From: Winfield Hill
Newsgroups: sci.electronics,sci.electronics.design
Subject: ceramic caps across the diodes on a bridge rectifier
Date: 21 Sep 2002 08:01:51 -0700
Organization: Rowland Institute
Message-ID:
References: <3d8b4028.29581636@news.btopenworld.com>
NNTP-Posting-Host: p-613.newsdawg.com
X-Newsreader: Direct Read News 2.97
Martin Griffith wrote...
>
>
http://martingriffith.users.btopenworld.com/PREAMP.HTM
From his web pages Martin wrote, "I built this Elektor preamp
in 1980, and I'm still using it. ... The only difference I
would make, would be to put 4 100nF ceramic caps across the
diodes on the bridge rectifier."
That's wise - it's for the purpose of dealing with the spike
resulting from diode snapoff when its reverse-recovery time
is finished. During the reverse-recovery time the diode
continues to conduct, even though the current is reversed.
When all the charge is finally swept out of the diode and it
does stop conducting, this happens suddenly, creating a spike.
The spike comes the transformer's leakage inductance, recall
the formula, V = L dI/dt. The spike's risetime and magnitude
are determined by parallel capacitance. Risetime, dV/dt = I/C,
and magnitude, V = I sqrt L/C as determined by matching the
inductive and capacitive energies, E = 0.5 L I^2 = 0.5 C V^2.
The current I in these equations is the reverse current flowing
in the diode just before turning off, which can easily approach
the average dc current being delivered to the load (remember
that peak diode forward current greatly exceeds the average).
We see that increasing C reduces dV/dt and V. Without an added
capacitor as Martin suggests the only capacitance helping us is
a low diode back-bias capacitance and the transformer's small
winding capacitance. Without any additional capacitance, it's
possible for a 50V spike lasting 5 to 20us to occur once every
ac half-cycle, not a good idea inside quiet audio equipment!
Martin suggests 0.1uF on each diode. The same good result is
achieved with two capacitors, one across each secondary winding.
There's another important issue, the resulting L-C tank circuit
has a high Q and will ring for many cycles. Because the 100nF
capacitor results in low frequencies, the ringing lasts an even
longer time. To deal with this, I like to add a series damping
resistor, damping the ringing in one cycle if properly selected.
Thanks,
- Win