Tom/Angel/Peter:
I went ahead and fired up my LTspice and entered a
schematic so I could see for myself what kinds of
stress we might see on a capacitor placed in parallel
with a filter choke.
I used parts values that I thought were believable if
not absolutely real-world. (No suggestions or other
feedback from the list, remember!)
The primary (mains) voltage was 120 at USA
frequency of 60 Hz. Transformer primary
inductance was 10 millihenries, second was a pair
of 1 henry windings with .5 ohms of resistance.
The filter choke was 10 henries, .1 ohms resistance.
The filter capacitor was 10 uF, load 1000 ohms.
Voltage across the load was a bit over 1000 volts
so we had a one-kilowatt system. The added
capacitor was .15 microfarad.
Ripple without the capacitor across the choke was
about 25 volts peak to peak, nearly sinusoidal.
With the added capacitor it dropped to about
5 volts peak to peak.
I found in my circuit that the circulating current in
the choke/capacitor pair was not at all high. But
one thing that really surprised me was the voltage
across those two items at turn-on. The voltage
across the capacitor (and also across the
choke) is the full output of the rectifier. If there are
any transients at turn on then the situation would be
even worse. Any transients coming in on the power
line would also appear right across that added
capacitor.
The only advantage I saw in putting the capacitor
in the circuit was that reduction of several dB in the
ripple level. Notching out the second harmonic of
the mains seems to be fairly easy but the other
harmonics are not particularly attenuated; the higher
ones actually seem to be increased in level.
But I would place money that the cause of the failure
of that added capacitor is voltage stress, particularly
at turnon or if there are transient voltages on the
power line. I would think that in a choke-input
power supply the choke would more easily tolerate
such transients; capacitors aren't so forgiving as
regards over-voltage.
- JimT
_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps
|