Why can AC ripple increase with bigger C in a CRC filter?

[leadbelly]

The context is I am about to start building a tube preamp which I have all the major parts for, all that's left is designing the PS filter stage.

I want to keep cost as low as possible, so I was thinking of using CRCRC filtering to avoid buying big $$$ chokes. I got the most recent copy of PSU Designer II and started to simulate, just playing with values of C and verifying how best to stagger the values for C in order to minimize ripple. In playing around, I noticed that leaving everything else alone and just increasing the value for one C out of the three can sometimes increase the ripple. Why? I always thought ripple would monotonically decrease with increasing capacitance no matter what the configuration.

[Stocker]

It could be an induced self-resonance caused by the particular values selected. I could also not know too much about it.

Mostly this is a bump-post. I would have thought the same as you.

[Sch3mat1c]

My vote is for resonance, although it shouldn't happen with a pure low pass filter system (no L). Might be the last C and the RL across it is acting as the excited element.

CRC for constant current loads is fine, use a moderate value for the first, enough to get maybe 5% ripple (for +300V, that'd be around 15V) then filter the rest with the second cap. This also assures a solid foothold for the circuits (one 100uF instead of a few feeble 10-20uF caps).

I don't like squishy long chains of R and/or L, they have a habit of causing phase shift oscillators. Better to rule out that possibility right away, even if there's no actual chance of it happening (aka, good practice).

Tim