Uzzors2k Banner

Microwave Oven Capacitor Bank
Microwave Oven Capacitor Bank

One of the "big HV projects" is blowing stuff up with high voltage capacitors. Now for some really good capacitors, you'll need to dish out a few hundred dollars, if you can even find them for sale. So I was quite surprised when I saw Steve Ward had used common microwave oven capacitors in a capacitor bank. These are the capacitors with 2kV AC, 1F ratings at 50Hz, making them pretty much unsuitable for any thing interesting. However, Steve found by experimenting that they can withstand up to 10kV DC for a limited time, and have remarkably low internal terminal inductance. So suddenly once useless capacitors appear suitable for some low energy pulse discharge experiments.

MOC bank

Before we continue I hope I don't need to warn you of how dangerous this is. If you don't experience with BOTH electronics and high voltage, then steer clear. One mistake and your heart WILL stop and you will lose a lot of flesh. This project has been planned for years. About two years ago I bought 20-something microwave oven capacitors, since I had a hard time finding them. A year after that I cleaned, soldered them together, and made the HV switch. Finally this year I made a small voltage doubler to rectify the "Big-Mofo" transformer, so I could charge the bank.

The capacitor bank itself consists of 20 capacitors averaging 1F each, and capable of withstanding 8-10kV for a short duration, according to the finds of Steve Ward. The measured capacitance of the bank is 21,8F, and the estimated energy in each shot is about 700 joules, give or take 100j. I couldn't read the final bank voltage with much accuracy, and I don't know how much it would have sunk while the switch closed. Bank voltage was measured using a 50A ammeter, with a 200M resistor giving 50A at 10kV.

HV Switch  can crusher setup

The HV switch was the only critical part of the project, and it serves two purposes. First it switches the bank into the load, and second it switches the charging circuit off of the bank. Leaving the charging circuit connected may result in failure of the rectifiers once the bank voltage reverses. When switching a 700 joule bank using a spark gap, any electrode surfaces are going to vaporize, and if they happen to be in contact they'll just weld together. So what's required is a heavy duty switch that won't make contact, but get very close, and is able to handle several thousand amps. To solve this, I used a metal rat trap, and rebuilt it to slam a copper bar into another bar. The spacing could be easily adjusted using rubber stoppers. The real beauty is that the charging circuit in connected to the holding pin, which is disconnected from the capacitors once the mechanism has fired. The switch is also easy to trigger from a distance. After 10 shots there is some wear on the electrodes, but not enough to impair it's function. The solder holding the copper to the iron(?) bar is what's showing the most wear.

Worn busbar

Crushing and Skrinking stuff
The first thing I tried was electromagnetically crushing beer cans. The cans are wrapped in 5,5 or 6,5 turns of 18AWG wire. Steve Ward was able to crush some dimes using his setup, so I hoped I could crush some Norwegian 50-rings, which are 97% copper. I didn't have much luck though, and even after several shots on the same coin there is almost no noticeable deformation. I'll have to investigate this further. The work coil became warm, and would pull in and bulge out a little as described on Bert Hickman's coin shrinking site. Ultimately my bank is much weaker than those which are usually used, but some degree of shrinkage should be possible.

Can in deathcoil  Pinched cans Coin shrinker coil.

Can Crushing Video


While cleaning old projects I came quite close to sending this one to the dump before thinking better of it. Steve Ward was able to skrink coins and tear cans apart using less energy than this bank can contain, so I should be able to do the same with my capacitors. After looking over the construction with new eyes I saw some potential for improvement. For one, the bank and switch could be redesigned for lower inductance and resistance, simply by rearranging things and using solid copper conductor. In the previous design multistrand wire and poor solder connections were abundant. To remedy this I opted to integrate one of the bus connections with the switch, to reduce the length of conductor needed, and simulatnously the inductance of the connection. The multistrand conductor was removed altogether, and replaced with thick solid copper. The distance between the capacitor bank terminals has been reduced as well. To reduce the contact resistance proper screw contacts have been soldered on.

Unlike with the previous capacitor bank, I decided to characterize this one to get some numbers for simulation. The test setup consisted of a SCR, current transformer, and external damping resistor. The HV switch was closed with a washer to short circuit it (in the nominal closed position it is spaced less than 1mm, too wide for 60V to jump across). With this simple setup all parameters of the capacitor bank can be determined. The capacitance was measured using my LCR meter. The leakage resistance, RL, can be determined by working backwards, realizing the feed resistor and the leakage resistance create a voltage divider. Once the capacitor bank has been charged to a known voltage, it can be discharged through the SCR, and the resulting current waveform captured on a storage scope. The frequency of the current waveform will only depend on the capacitive and inductive components of the bank, and with the capacitance known, the inductance can be worked out using the formula for LC resonance frequency. The dampning of the waveform will depend on the resistance of the circuit, which consists of the conductor resistance, contact resistances, and also the external damping resistance. By noting the peak current and using Ohm's Law, the system resistance can be determined, and the bank resistance after subtracting any external resistance. I've placed all the required formula in a spreadsheet for easy use.

Test setup schematic. Scope traces.

Characterization schematic, and captured discharge waveform.

The determined specifications of the bank are:

C = 21920nF +/- 10nF ~= 21.9F
ESL = 679,5nH
ESR = 270mOhms
RL = ~500k

Remaining upgrades:
* Difficulty reading the voltage of the bank, could instead use op-amps and voltage dividers of defined ranges, with LED indication of each voltage threshold. Would allow for consistent voltage between shots, and hopefully avoid overvoltage situations.
* Stronger capacitor charger, the previous model was just barely able to bring the bank to full charge.
* Determine how the number of turns is calculated. The inductance must have some impact, of perhaps just the number of turns? Unsure if some kind of inductance matching is needed, or if some sweet spot in terms of current rise is present. The naive idea would be that many turns = stronger magnetic field = more crushing force. Only draw back is how much current the wire can carry. Maybe the coil resistance needs to match the ESR of the bank, for maximum power transfer/impedance matching? Simulation seems to show that a lower inductance gives a much higher current peak, and shorter pulse duration. Prime area seems to be 2H for this bank configuration.

Similar High Voltage Capacitor Banks:
Bert Hickman's Page
Microwave oven capacitors and a Heavy duty Maxwell setup at Steve Ward's page
Destructotron at Mike's Electric Stuff
TeslaDownUnder - Can Crushing
Kaizer Power Electronics - 333 Joule MOC capacitor bank

Youtube Flickr Twitter LinkedIn To top

Disclaimer: I do not take responsibility for any injury, death, hurt ego, or other forms of personal damage which may result from recreating these experiments. Projects are merely presented as a source of inspiration, and should only be conducted by responsible individuals, or under the supervision of responsible individuals. It is your own life, so proceed at your own risk! All projects are for noncommercial use only.

Creative Commons License This work is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 Unported License.