0
\$\begingroup\$

I'm building a small class D amplifier: some op amp circuits for pre-amplification, a pwm modulator based on a dsPIC, some power buffers for controlling the gates of an h-bridge power stage, and a LC output filter.

To power the system I bought a cheap (7€) power supply 12V, 5A. A photo below to give the idea of the device: enter image description here

I tested it and it seems quite strong: while increasing the current flow the voltage always remained at a fixed value.

The problem is that every 15us there is a very high frequency oscillation (~1MHz, ~2V amplitude) on the output. Images below:

enter image description here

To reduce the output of these pulses i tried:

  • An additional LC filter (L=200uH, C=30uF tantalum) on the output. The ripple still remains and I don't know why.
  • A 12v standard zener regulator. The ripple still remains (maybe because the frequency is too high for a zener diode???)

What else do you recommend to do?

Thanks

\$\endgroup\$
5
  • 1
    \$\begingroup\$ how are you probing your psu? maybe its radiated noise, not conducted \$\endgroup\$
    – Wesley Lee
    Commented Jul 27, 2016 at 17:40
  • \$\begingroup\$ @Wesley Lee I don't think it is radiated noise because it is not a constant sine waveform but is an oscillation that happens every 15us with a very high frequency and dies in about 500ns. It's like a high frequency heart beat. \$\endgroup\$
    – Dukenukem
    Commented Jul 27, 2016 at 18:46
  • 1
    \$\begingroup\$ radiated noise does not have to be a sine wave.. \$\endgroup\$
    – Wesley Lee
    Commented Jul 27, 2016 at 19:08
  • \$\begingroup\$ Try putting a load on the output. \$\endgroup\$
    – Andy aka
    Commented Jul 27, 2016 at 19:09
  • \$\begingroup\$ Tantalum capacitors do not work well at high frequencies. Try a ceramic instead. You don't need to be down at fc=2kHz (which may be detrimental to your audio), as you are targeting noise that is around 50MHz. \$\endgroup\$
    – Mark
    Commented Jul 27, 2016 at 20:00

1 Answer 1

Reset to default
0
\$\begingroup\$

Your power supply appears to be missing the Y capacitor which will make it more difficult to measure the output noise (the 2V you are seeing is almost surely not real). It does have what look likes a nice common-mode choke and X capacitor(s).

enter image description here

The purpose of this capacitor is to conduct some of the 67kHz switching noise out to the earth terminal. It must be a safety rated capacitor designed for Y use. Don't just stick any old cap in there- this is a safety issue.

http://photos.prnewswire.com/prnvar/20130521/CL18315

To get a better measurement, try grounding the supply output to earth and using a differential measurement on the output terminals (A-B with two scope probes). Just grounding the output deliberately (not through the scope probe) should help.

\$\endgroup\$
6
  • 1
    \$\begingroup\$ Thank you @Spehro Pefhany, using an Y1, 1000pF capacitor reduced the noise to 50mVpp. Considering that this is a classD power supply, would it be usefull to use an additional LC filter (with very low cut frequency) on the output of the smps? \$\endgroup\$
    – Dukenukem
    Commented Aug 4, 2016 at 17:30
  • \$\begingroup\$ Good to hear it made a profound difference in the measured noise. You could consider adding a common mode L-C filter (google common mode choke- something like 1mH and rated for your required current or above) but the switching frequency (and the harmonics) are way above audible so a bit of noise may not have much effect on the quality of sound. \$\endgroup\$
    – Spehro 'speff' Pefhany
    Commented Aug 4, 2016 at 18:25
  • \$\begingroup\$ Ok but, a common mode choke is only for AC (so it should be placed in the 220V AC line), am I right? Would be useful to add an extra LC (with a simple 1-wire-inductor) filter between the output of the smps and the classD circuit? \$\endgroup\$
    – Dukenukem
    Commented Aug 5, 2016 at 16:47
  • \$\begingroup\$ You can use a common mode choke equally well on DC! \$\endgroup\$
    – Spehro 'speff' Pefhany
    Commented Aug 5, 2016 at 17:12
  • \$\begingroup\$ Can you please explain me why it would be better to use a common mode choke instead of a standard LC filter? I dont get exactly why someone should use one instead of the other. \$\endgroup\$
    – Dukenukem
    Commented Aug 8, 2016 at 12:51

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.