I stumbled on this little experiment at vt52.com. Same three capacitors of different values in a CLCLC power supply. But different results just by swapping them positionally. In general it seems going from small to large capacitors in a filter chain is best. I wonder why.
http://www.vt52.com/node/180
http://www.vt52.com/node/180
The signal current primarily flows through the final capacitor. The lower its impedance, the better.
More specifically i would say the lower the ESR the better. With these small values, not too hard to find. Also, with this small load i would suspect chokes are overkill - perhaps resistors would suffice. Worth modelling anyway.
Last edited by a moderator:
The ripple current spikes of the first capacitor after the rectifier diodes are proportional to its capacitance. These spikes contain a lot of spectral components well into the MHz range. They will propagate through the ground conductor or radiated to nearby sensitive components by electromagnetic coupling. Smaller first capacitor - lower amplitude of spikes. The next capacitor gets mostly 100Hz ripple.
It all depends on your priorities: if ripple rejection is your main concern, the lowest impedance components (large cap and small inductor) should be placed upstream. That way, you can come close to the 6dB/octave afforded by isolated reactive components.
However, in real life, you need to take many other factors into account: some have been evoked by the other posters
However, in real life, you need to take many other factors into account: some have been evoked by the other posters