I would like to design a transient protection solution for HVDC_pulse_unit.

HVDC voltage = 3500V , current > 40kA.
D1 flyback diode survives(50kA surge forward current).
MOVs section did not trigger (1200j, Varistor voltage = 5400V).
D4 12x low power serial diodes (300mA, 16kV each) rated 192kV burned.
EDIT:D4 is placed wrongly on the circuit. It protects the rectifier bridge of the charger inside the pulse unit. It was mentioned to acknowledge the magnitude of the voltage spike. Please ignore it.


Possible solutions:
Op1 - Using RC Snubber method for 3kA pulses (picture below), R C values where computed. The problem is voltage rating for C1. If D4, rated at 192kV failed how would a snubber capacitor survive?

3300 ampere


Op2 - Would "normal" 1500W peak power TVS diodes survive until the varistors start to conduct? A small inductor in series with the TVS diode in order to limit the current rate of change?

Are the two solutions viable? If so, which one is better? I have no formal schooling in electronics, so all kind of errors might plague this post. Thank you.

EDIT: I did a few tests (below 10kA) using TVS, 10 bidirectional 1.5KE400CA in series. Breakdown voltage 400V per diode. Still got 8kV voltage spikes (picture below, x100000 attenuation)

enter image description here

  • \$\begingroup\$ Can you edit your question to provide more specifications on D4? Current, power, recovery time, etc? Also, what are the characteristics of the pulsed DC? \$\endgroup\$ – Edgar Brown Nov 5 '18 at 0:36
  • \$\begingroup\$ How is the reverse voltage over the diodes in D4 distributed? Parallel resistors over each of them, or do you trust them to have equal leakage currents, being all from the same batch? Maybe the circuit was still operating within its operating range (MOV didn' t trigger) but one of the diodes in D4 was unbalanced in leakage, setting off a cascade of failing diodes. If you want a capacitor to survive, is it a valid suggestion to dimension it large enough to absorb all the energy without reaching its breakdown voltage, or would that lead to improbable large values? \$\endgroup\$ – joe electro Nov 6 '18 at 3:49
  • \$\begingroup\$ Same batch, no parallel resistors. If the blocking voltage per diode is 16kV and the first one in chain failed, then the transient is over 16kV? If that's correct, then the MOV should have triggered. I think MOV response time is too slow. The released energy > 10kJ, contained by the crowbar diode D1. The transient in not that energetic, at least not in the range of the pulse, it destoyed D4 diodes, and maybe damaged one HV transformer protected by D4 ( There is a series of balanced TVS diodes across the transformer, it was not connected). It is not possible for C1 to absorb all the energy. \$\endgroup\$ – Maris Nov 6 '18 at 11:24
  • \$\begingroup\$ It is not possible for C1 to absorb all the energy of the pulse, the energy of the transient is unknown. \$\endgroup\$ – Maris Nov 6 '18 at 11:32
  • \$\begingroup\$ You have a good estimate of the energy of the transient just with the maximum current through the inductor. \$\endgroup\$ – Edgar Brown Nov 6 '18 at 11:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.