What tells us more, practice, or theory?

[Amplifier dude]

Hey guys, I read the review on the Hypex MP502 amp and I was wondering, if this same measurement was conducted with the signal at 15k, would we see the exact same SNR over the amps audible bandwidth?

 

[LTig]

If the frequency response is flat (usually true) then yes.

 

[Amplifier dude]

If the frequency response is flat (usually true) then yes.

Is there any possible way different frequencies could agitate an amplifier more than others. Which in turn could create a different noise profile?

 

[mhardy6647]

 

[Amplifier dude]

I’ve been working on an amplifier. I tuned my last build up by ear, then took a measurement. But I’m being told I did things wrong. Would the SNR be worse if I fed it a 1khz signal instead?

 

[JW001]

Here is one of the factors which may have impact: the amplifier's slew rate. Generally, it is defined as du(t)/dt - the speed of signal change for which the amplifier will provide a distortionless output. For a sinewave of frequency f and amplitude A, its value is proportional to 2*pi*f*A.
A test signal of frequency 15 kHz may exceed the amplifier's slew rate, causing distortions impacting the THD chart.

 

[Amplifier dude]

Thanks. That’s what I like about ASR. Some real experts on here.

 

[Amplifier dude]

I’m kinda new with my analyzer. How do I measure to figure out the THD in the audible band with a 15khz signal? I’m using the dScope software.

 

[peng]

Here is one of the factors which may have impact: the amplifier's slew rate. Generally, it is defined as du(t)/dt - the speed of signal change for which the amplifier will provide a distortionless output. For a sinewave of frequency f and amplitude A, its value is proportional to 2*pi*f*A.
A test signal of frequency 15 kHz may exceed the amplifier's slew rate, causing distortions impacting the THD chart.

15 kHz signal should not be an issue for the NC502MP in terms of slew rate but the switching frequency may be an issue, if not filtered for the test?

 

[Amplifier dude]

15 kHz signal should not be an issue for the NC502MP in terms of slew rate but the switching frequency may be an issue, if not filtered for the test?

But you‘re talking theory right? I just meant the exact same settings Amir used, but only changing the frequency to 15k.

 

[Ken Tajalli]

I think the explanation lies with keeping standards.
Ofcourse feeding an amp 15KHz test signal may produce different distortion patterns compared to 1KHz,
BUT
Since we are comparing many many amps, we need to set a standard, for the measurements to be meaningful when compared against each other.
That standard demands 1KHz.
Me thinks, that's all there is to it.

 

[Amplifier dude]

I think the explanation lies with keeping standards.
Ofcourse feeding an amp 15KHz test signal may produce different distortion patterns compared to 1KHz,
BUT
Since we are comparing many many amps, we need to set a standard, for the measurements to be meaningful when compared against each other.
That standard demands 1KHz.
Me thinks, that's all there is to it.

I understand for standards purposes. But standards are really just advertising. To prove to the potential client that the design is proficient. But what if we want to dive deeper to get a better understanding?

 

[Ken Tajalli]

I understand for standards purposes. But standards are really just advertising. To prove to the potential client that the design is proficient. But what if we want to dive deeper to get a better understanding?

Anyone who buys hifi, purely on advertised test results, hype, graphs and numbers - deserves to be screwed!
BTW, Amir does test amps with different tones, for IMD, even a 32 tone test. One can see the results if one wants to.
But definition of SINAD is related to what frequency is used. You can change the signal, get another figure for noise and distortion, you just can't call it SINAD!
it won't be fair.

 

[Amplifier dude]

Anyone who buys hifi, purely on advertised test results, hype, graphs and numbers - deserves to be screwed!
BTW, Amir does test amps with different tones, for IMD, even a 32 tone test. One can see the results if one wants to.
But definition of SINAD is related to what frequency is used. You can change the signal, get another figure for noise and distortion, you just can't call it SINAD!
it won't be fair.

How can I figure out if a 15k signal produces more noise and distortion in the audible bandwidth?

 

[peng]

But you‘re talking theory right? I just meant the exact same settings Amir used, but only changing the frequency to 15k.

Not even that, I was only referring to someone's comment on the effects on slew rate, not SNR that I think you are asking.

For SNR measurement, I don't think 15 kHz or 1 kHz would make much difference if at all. SINAD would be diiferent obviously.

 

[Amplifier dude]

Not even that, I was only referring to someone's comment on the effects on slew rate, not SNR that I think you are asking.

For SNR measurement, I don't think 15 kHz or 1 kHz would make much difference if at all. SINAD would be diiferent obviously.

But you don’t know for sure right? Can anyone demonstrate this?

 

[tmtomh]

I understand for standards purposes. But standards are really just advertising. To prove to the potential client that the design is proficient. But what if we want to dive deeper to get a better understanding?

I would say it's not so much about theory vs practice, but rather about the practical application of theory. Along these lines, I agree strongly with @KenTajalli that there has to be a measurement standard, not only because many amps are being tested but also more deeply because the entire endeavor of relying on measurements is in turn based on using fidelity as a standard. We use fidelity as a standard for the same reason that Ken cites for using 1kHz tones as a standard: not everyone has to prefer maximum neutrality/fidelity, but it's the only sensible standard since different people have different preferences for how they want their gear to be colored/"voiced." In fact, I'm always struck by self-identified subjectivists' hostility to measurements: if you like your gear to impart a "warm," "organic," "ambient," "tube-like" or whatever kind of sound, it's only by searching for deviations from maximum fidelity, through measurements, that you can make intelligent purchasing decisions. Otherwise you're just shooting in the dark and making almost random guesses about whether an amp you buy will sound "musical" or "sterile" - and this constant carousel of random purchases, along with a health dose of confirmation bias, is in fact what many audiophiles do and what keeps segments of the industry afloat.

I mention all this because, per your last comment, of course there's nothing wrong with measuring an amp at 15kHz if you are building one and you want to see if changes to the circuit produce changes in that measurement even if those changes do not produce changes when you measure at 1kHz.

But unless or until someone provides evidence of some repeatable mechanism or variable that can produce audible differences in an amp's sound and can only be measured by using higher-frequency tones, testing with a single 15kHz tone is not necessarily or helpful for someone who's reviewing or measuring multiple, already finalized and shipping amps made by multiple people/companies.

It's just a different use case for the use of the measurements.

 

[Ken Tajalli]

How can I figure out if a 15k signal produces more noise and distortion in the audible bandwidth?

If you got the test gear, then replace the 1KHz with any frequency you like, and measure the total for noise and distortion in audio band.
Bear in mind a few things though:
- amps tend to add distortion artifacts that are higher in order, not lower. Meaning the distortions are 2x fs 3xfs and so on, so 15KHz may not produce enough distortion in audio band!
- But amps also usually have higher persentage of distortions as frequency goes up, so distortions produced by higher frequencies may be of higher persentage, compared to say 1KHz.

 

[Amplifier dude]

I would say it's not so much about theory vs practice, but rather about the practical application of theory. Along these lines, I agree strongly with @KenTajalli that there has to be a measurement standard, not only because many amps are being tested but also more deeply because the entire endeavor of relying on measurements is in turn based on using fidelity as a standard. We use fidelity as a standard for the same reason that Ken cites for using 1kHz tones as a standard: not everyone has to prefer maximum neutrality/fidelity, but it's the only sensible standard since different people have different preferences for how they want their gear to be colored/"voiced." In fact, I'm always struck by self-identified subjectivists' hostility to measurements: if you like your gear to impart a "warm," "organic," "ambient," "tube-like" or whatever kind of sound, it's only by searching for deviations from maximum fidelity, through measurements, that you can make intelligent purchasing decisions. Otherwise you're just shooting in the dark and making almost random guesses about whether an amp you buy will sound "musical" or "sterile" - and this constant carousel of random purchases, along with a health dose of confirmation bias, is in fact what many audiophiles do and what keeps segments of the industry afloat.

I mention all this because, per your last comment, of course there's nothing wrong with measuring an amp at 15kHz if you are building one and you want to see if changes to the circuit produce changes in that measurement even if those changes do not produce changes when you measure at 1kHz.

But unless or until someone provides evidence of some repeatable mechanism or variable that can produce audible differences in an amp's sound and can only be measured by using higher-frequency tones, testing with a single 15kHz tone is not necessarily or helpful for someone who's reviewing or measuring multiple, already finalized and shipping amps made by multiple people/companies.

It's just a different use case for the use of the measurements.

I hear what you’re saying for measurement standards that are used to market electronics. But I want to go deeper than that on this thread. And uncover some new knowledge deeper than the standards used to market amplifiers.

 

[Amplifier dude]

If you got the test gear, then replace the 1KHz with any frequency you like, and measure the total for noise and distortion in audio band.
Bear in mind a few things though:
- amps tend to add distortion artifacts that are higher in order, not lower. Meaning the distortions are 2x fs 3xfs and so on, so 15KHz may not produce enough distortion in audio band!
- But amps also usually have higher persentage of distortions as frequency goes up, so distortions produced by higher frequencies may be of higher persentage, compared to say 1KHz.

You mean like this:

If not, how do I measure the distortion in the audible band, with a 15k signal?