high voltage power supply

Under construction

The circuit:

Here you can see the circuit of my first solid state power supply. It uses an astabile multivibrator for generating the frequenzies. It still swings too slow (3000Hz up to now) because of the 22K-resistors and the 10n-caps. The other two transistors are for amplifying the signal and making it more rectangular. Now you can power a Mosfet with it. Try some different types of N-channel Fets and pay attention on the total gate charge (get some data sheets at www.conrad.com ). It should be as low as possible because every time the signal changes it has to be (un)charged. So you need some time to switch, which depends on the resistor before the gate. With a gate capacity of 200nF at 10V (as the IRL3803 has at this voltage) and R_Gate=100 Ohm (average current at 10V=0.05A) you have a gate charge of 2µC and so it takes about 50µs which are 1/3 of the total on-time of the Mosfet. There are two disadvantages of a high switching time: The first is the low secondary voltage in the used transformer (U_out~dU_in/dt), the second is the Mosfet getting very hot at high currents even if it has a low R_on (R_Drain-Source when it is fully switched on ). Note: The IRL3803 has a R_on of 0.006Ohm but up to now I use only a IRFP064N with 0.008Ohm (which is quite enough). Of course I can increase the gate-current by replacing the 100R by a 10R-resistor and the 200R one used by the fourth NPN-transistor by an also much lower one. Then I just need a transistor capable of a higher current then 0.2A and a headsink for the R_Collector.

No I replaced the 22K by 4K7-resistors. I had much heigher frequenxies (about 15KHz) and input current was much lower than before also when I used the same transformer.

The used transformer:

I have a TV flyback transformer which gives about 5kV when powered with 20V. Because of the bad Mosfet driver (the last transistor with its resistors) the Mosfet got very hot (without headsink). Secondary current was not really what I expected; maybe because of the low frequency (I've heard 20KHz are much better than 3000Hz). At one time I replaced the IRFP064N by a IRF530N which has a gate capacity of only 44nF (but a R_on of 0.11Ohm!) my TV transformer burned because of internally arc over! So I'm hopefully of getting much higher output voltages when degreasing the switching time dramatically. I also used some commercial 12V 10W transformers which gave also some KV at lower currents. With two transformers in series I still was able to trigger a small Xenon flash lamp (must be about 4KV).

Some time later I used a self-ade transformer (built from a transformer kit from conrad-electronics ). It uses a ferrit core, 5 windings primary (5 Ampere currentand about 350 windings secondary. When I powered it with the described circuit it gave about 1000 V (13.8 V Input). Guess why! 350/5=13.8=966V...). It was a some type of AC-Voltage with a peak/peak-value of about 2000 V, so I tried to feed a cascade witfh it. Again I was able to trigger the Xenon flash lamp, but this time it was much brighter.

A better circuit: