Improve EMC In Class D Amplifier Applications

[gmphadte]

For info

Improve EMC In Class D Amplifier Applications
http://www.elecdesign.com/Articles/I...rticleID=12004

Gajanan Phadte

[Pierre]

Well, interesting. Thanks for the link.

A question that I have is... it is true that typical audio analyzers provide false THD and noise results with Class-D amplifiers (I am not talking about that rubbish filterless ones, but of half-bridge modules, such as ZapPulse, Hypex, Coldamp, Tripath...)?

Perhaps real designers such as Lars, Bruno or Sergio can tell us how do they test their amplifiers' THD. The question is: is it really necessary to use a AES filter or similar, or on the other hand standard AP, HP8903, etc system can provide reliable measurements (with only their standard 30KHz filtering, for example)?

[Pano]

Originally posted by Pierre
it is true that typical audio analyzers provide false THD and noise results with Class-D amplifiers?

It is true, the ultrasonic switching noise can hurt the measurements, as most devices were not designed to cope with the switching residue.
Audio Precision, makers of the industry standard audio analyzers, have a "switching amp" passive low pass filter, the AUX-0025.

I just had a friend at a major amplifier company run 4 Tripath amps thru tests on an AP. He used the filter for all measurements, as one should for class-d amp. The results will be on my website in a few days.

[Lars Clausen]

My experience is, that for safety you should always use the AES 'brickwall' filter. If you dont you may or may not get
faulty results on your measurements.

In my own setup, which is a ESI Juli@ soundcard, and software from www.ymec.com , the systems seems to be immune to the switching residues, even if they are several Volts. Meaning the THD results are exactly the same with and without the AES 17 filter. I suspsct this is owing to the fact that the software uses FFT to dig out harmonics, and so the out of sync switching residue, will not appear on the measurement.
If the measurement was THD+N, i expect the result would be somewhat different......

In any case i would recommend using the AES filter. Preferably passive, or if active, populated with OPA627 opams. This is the only opamp i have found with low enough THD results. (Despite what the datasheet of many other opamps promise).

[Pabo]

Lars Clausen

A PC soundcard has a A/D converter on the input so I suspect that it also has a very steep low pass filter before this A/D converter. I have also seen that soundcards are unaffected by switching residuals.