Ripple Filter Damping Techniques

Found this very useful article by Kevin M. Tompsett on damping of a PSU output filter:

http://www.analog.com/media/en/technical-documentation/technical-articles/Designing-Second-Stage-Output-Filters-for-Swit… 

In Fig 4 there are 3 techniques shown to damp oscillation in the output C-L-C pi filter:

1. Resistor across the inductor

2. RC load on input

3. RC load on output

I can understand how techniques no. 1 & 3 work as they are loading the output of the filter to damp the oscillation, but I do not understand how no. 2 works. How can a RC load on the input of the filter stop oscillations on the output? It seems to me that the output LC can still resonate.

Any clarification of how technique no. 2 works would be well appreciated!

Thanks, Ken

Parents
  • Hi James, Kevin,

    Just to let you know I have tested the filter on a LT8643S supply (OK, it's not a part from Analog, but now LT is part of AD ). Fsw = 1.1MHz, Vout = 1V1, L1 = 1uH, C1 = 2uF, CD = 2uF, RD = 0R47, LFILT = 110nH, COUT = 300uF (all ceramic). Works a treat! Thanks Kevin!

    On the bode plot you can see the Pi filter resonant peak at about 350kHz (Tracer: ChA), which is well beyond the gain crossover at 70kHz (M1). [Plot taken with a CS328-FRA]. I would have liked the resonant freq to be a bit lower, eg. Fsw/5, but it was upsetting the gain margin, and as it is I still get very good ripple filtering even though it is Fsw/3.

    Here is the ripple before (ChA, about 100mVpp) and after (ChB, about 2mVpp) the ripple filter. This plot also shows the response to a 3.3-6.6A load transient. The great advantage of taking the feedback after the ripple filter is that the control loop can compensate for the IxR drop on the filter inductor, which is 12mOhm in this case and the drop would have been 3.3x12m=40mV. Here the IxR drop is only 1096 (M1) -1091 (Tracer) = 5mV.

    Good luck with your designs!

    Cheers, Ken

Reply
  • Hi James, Kevin,

    Just to let you know I have tested the filter on a LT8643S supply (OK, it's not a part from Analog, but now LT is part of AD ). Fsw = 1.1MHz, Vout = 1V1, L1 = 1uH, C1 = 2uF, CD = 2uF, RD = 0R47, LFILT = 110nH, COUT = 300uF (all ceramic). Works a treat! Thanks Kevin!

    On the bode plot you can see the Pi filter resonant peak at about 350kHz (Tracer: ChA), which is well beyond the gain crossover at 70kHz (M1). [Plot taken with a CS328-FRA]. I would have liked the resonant freq to be a bit lower, eg. Fsw/5, but it was upsetting the gain margin, and as it is I still get very good ripple filtering even though it is Fsw/3.

    Here is the ripple before (ChA, about 100mVpp) and after (ChB, about 2mVpp) the ripple filter. This plot also shows the response to a 3.3-6.6A load transient. The great advantage of taking the feedback after the ripple filter is that the control loop can compensate for the IxR drop on the filter inductor, which is 12mOhm in this case and the drop would have been 3.3x12m=40mV. Here the IxR drop is only 1096 (M1) -1091 (Tracer) = 5mV.

    Good luck with your designs!

    Cheers, Ken

Children
No Data