[Prev][Next][Index][Thread]
Re: Wiring caps in series
I have seen a lot of cap failures due to bad voltage sharing in many
computer power supplies I've repaired. When you run leaky caps in
series, you must use a string of voltage sharing resistors in
parallel. The bleed current in the string must be twice the current
passed by the leakiest cap for safety. You also have to choose
resistors that can handle the voltage you impress across them. ALL
the failures I mentioned earlier were due to an inappropriate choice
of resistor. By using low power resistors, you run the risk of them
going high and even open circuit which means that the cap unfortunate
enough to have an open circuit resistor across it is going to be
charged to a voltage well beyond what it can handle.
When using electrolytics at high voltages you should ALWAYS have
an appropriately rated bleed resistor across it. Electrolytics possess
the "memory" effect like no other.
Finally, all caps in a series string should have the same value
or trouble will arise from uneven charge/discharge voltages.
Malcolm
You wrote....
> || || || || || ||
> o-||--||--||--||--||--||--o
> || || || || || ||
>
> A B C D E F G
>
> The first time I charged them up, each of the caps had about the same
> voltage across them (Vab=Vbc=Vcd=Vde=Vef=Vfg=250 volts). After
> dischaging them by bridging A and G with a wire, I assumed that all
> of the caps would have discharged to some low value. What I found
> instead is that the caps had more or less random voltages on them
> (between -50 and +50 volts). Charging the set up again (by putting
> 1500 volts across points A and G) resulted in widely varying voltages
> across the various capacitors.
>
> Given this situation, I could easily see some of the caps failing
> due to over-voltage conditions, which are caused by the varying
> voltages that result when the capacitors are charged and discharged
> in series.
<snip>