You have started from a wrong assumption. The spectrum produced by musical instruments goes way beyond 20KHz.
There's life above 20 kilohertz! A survey of musical instrument spectra to 102.4 kHz
If the phase margin is less than 45 degrees at 0dB gain, an excitation signal outside the audio band will excite the amplifier into instability.
What kind of signal outside the audio band ?
There are not present in the sources, limited to 20 or 22 kHz.
Noise ?
You should learn about how oscillator work.
An amplifier become oscillator if not design properly.
You can read Nyquist criteria of amplifier stability.
Yes but the recording media doesn't, except vinyl...and well theres plenty of other evils to worry about in that case!
Noise and RF ingress...yes that's something to be worried about.
Anti-aliasing may refer to any of a number of techniques to combat the problems of aliasing in a sampled signal such as a digital image or digital audio recording. Specific topics in anti-aliasing include: Anti-aliasing filter, a filter used before a signal sampler, to restrict the bandwidth of a signal such as in audio applications
I design USB oscilloscopes and unless the bandwidth is limited to about half max sampling speed I get noisy results.
I tend to leave front end op amps as fast as possible but LPF the signal digitally out of the A2D.
I design USB oscilloscopes and unless the bandwidth is limited to about half max sampling speed I get noisy results.
I tend to leave front end op amps as fast as possible but LPF the signal digitally out of the A2D.
>How this resonance could be excited (amplifier loaded with a speaker)
How about Murphy's law; your speaker just happens to present a near infinite load at the very same frequency, due to Xover design?
How about Murphy's law; your speaker just happens to present a near infinite load at the very same frequency, due to Xover design?
Noisy switch mode wall wart, a.m. broadcast radio station, police or ham radio transmitter, triac lamp dimmer...
I dont think it has much to do with Fourier and harmonic series tbh Joe.
I think it has an awful lot more to do with sub harmonics generated in the audible frequency band as a result of intermodulation caused by frequencies above audibility.
There is evidence that virgin sounds (not reproduced) contain information humans can interpret, though not hear, in the conventional sense.
Of course, external noise and RF must be suppressed, and susceptibility to them, reduced.
I can still hear quite nicely to 16kHz. My amplifiers, roll off at about 20kHz.
Even if theres 0dB ultrasonic in the signal, my ears add maybe -3dB, and the amp, another -3dB.
Even if my source could reproduce it, as well as all the RF junk it picks up in that range, the amp couldnt reproduce it at realistic levels.
So then the wise answer is..."let my amplifier BW reach to the moon, well 100kHz".
Why extend source and amplifier BW 2.5 octaves above audibility, just to amplify signals that may be -20, or -30, or -40dB or more attenuated and alone inaudible?
As far as Tape...sure there is high frequency information there, buried in the hiss, and the remnants of the ultrasonic erase signal.
Unless you're running a studio grade reel to reel at high tape speeds, not likely to have the bandwidth of CD even. (I used to have a Tascam some years ago, and always ran it double standard speed.
Then you got some HF that was half decent, with a top of the line TDK cassette. Not many albums that fit on 22mins per side though...)
Phono definitely has some ultrasonic output, if it weren't for anything else, merely because of the RIAA EQ. And sometimes, occasionally, it can be said, theres a little perceived added realism, with the right media. But you're listening to a whisper through the rustle of leaves, blown in the wind.
I think it has an awful lot more to do with sub harmonics generated in the audible frequency band as a result of intermodulation caused by frequencies above audibility.
There is evidence that virgin sounds (not reproduced) contain information humans can interpret, though not hear, in the conventional sense.
Of course, external noise and RF must be suppressed, and susceptibility to them, reduced.
I can still hear quite nicely to 16kHz. My amplifiers, roll off at about 20kHz.
Even if theres 0dB ultrasonic in the signal, my ears add maybe -3dB, and the amp, another -3dB.
Even if my source could reproduce it, as well as all the RF junk it picks up in that range, the amp couldnt reproduce it at realistic levels.
So then the wise answer is..."let my amplifier BW reach to the moon, well 100kHz".
Why extend source and amplifier BW 2.5 octaves above audibility, just to amplify signals that may be -20, or -30, or -40dB or more attenuated and alone inaudible?
As far as Tape...sure there is high frequency information there, buried in the hiss, and the remnants of the ultrasonic erase signal.
Unless you're running a studio grade reel to reel at high tape speeds, not likely to have the bandwidth of CD even. (I used to have a Tascam some years ago, and always ran it double standard speed.
Then you got some HF that was half decent, with a top of the line TDK cassette. Not many albums that fit on 22mins per side though...)
Phono definitely has some ultrasonic output, if it weren't for anything else, merely because of the RIAA EQ. And sometimes, occasionally, it can be said, theres a little perceived added realism, with the right media. But you're listening to a whisper through the rustle of leaves, blown in the wind.
Last edited:
Sorry, I kow everything about Fourier Analysis
I just ask what kind of signals could be over 20 kHz, as there is no audio signals in sources above 20 or 22 kHz (except for vinyls).
Thanks to some answers, I now think that noises (internal and external), intermod, vinyl signals etc can be over 20 kHz and trigger instability if any.
Well they often have (and need) filters, its common to have selectable roofing filters for recording setups - for instance the Neumann lathes have 8kHz/11.2kHz/15kHz roofing treble filter used for control of sibilant signals, since ultrasonic energy can be problematic for tracking on playback.
Whatever the instrument generates, the microphone and EQ may severely damp down the ultrasonic high end, even the RF-suppression filter may be kicking in before 40kHz.
For tape recording you would have a big problem with significant ultrasonic energy that came close in frequency to the bias oscillator - it would heterodyne with it and produce spurious tones.
My suspicion is that when non-band-limited sources like synthesizers appeared recording studios all routinely started filtering/EQing out ultrasonics for them, partly for mastering issues and partly for cross-talk issues within the mixing desk - hard edged square waves are the worst for crosstalk.
+1
Tape bias signal was what I meant (rather than erase signal)
I knew what I meant then realised I didn't know what I meant, then realised what I erroneously meant, was actually something else, but the point; lost on everyone, was vaguely correct.
Phono, at least, though band limited at cutting (refs above) will have a fair bit of harmonic distortion, extending its output above the limits of audibility.
Especially when tracking error, groove/stylus wear all add their contribution.
10% THD at 10 or 15kHz isnt going to be directly audible to a human, but it will generate IM products that will be the AF range.
Trying to "preserve" that along with the actual signal, isnt going to add realism, on the contrary, it needs to be filtered out of band in order to preserve signal integrity.
Tape bias signal was what I meant (rather than erase signal)
I knew what I meant then realised I didn't know what I meant, then realised what I erroneously meant, was actually something else, but the point; lost on everyone, was vaguely correct.
Phono, at least, though band limited at cutting (refs above) will have a fair bit of harmonic distortion, extending its output above the limits of audibility.
Especially when tracking error, groove/stylus wear all add their contribution.
10% THD at 10 or 15kHz isnt going to be directly audible to a human, but it will generate IM products that will be the AF range.
Trying to "preserve" that along with the actual signal, isnt going to add realism, on the contrary, it needs to be filtered out of band in order to preserve signal integrity.
Last edited:
10% sounds like a tube amp
IMF which other frequency are you mixing with?
Is it desireble to hear these products?
If I recall correctly,
If f1=10000hz, f2 = 20000Hz then the IMf = 1/2Hz amplitude modulation, (or the inverse case, 2Hz, if I have the fraction upside-down) which depending on the modulation depth should be audible to a listener.
This, I guess, should be preserved, but is very difficult to do so, without increasing susceptibility to RF.
If f1=10000hz, f2 = 20000Hz then the IMf = 1/2Hz amplitude modulation, (or the inverse case, 2Hz, if I have the fraction upside-down) which depending on the modulation depth should be audible to a listener.
This, I guess, should be preserved, but is very difficult to do so, without increasing susceptibility to RF.
Of course a nice one octave harmonic gives a sound a pleasant IMD side effect, the beat frequency is low, and creates a mild tremelo effect.
The IMD caused by interference, DAC filtering and any other source is dissonant and more objectionable, and absolutely needs to be reduced to a level that isnt noticable.
So while a ultrasonic harmonious harmonic being faithfully reproduced from a recording capable of recording it, is desirable for the ultimate in realism - the possibility that something like for e.g. the bias signal of tape playback causing intermodulation in the reproduction of HF, that is unwanted.
Like it all, it's a compromise, and a gamble what you think you will gain, or what you find most objectionable, potentially missing that 1ppm of the original sound, or gaining 1ppm of something artificial.
The IMD caused by interference, DAC filtering and any other source is dissonant and more objectionable, and absolutely needs to be reduced to a level that isnt noticable.
So while a ultrasonic harmonious harmonic being faithfully reproduced from a recording capable of recording it, is desirable for the ultimate in realism - the possibility that something like for e.g. the bias signal of tape playback causing intermodulation in the reproduction of HF, that is unwanted.
Like it all, it's a compromise, and a gamble what you think you will gain, or what you find most objectionable, potentially missing that 1ppm of the original sound, or gaining 1ppm of something artificial.
Sorry, no beat frequency at all for harmonics.
Try to hear the difference between a squarewave and a sine at 6 to 10kHz. The sqare has a 3rd harmonic which has also 1/3 of the amplitude of the fundamental, the 5th has 1/5 and so on. That would be 43% of distortion with only unpleasant harmonics.
We can not distinguish between the two.
Try to hear the difference between a squarewave and a sine at 6 to 10kHz. The sqare has a 3rd harmonic which has also 1/3 of the amplitude of the fundamental, the 5th has 1/5 and so on. That would be 43% of distortion with only unpleasant harmonics.
We can not distinguish between the two.