kHz.
If you want to maintain that bandwidth, the amplifier needs to have an essentially flat response up to 23.333... kHz. On top of that, whatever signal you want to play should not drive it into or close to slew rate limiting (see post 4). I think those two things together answer the question.
Rise time is definitely measured in seconds, not in watt hertz like your equation says. The required slew rate does depend on power, on the square root of power to be precise. (Four times the power means only twice the voltage.)
All an amplifier does is multiply the input voltage at a moment in time, it doesn't know or care what that single voltage is part of. So long as it can "stay in the subject" all is well.
Slew rate of an amplifier seems well defined. Not so, it is misleading.
The SR you see in op amp datasheet for instance tells how fast goes the output when the input is over driven. That is a non linear behavior parameter. It is NOT relevant to audio because we do look for a linear behavior.
The SR that makes sense to audio is of a lower value and cannot be guess estimated with some safety margin applied to the non linear behavior SR.
The SR that matters for audio must be determined with sine signals of large amplitude and frequencies to find out the maximum signal slope the amplifier can output without any abnormal raise of THD. This SR is within the linear behavior of the amplifier, a bit more amplitude or frequency enters non linear behavior. With such SR definition there is no need for large safety margins, there is no more need for margin than what's usually chosen about amplitude head room before clipping.
The SR you see in op amp datasheet for instance tells how fast goes the output when the input is over driven. That is a non linear behavior parameter. It is NOT relevant to audio because we do look for a linear behavior.
The SR that makes sense to audio is of a lower value and cannot be guess estimated with some safety margin applied to the non linear behavior SR.
The SR that matters for audio must be determined with sine signals of large amplitude and frequencies to find out the maximum signal slope the amplifier can output without any abnormal raise of THD. This SR is within the linear behavior of the amplifier, a bit more amplitude or frequency enters non linear behavior. With such SR definition there is no need for large safety margins, there is no more need for margin than what's usually chosen about amplitude head room before clipping.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.