How to calculate RC comp network for diode stack?

[scoobyrollz]

I am trying to design HV diode stacks for T/R sets. I was wondering about when adding an RC compensation network, if I should match the line frequency or the reverse recovery time of the diode? Also should I use T = RC and then f = 1/T or f = 1/(2*pi*RC) to calculate the frequency characteristics of the RC network? Any help will be greatly appreciated.

 

[Skogsgurra]

No need to double-post. You got answer in your first post.

Gunnar Englund

http://www.gke.org

 

[VEBill]

Assuming that the other thread being referred to is the one about the 'FWBR', then that one might be a slightly different topic than a 'diode stack'.

Diode stacks (meaining: diodes in series) are typically used when you can't find diodes with sufficient reverse breakdown voltage. Because the individual diodes might have slightly different reverse characteristics, one uses parallel resistors (like 100k) and capacitors (maybe nF) to swamp out these variations to ensure that the reverse voltages are distributed equally across the series diodes both during the reversal and during the reverse peak.

Reason: There's no point in putting diodes in series if one diode takes most of the reverse voltage at some point during the waveform.

Even nearly 30 years ago (when I was building an HV PS for my HF RT), the generally accepted advice was to try to find better diodes rather than use this technique. But I had to use this technique due to lack of better parts.

Long and short of it, although the parallel RC looks like an RC network, but it can be examined more properly as a capacitor and a resistor. The values are chosen to swamp (x10) the diode reverse characteristics.

Watch out for HV across these components. Might have to use higher wattage resistors simply for the peak voltage.

As was the case 30 years ago, it is much better to simply find diodes that don't need to be stacked. That is much easier these days.