I am using Audio Precision 2700 to test some test vectors.

I have connected resistor banks as class - d power amp's load, and tap from this resistor using a XLR cable to the AP.

When my power amp is powered up, but no song is playing(song is paused, not muted) i checked fft spectrum of the input channel , i get the following spectrum

enter image description here

You can see the three peaks in frequency above 4k. (Likely, the 3 peaks in frontier band upto 200 hz). Which causes this harmonics?

The test vector is playing at 48khz. Note : There's a passive filter after the power amp's output having cut off at 30khz.

Please point me some insight. Does the passive filter aliases the input signal, so that i get undesired signal at folded frequency?[This is my initial doubt, because this is what i believe should be the cause, if it's not then some one please tell me why this couldn't be the problem]

Thanks folks in advance...

  • 7
    $\begingroup$ A Class-D amplifier uses pulse-width modulation or pulse density modulation internally. When a song (digital signal) is muted, there still can be some signal being generated as input to the power amplifier because some PWM schemes do not have an exact representation of $0$; with $256$ levels, say, arranged symmetrically about $0$, there is no exact representation of $0$, and you get very low-level noise because of $\pm \epsilon$ is used to represent $0$. Note that $-110$ dB is great SNR for any power amp! $\endgroup$ Commented Oct 27, 2014 at 15:26
  • 2
    $\begingroup$ Your passive filter is an analog filter (as opposed to digital) so it is not causing any aliasing. Aliasing is strictly a digital phenomenon. $\endgroup$
    – Jim Clay
    Commented Oct 27, 2014 at 15:41
  • $\begingroup$ Could you tell us more about your setup? Which device generates the input signal sent to your power amp (is that the AP or something else)? What is the switching frequency of your amp? $\endgroup$ Commented Oct 27, 2014 at 18:34
  • $\begingroup$ I can think of 3 explanations for your measurement: noise floor of the device generating the input signal (if it is not the AP), noise floor of your class-D amp (all kinds of grounding/layout issues could explain that it is picking up noise from the PSU or EMI); or high frequency noise from your amp (near its switching frequency) not fully attenuated by your passive filter and not being adequately attenuated by the band-limiting filter on the AP inputs (unlikely). $\endgroup$ Commented Oct 27, 2014 at 18:35
  • $\begingroup$ @pichenettes, The test tone is played back from test device which has the power amp discussed here. Switching frrequency of popwer amp is 384 Khz $\endgroup$
    – nmxprime
    Commented Oct 28, 2014 at 12:39

2 Answers 2


If your class-D amplifier's switching frequency is 5 kHz above a multiple of the sample rate of your audio analyzer, then the aliases of the attenuated harmonics of the switching frequency will march upwards in frequency by 5 kHz across (and around) the spectrum displayed by the analyzer, until they disappear below the noise floor of the system and analyzer.

  • $\begingroup$ That was one of my hypotheses. I looked at the datasheet of the AP2700 (ap.com/download/file/183), and I'm not even sure the analyzer is actually always sampling the input audio signal (the spectrum could very well be measured by other "esoteric" methods - bandpass filtering or sampling for example). But when it samples (p. 22), the alias rejection is above 115dB. In any case the analyzer accepts fixed-frequency optional filter cards which should help cleaning this up. $\endgroup$ Commented Oct 27, 2014 at 20:22
  • $\begingroup$ @pichenettes, using optional filter cards that comes with AP doesn't help much. It just floor down overall noise floor by 10 dB $\endgroup$
    – nmxprime
    Commented Oct 28, 2014 at 12:42
  • $\begingroup$ @hotpaw2, The switching freq is 384 KHz, Then, 384 KHz / 48KHz(Sample rate of test tone played) is exactly 8, also above 5KHz. If as you said the aliases of the attenuated harmonics of the switching frequency will march upwards in frequency by 5 kHz across (and around) the spectrum displayed by the analyzer, i presume i had it to see around 8 KHz, is it so? $\endgroup$
    – nmxprime
    Commented Oct 28, 2014 at 12:45
  • $\begingroup$ @pichenettes, Consider (actually)The analyzer is not always sampling, if so, what about the following case . The tone is 50 Hz, and it's complete one period is not covered in the AP's single sampling duration(There's no time sync), i.e, it's sample window. Now is harmonics like these are aliases produced by the inadequate sampling of AP? $\endgroup$
    – nmxprime
    Commented Oct 28, 2014 at 12:49

Like the 50 Hz mains hum and its harmonics, the easiest explanation for the 5 kHz harmonic interference is a nearby switching power supply, like a wall wart. Try disconnecting appliances from mains, turning off monitors, moving things around, and altering your grounding scheme or altering the capacitance of things by touching them to reduce interference.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.