How do high voltage high current pulse power supplies work? For example in electrical discharge machining (generating sparks that remove metal,) a single square pulse can be as small as 1 microsecond (1MHz) with a 1 microsecond off-time, with the voltages up to around 200 - 300 volts and the current up to 30 - 50 A (when a spark occurs the voltage drops down to around 10-30 V.)

I've tried researching premade systems, but they are either high voltage low current (kV and mA,) or high current low voltage (low double digit current and low double digit voltage.) The simplest way to produce these pulsed power supplies would be to use a waveform generator and link it to a power supply that can switch at 1MHz or faster, but once again it is either high voltage low current, or low current high voltage.

Does anyone have any idea how they make these pulse power supplies?

  • \$\begingroup\$ Are you searching "spot welding" generator? \$\endgroup\$
    – Antonio51
    Nov 11, 2022 at 18:20

3 Answers 3


300 Volts at 50 Amperes is 15kW of power - a very large industrial machine. Can get away with a tenth of that if cutting speed isn't a huge concern.

Such a power supply would be very large (and costly) and consist of a three-phase 480 VAC input. This would be bridge-rectified to ~700 VDC, which is of course very dangerous. One or more 300V adjustable bucking DC regulator(s) working in synchronicity would drop this to the target voltage. Many bulk energy storage capacitors would be used to keep this level constant. And some fast means of switching these large currents on and off (MOSFET or IGBT) would be used. Each pulse would have very fast (tens of MHz) current-loop monitoring, to ensure the current stays where it should during each pulse, and doesn't blow anything up.

You won't find such a schematic because the only team of people whom have them, spent years and a whole lot of money fine-tuning a niche device for a niche application. They won't share them, ever.

The supply of power isn't really all that complicated; the power just burns away the metal. A welder can burn away metal. But EDM is a rather unique use-case for a power supply; special considerations must be made such as what to do when pulse current gets too low (should the voltage increase?) or too high (reduce voltage or stop machine?) And should current stability be prioritized over voltage stability? And in the case of wire EDM, too much current can break the wire; how to handle that? All of this must be analyzed on a pulse-by-pulse basis, including sufficient dead-time for the fluid to flush out the swarf. Then a whole host of ancillary machine concerns arise, such as fluid conductivity monitoring / filtering, temperature regulation, quill/wire feed control, etc.

Anyways the "cheap EDM machines" you can find on eBay and Aliexpress are just that - cheap. Very cheap. They will output some volts and some current, and could be used in a pinch to EDM-out a hardened bolt a few times before they die. Inside they use a primitive voltage and current regulator (primitive like a single big capacitor, inductor, and resistor.) They have none of the engineering design provisions for robustness, protection, optimization, and feedback control that a 15kW machine would require.

Suggest researching the BAX EDM website and videos for more details.


You can use a high-voltage low-current power supply to produce high-voltage high-current pulses, at a low duty cycle, by charging up a capacitor from the power supply and then rapidly discharging it. A power supply limited to 100 mA can then theoretically provide 1 A of current in 10% duty pulses, or 10 A in 1% pulses, and so on. The maximum length of a pulse is determined by the capacitance; larger capacitors can do longer pulses.

This is the principle on which most pulse-only power supplies (commonly used for semiconductor testing, for instance) work; they have enough capacitance to hold the voltage steady for the length of a pulse, and a fast switch to turn on and off. I've personally designed several ad-hoc test systems using this idea.

This wouldn't help quite so much for your 1 μs on/1 μs off scenario, since you'd only manage to double the theoretical peak current, but it does work quite well for my 3 μs on/10 ms off scenario (I'm using it to get 60 A pulses out of a 50 mA power supply).


How do high voltage high current pulse power supplies work?

There are many schematics, depending on polarity, modes, and devices used ...

Here is a basic example of such a "pulse power supply".
Here for just ~ 400 A, 2 kW on R1 (NB: this resistor is "really" unknown.
Note also the power needed ... and power dissipation on devices.
Cooling used may be "circulating water".

enter image description here


Your Answer

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

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