Hi mchambin,
you are posing a very clear question in post #1, and I admit that I did not read all posts until here.
Are you aware of any results relating to the real-world requirements to an amp, i.e. when working in a noisy (power supplies, non-ideal cables, high-frequency ground loops, non-ideal filtering in DAC output, ...) environment?
I'm wondering why some people requesting high slew rate, especially from the CFA community, are that content with the sound of their amps, although in theory the high "voltage swing velocities" are (perhaps?) unnecessary.
Might it be that this cannot be explained when assuming well-engineered or even idealized environments?
Kind regards,
Matthias
you are posing a very clear question in post #1, and I admit that I did not read all posts until here.
Are you aware of any results relating to the real-world requirements to an amp, i.e. when working in a noisy (power supplies, non-ideal cables, high-frequency ground loops, non-ideal filtering in DAC output, ...) environment?
I'm wondering why some people requesting high slew rate, especially from the CFA community, are that content with the sound of their amps, although in theory the high "voltage swing velocities" are (perhaps?) unnecessary.
Might it be that this cannot be explained when assuming well-engineered or even idealized environments?
Kind regards,
Matthias
So do you know results showing that disturbances from "outside" should or should not influence slew-rate targets in reasonable engineering?
Matthias
So an amp that can barely reproduce the audio spectrum, has an equally fast error correction with an amp that extends to MΗΖ, and the only difference is that the former has a large phase shift. Sorry, that doesn't make any sense.
The term "response" says it all. Everything else equal, a high frequency response amp is able to quickly respond and correct induced distortions in the audio frequencies, and thus reach a higher level of faithful reproduction. That is a fact.
Therefore, phase shift caused by low bandwidth, has the same effect as delay in error correction, except the low-pass filtering of the delayed signal, whether you measure it in degrees for a given frequency or in microseconds.
Even 10μs is still a huge difference vs a high bandwidth audio amp that can respond down to a few tens or hundreds of nanoseconds.
A power bandwidth of 20kHz does not equate to a delay of 50us. You may also be confusing open loop bandwidth with closed loop bandwidth.MrMagic said:
Provided that distortion does not rise as you approach the slew rate limit, and you only listen to real music, then you only need a power bandwidth of a few kHz (I can't remember what the actual figure is). Almost always distortion does rise as you approach the slew rate limit, so you need a margin. How much margin depends on amplifier circuit details.mrchambin said:
Yes, 50μs was too much -of course it depends on the circuit design -I've seen a single stage causing a group delay of a couple of microseconds with a gain of 1.
With just a few KHZ of power bandwidth, most likely it will cause a drop on the magnitude of high frequency content eg on a music pass with intense cymbals, or the string section of an orchestra. It would be too good-to-be-true to leave them intact.
Unless we don't talk, nor care about Hi-Fi -oh wait, we don't, this conversation is from the 30's
No, this conversation is about hi-fi in the 2010's. I seem to recall seeing a figure of 2.2kHz as the power bandwidth needed for reproducing music from an MM cartridge. A bit more may be needed for digital sources. Thus a power bandwidth of 20kHz already contains a significant margin for error.
If you can make an amp with just 20Khz of native power bandwidth that will have 1-2ppm of distortion @20Khz, then yes, but you can't, it's impossible by definition.
So for those of us who want the best possible because we can have it for about the same amount of money, we'll continue looking at the best and improving what we've got, for the challenge and joy of perfection.
For those who say "you don't need this", "you don't need that", I suggest them to follow their heart and dust off the grandpa's old gramophone they have hidden in the attic, and enjoy themselves, the way they always wanted.
No. The inference is that real music contains less energy at high frequencies.
It is not "impossible by definition", although it may be hard to achieve in practice. I am not aware of any reliable data to show that 1-2ppm of distortion (i.e. -114 to -120db) is required at any frequency. That is convenient, because achieving it might be difficult. Obviously you have access to transducers and recordings which go way beyond what is available to the rest of us.MrMagic said:
Who cares about distortion of a 20kHz sine?
The second harmonic is 40khz
The third harmonic is 60khz
The fourth harmonic is 80khz
...guess what? Nobody can hear the difference between 20kHz sinewave and 20khz rectanctle. period.
Certainly you can aim at bandwiths of 100khz or more.
This might be fun, but for serious audio applications it is simple overengineering: because we can....
The second harmonic is 40khz
The third harmonic is 60khz
The fourth harmonic is 80khz
...guess what? Nobody can hear the difference between 20kHz sinewave and 20khz rectanctle. period.
Certainly you can aim at bandwiths of 100khz or more.
This might be fun, but for serious audio applications it is simple overengineering: because we can....
You dont need high power high freqs to get very fast rise times. How much 20khz is there in a 1khz square wave? Before any one says that's not music, it's just to show if you add the right lower freqs together you can get very fast rise times. ( and synths do put out square waves, more or less.
Hi
I admit to not reading the whole thread, so my comment might be already covered or irrelevent... hehe
The best PAs I've seen all aim for 100kHz bandwidth. This does not mean that the raw amplifier is limited to 100kHz: rather, the "packaged" amplifier is limited usually by input filtering. <1ppm at 20kHz is definitely possible and IS DONE in modern amps, but this requires a flat phase response way past 100kHz. obviously, we do not want to amplifiy low-RF and above, and we need to keep the amp stable, so the output zobel and/or coil takes care of that. The input RC filter keeps very-high-frequency signals and noise from invading the amplifier input, so the amp never slew-limits.
It was proven back in the 1990s that music with a higher band-limit of 90-100kHz was perceived as being more natural than music that is brick-walled at 20-22kHz. There is more to sound than what our ears pick up.
In my view, all electronic distortion in a playback system is unnatural to us and must be made as small as possible. Our senses detect parts-per-billion so why should we be lazy in our audio design and accept obsolete performance?
So, related to the OP's Q: higher slew rates than required to simply not distort the maximum frequency is definitely worth striving for. It happens to be so easy with modern devices that you almost don't have to consciously target slew-rates or rise times and will end up with a fine amplifier.
I admit to not reading the whole thread, so my comment might be already covered or irrelevent... hehe
The best PAs I've seen all aim for 100kHz bandwidth. This does not mean that the raw amplifier is limited to 100kHz: rather, the "packaged" amplifier is limited usually by input filtering. <1ppm at 20kHz is definitely possible and IS DONE in modern amps, but this requires a flat phase response way past 100kHz. obviously, we do not want to amplifiy low-RF and above, and we need to keep the amp stable, so the output zobel and/or coil takes care of that. The input RC filter keeps very-high-frequency signals and noise from invading the amplifier input, so the amp never slew-limits.
It was proven back in the 1990s that music with a higher band-limit of 90-100kHz was perceived as being more natural than music that is brick-walled at 20-22kHz. There is more to sound than what our ears pick up.
In my view, all electronic distortion in a playback system is unnatural to us and must be made as small as possible. Our senses detect parts-per-billion so why should we be lazy in our audio design and accept obsolete performance?
So, related to the OP's Q: higher slew rates than required to simply not distort the maximum frequency is definitely worth striving for. It happens to be so easy with modern devices that you almost don't have to consciously target slew-rates or rise times and will end up with a fine amplifier.
Here's a low powered amp I built last year. 20 volt supply rails, 35 volt slew in 320 nanoseconds = 109V/usec = 5.4 volts / usec / volt of supply rail. (Photo was displayed earlier, as an attachment to post #61.)
The amp's design closely follows the analysis and equations found in the online Tutorial pdf document written by James E Solomon. Those same equations are printed in Bob Cordell's power amp book too. Starting from the theory, it wasn't particularly difficult to achieve 5.4V/us/V slew rate. And of course, lower slew rates, for those designers who prefer them, are even easier to achieve. Just crank down the appropriate term in the equation and follow your nose.
_
The amp's design closely follows the analysis and equations found in the online Tutorial pdf document written by James E Solomon. Those same equations are printed in Bob Cordell's power amp book too. Starting from the theory, it wasn't particularly difficult to achieve 5.4V/us/V slew rate. And of course, lower slew rates, for those designers who prefer them, are even easier to achieve. Just crank down the appropriate term in the equation and follow your nose.
_
Attachments
Last edited:
In order to control and correct harmonics beyond a given fundamental, you need an amp that at least can operate at those frequencies. Asymmetries, crossover distortion, phase distortion, all will prevent you from reaching that level of linearity.
Neither the opposite. Consider that level of ultra low distortion (+ high bandwidth and ultra low noise) as a safe margin and guarantee that the amp will be totally absent between the transducer and the source, beyond any subjective interpretations and colorful opinions about how it performs.
Add to this a reasonable cost comparable to the rest ones, and you have no reason to prefer something worse than that.
Not sure what you mean. That distortion is difficult but doable, -I have personally achieved a fraction of ppm at 1300V output pp 20Khz, but I recently redesigned the ES amp to improve other specs, and I'm already down at 3ppm and improving it.
Nope, no special access. I just have a DIY electrostatic headset that almost anyone can make (that obviously has >100 times less distortion than the best dynamic ones + extended high freq. response), and a low-latency 24bit/96Khz sound card -to be replaced soon with a 192Khz one which is common today at low cost.
But you gave me an idea: after getting that card and finishing the amp, it would be interesting to do some experiments making recordings and music enriched with high frequency timbre at the edge, or above the audio spectrum.
And making a couple of electrostatic microphones (which is easier than ES headphones) with the goal to extend to ultrasound, would be very interesting too.
You can't hear individual HF tones, but you can definitely hear the difference in a complex music pass with HF content. That's what everybody confuses.
There is no such thing as "overengineering". Such ultimate goals, make you reconsider what is considered as "given" on design practices, and you finally end up with novel techniques that improve the overall quality of the device, further evolving what we call technology.Certainly you can aim at bandwiths of 100khz or more.
This might be fun, but for serious audio applications it is simple overengineering: because we can....
But indeed, we do that because we can and because it's fun, plus the guarantee of complete transparency as I mentioned above.
@ nauta
+1
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.