I'm planning to discharge a 1500VDC 150uF~250uF capacitor into a coil 20 times per second. I've found a few capacitors that might do the work, but I have some questions about their charge/discharge frequency capability.

This is one that I think would work.

It seems to be designed for DC links and not for pulses. Is there any way to know if it can handle the charge/discharge pulses in 20Hz? Maybe its dV/dt parameter?

Another one that looks promising is this.

Since it's designed for filters, I suppose it's more suitable for high frequency charge/discharge. Does it make sense? Both are gonna work current and voltage wise, but how about frequency?

  • \$\begingroup\$ Which one specifically in the range? There are literally hundreds in that data sheet. \$\endgroup\$
    – Andy aka
    Apr 28, 2020 at 17:49
  • \$\begingroup\$ B25620B1217K983 in the first datasheet and B32373A7157J080 in the second one. \$\endgroup\$ Apr 28, 2020 at 18:08
  • \$\begingroup\$ State uF and voltage rating. I can’t keep referring to a long number on my stupid tablet because it makes me want to put a hammer thru the screen lol. \$\endgroup\$
    – Andy aka
    Apr 28, 2020 at 18:20
  • \$\begingroup\$ The 215 uF part won’t do. Why not read the data sheet and see why. \$\endgroup\$
    – Andy aka
    Apr 28, 2020 at 18:24
  • 2
    \$\begingroup\$ Those caps have a big difference between Is (non-rep surge) and Ihat (repetitive peak current). In addition, in pulse use, you may have to derate the voltage if there's any voltage reversal, due to dielectric relaxation over-stressing the dielectric \$\endgroup\$
    – Neil_UK
    Apr 28, 2020 at 18:27

1 Answer 1


Edit: Corrected timings.

For the B25620B1217K983, \$I_S\$ (surge current) may be 9.6kA, but \$I_{MAX}\$ is 60A... if you discharge 5kA within 50µs (of 1/20Hz = 50ms period), then:

$$D = \frac{\over{P}}{P_i} = \frac{\tau}{T}$$ $$D = \frac{50µs}{50ms} = 0.001$$ $$A_{AVG} = D\cdot P_i$$ $$A_{AVG} = 0.001\cdot5kA = 5A$$

  • Discharge component: \$\frac{5kA/50µs}{1/50ms} =\$ 5A/cycle average
  • Charge component remaining: \$I_{MAX}\$ - 5A = 55A per cycle

It sounds like a lot is remaining, but charging will take much more time where current is non-zero, leading to higher RMS current. I think this could pose trouble depending on how it is charged.

For instance, if you chose a 50Ω charging resistor, $$V_C = V_{IN}[1-e^{-t/RC} ]$$ $$t=-RC\cdot ln\frac{1-Vc}{V_{IN}}$$ $$t=-50Ω\cdot 215µF\cdot ln\frac{1-0V}{2kV}$$ $$t=-0.01075‬\cdot ln(0.0005)$$ $$t=-0.01075‬\cdot -3.30103$$ $$t=35.5 \text{ms (99% at 49ms)}$$ At charging \$t=0\$, 2kV is across 50.0029Ω = 40A

Resistor pulse rating: 2kV * 40A = 80kW!

Initial maximum theoretical inrush charging current = \$\frac{2kV}{2.9mΩ} = 690\text{kA}\$

To avoid pages of math and get a quick idea of the total average current, I came up with this:

LTspice simulation

Checking the datasheet, \$I_{MAX}\$ RMS has been exceeded by 117.61/60 = 1.96 times. Perhaps the values could be tweaked some, but this is worrisome. An alternative could be to use multiple caps and take turns firing them sequentially.

Needless to say, 2kV charged caps are extremely dangerous.

  • \$\begingroup\$ Thanks for your reply. Do you have any tip on how I could charge the capacitor without those huge power and current? Maybe switching? \$\endgroup\$ Apr 29, 2020 at 4:24
  • \$\begingroup\$ You made a little mistake. 20Hz has a period of 50ms not 5ms. But I got the idea. \$\endgroup\$ Apr 29, 2020 at 4:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.