0
\$\begingroup\$

I've got an AC-DC power supply which has an amount of ripple which is unsatisfatory for my application, ~250 mVpp, sometimes even 600 mVpp when there is a DC-DC downstream (for some reason).

I'd like to bring that down to 100 mVpp or less using an LC low-pass filter - which should also, I think, decouple the two switching controllers that may cause this increased ripple when a DC-DC is present downstream.

I thought to myself this would be easily fixed with an LC filter downstream of the AC-DC. The switching frequency of the AC-DC is 360 kHz, so I chose an inductor of 10 μH capable of handling my load current of 4.5 A without saturating, and slapped a 10 μF ceramic cap (taken from an assortment) rated for 50 V when my bus is 35 V, meaning the cut-off is supposed to be 16 kHz for an attenuation of that ripple by a factor of 60 in theory.

enter image description here

I know my filter is soldered correctly, because I checked it a gazillion times using continuity checks and the LCR meter (and the inductor on its own), and yet the AC-DC's output passes through nearly unencumbered - barely divided by half.

I know there are parasitics to take into account such as SRF, but I should not get as low as a factor 2, should I? Especially given that I've increased the capacitance by a factor 10 (120 μF) and it doesn't do anything more. It even worsens the noise amplitude slightly.

Am I wrong in thinking a LC filter should do the job here and that something's afoot?

Note that, to my great demise especially in this instant, I do not own a function generator to actually plot the bode chart of the filter.

\$\endgroup\$
13
  • 1
    \$\begingroup\$ What are the load conditions when you measured 600 mV or ripple? What is the source impedance of the regulator? When you calculated the attenuation, how did you model the load? Open circuit or with some current flowing into the load? \$\endgroup\$
    – user57037
    Commented Oct 8, 2022 at 2:46
  • 1
    \$\begingroup\$ In order to design a filter you need to know the source impedance and the load impedance. \$\endgroup\$
    – user57037
    Commented Oct 8, 2022 at 2:47
  • 1
    \$\begingroup\$ I am just probing. One thing, though, at 360 kHz the output impedance of the regulator will not be 0.1 Ohm. It will be complex. You will be "feeling" the output filter on the DC-DC. \$\endgroup\$
    – user57037
    Commented Oct 8, 2022 at 3:08
  • 1
    \$\begingroup\$ Maybe the inductor has a lot of parasitic capacitance at 360 kHz? The windings couple to each other capacitively. What is the SRF of the inductor, or do you have a model or chart for it that models the high-frequency behavior? \$\endgroup\$
    – user57037
    Commented Oct 8, 2022 at 3:10
  • 3
    \$\begingroup\$ Do you feel like your oscilloscope probing technique is beyond reproach? It is not so difficult for 100 mV or more to couple into an oscilloscope one way or another. \$\endgroup\$
    – user57037
    Commented Oct 8, 2022 at 3:18

0

Reset to default

Browse other questions tagged or ask your own question.