On Sun, 20 Nov 2005, Wouter van Ooijen wrote:

>> I was recently talking to a guy a an EMC test house. He says
>> that caps across a bridge make a big
>> difference to conducted emissions, and it's the first thing
>> they try when they have a product fail,
>> and it frequently is the only thing they need to do to make it pass.
>
> Any hints at the theory behind this, and how do you select the
> appropriate values (and maybe type) of the capacitors?

Afaik choose C~=50-100 times the capacitance between primary and 
secondary of the mains transformer (measured with the core grounded if 
so fitted, else 'in air').

>> From the fact that they are over each diode and not over the trafo
> secondary and/or after the bridge I suspect that they might have
> something to do with the diodes themselves. Maybe prevent nasty effects
> of high dv/dt spikes?

That too, when you have a RFI event in the primary it can kill the 
diodes if the caps are not on.

When the diodes switch off hard the transformer secondary still carries 
current and the impedance it sees is nearly open circuit. The resulting 
high speed pulse is similar to the one on the drain of a MOSFET being 
turned off fast and propagates all over the place. It can also mix with 
other signals (f.ex. radiated and conducted RFI from within the project 
box which will be mixed with 50Hz ac in the 'modulator' bridge).

I never had trouble with this so far but I nearly always put the 
capacitors in. I usually put in just two caps, from each transformer 
secondary wire to the circuit gnd. If you have long wires then this may 
not be enough. If you use real cheap diodes which leak and have large Vf 
drop, and a small filter cap which increases the conduction angle, then 
the problem will not happen in the first place ;-)

This diode turnoff trick is one of the things that can be modelled 
nicely with Spice (be sure to model the stray capacitances and 
inductances). It looks very nice, almost like a spark gap transmitter's 
output ;-)

Peter
-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist