Well, you certainly aren't going to run out of bandwidth anytime soon 🙂 Personally, I try to stay with as low a GBW as I can get away with as long as the rest of the specs are workable - the faster the unit the more susceptible to parasitics and more trouble keeping it stable. Currently working on a front end for a high speed ADC and using an AD8099 - nice part but finicky. Miss the good old days of just putting in an OP27 and calling it good...
Hal
Hal
@MarcelvdG - that is what I was looking for if it had been data taken before the recording. Been trying to Google something useful but nothing so far... Or, does the article have before/after data?
Thanks for the reply, though.
Thanks for the reply, though.
No, this was all based on vinyl record playback. Well, at least you know now that German marching music is of interest.
The same may hold for gamelan music. A Japanese team used it for comparing music reproduction with and without the spectral content above 26 kHz.
If it's a microphone feed you are interested in, there could be a bat flying by the microphone.
The same may hold for gamelan music. A Japanese team used it for comparing music reproduction with and without the spectral content above 26 kHz.
If it's a microphone feed you are interested in, there could be a bat flying by the microphone.
I think I understand what you are asking... and when it's all said and done the answer is still going to be something in the vicinity of a 20Khz signal. it's the fastest signal we can hear. If there was a transient of something that had bigger power response, it would be a lower frequency, that is, slower. If there are ultrasonic components to the higher order effects of something, say a cymbal crash or the like, it doesn't matter much as it's not going to be perceived. (well, maybe the bats will hear it, as others mention...)
There's a lot of talk about leading edge transients in music, usually percussive like the hit of a drumstick, a picked guitar, some electronica, bassoon concertos, (well, maybe not that one) but all that is slower than a 20khz signal because it would have 20khz information if it did. (And yes, some does.)
Here are a couple of related acoustical papers, but you will find spectral analysis (up to 100kHz), not time analysis.
Last edited:
It may be true that our hearing is limited to 20 KHz maximum. But defining an amplifier's performance by sine waves is misleading. A cymbal or rim shot's waveform is more like a square wave than a sine. A square wave requires 10 times the sine wave bandwidth to replicate effectively; so if we need 1.88v/uSec for a sine wave, we need at least 18.8 v/uSec for a square wave.
No. No combination of lower audio frequencies that does not clip has a higher slew rate than a ~20KHz sine wave. For example, if the wave is 50% 20K and 50% 10K then the peak slew rate is 1/2+1/4 = 3/4 that of 100% 20K. However, simply avoiding slew limiting may not be good enough since high slew signals tend to overdrive the IPS, resulting in poorer THD, albeit not slew limited. A rational requirement is a 6dB margin, ie a slew rate of 2x (2π *20K * Vp). You can also argue that clipping transients is not unusual, and the slew rate should avoid slew limiting of signals that are otherwise highly distorted.
I'm with John Curl on this one: Maximize slew rate! --- and then you don't have to worry that your low-gm input diffpair (of degenerated JFETs or BJTs) needs more than a tiny handful of millivolts of overdrive to guarantee max-available-slewing. Kill the fly with a bazooka and feel confident it is dead, dead, dead TF dead.
If the region where your input stage's dVin/dIout curve is beautifully linear extends across "V" volts, aim for a minimum slew rate of (V/0.2)*5E+6 volts per microsecond per volt of half-supply. Ask diyAudio member "Bonsai" to tutor you about The Solomon Paper if you need further technical support.
If the region where your input stage's dVin/dIout curve is beautifully linear extends across "V" volts, aim for a minimum slew rate of (V/0.2)*5E+6 volts per microsecond per volt of half-supply. Ask diyAudio member "Bonsai" to tutor you about The Solomon Paper if you need further technical support.
You think you are gonna have to have enough slew rate for 30 kHz components or something?
@dotneck335 - Thanks! Finally someone that actually understands what I am asking - and evidently more importantly, not asking. Probably my mistake was asking in an audio related forum where, no matter how much I state otherwise, the assumption is that actual audio gear - along with it's imperfections and compromises - must be involved somehow; rather, my interest is from a signal processing viewpoint. As you point out, a portion of the waveform can easily have a much higher DV/DT than that of a pure sine wave - I was just hoping to find the real world examples and the relevant data of such a case. It would of course not be an audible artifact but, then, that was never part of the consideration or the question.
Hal
Hal
Unfortunately, designing for maximum slew rate leads to stability and other problems, so some discretion is wise. But today with 40MHz BJT and MOSFET power transistors, a more than adequate slew rate should be easy. Poor slew happens when the designer puts too much OLG in the circuit, forcing draconian amounts of compensation. This is why amps with an op-amp IPS are often unstable and have a poor slew rate.
The Mark Alexander power amp using an opamp ips is one example of not having a slew rate issue. One day I’ll give it a try using some latfets
It depends on bandwidth. Tom's numbers are for a brickwall response eg red book CD.
If the roll-off is gentler, eg -3dB 20kHz 6dB/8ve, maximum slew happens for a peak to peak square wave and is twice Tom's numbers. Anything higher will overload and the audible concern isn't slew but overload recovery.
You only need to worry when doing live recording. The usual bandwidth limiter is the microphone and it sets max slew until your A/D and its anti-aliasing filters.
'Real life' max slew are with clap sticks and drum rim shots. But these are incredibly difficult to record without overloading, microphone & A/D. The good news is that you usually can't hear these overload on a good microphone & recorder. 🙂
How does one measure the rate of change of acoustical signals without some form of audio gear? Even if you would use a 19th-century purely mechanical set-up for the measurement, it would be imperfect equipment processing the acoustical signal.
My bet for high dp/dt values is still on the bats.
in real life, probably!
So it is more a question of how audio and music is defined.
Is it still "music" or "audio" (latin "I hear") if we can't hear it?
a 20 kHz square wave recorded on vinyl, cd or as we percieve it (if we're young enough), is identical to a sine.
Slew rate is an artifact of a capacitor being charged by a constant current source (as in solid-state amplifiers). Mechanical systems can also slew. Outside of those, one normally considers bandwidth.
Ed
Ed
So, did a simple sym. A single pole RC lowpass with 20 kHz -3dB corner. Not close to a "brick wall". This is a reasonable worst case frequency roll off whether it be the ears or the program material. In reality there are many stacked poles in the 20 kHz region.
Results with a 10nS rising step to 1V gives 121mV/uS slew rate output. As this is a linear circuit this result can be multiplied by the actual input signal level to give needed slew rate maximum.
For 100w peak power into 8 Ohms, the signal peak would need to be 28.3 volts. Slew max slew rate needed would be 3.4 V/uS.
Results with a 10nS rising step to 1V gives 121mV/uS slew rate output. As this is a linear circuit this result can be multiplied by the actual input signal level to give needed slew rate maximum.
For 100w peak power into 8 Ohms, the signal peak would need to be 28.3 volts. Slew max slew rate needed would be 3.4 V/uS.