High Voltage Issues

[Anthony P]

I am working on the power supply for my DIY dye laser. It is a 20KV supply using a variac to control a 5KV, 30mA neon sign transformer, which feeds a 4 stage cockroft-walton multiplier. This charges a 1uF 20kv cap through a 20k resistor to about 15kv for the flashlamp.
The problem is 2 part. 1) according to PR-28A high voltage probe, the voltage fluctuates wildly +/-100v. I expected some ripple from the multiplier, but this seems extreme. Should I try larger caps in multiplier (.oo1uF now) or is this amount of ripple normal and just fine?

2)The second issue is metering the charge on main cap. I have tried several meter configurations. One was a voltage divider with over a Gig-ohm total resistence. The next was a uA meter with a 100meg HV resistor. Both did an excellent job of monitoring the cahrge on the cap. The problem is when the lamp was fired, the meter died.

Any thoughts would be appreciated. I am not afraid to experiment and purchase odds and ends as needed.


EDIT:
I have done more research and I believe the voltage fluctuation is from a combination of resistive loading and diode losses. A 12KV diode has a forward voltage drop of over 6V and there are 4 of them in the circuit.

I am still open to meter ideas.

 

[Cyparagon]

Yeah, high voltage is tricky to meter - at least with high impedance outputs like you're working with. It shares similarities to quantum mechanics, in that you can alter the parameter merely by measuring it. 10meg of stray parasitic resistance in this test lead or a couple nanofarads of stray parasitic capacitance with your hand nearby would never affect a more standard low voltage DC value. But with high voltage high impedance, it's a different game. Sometimes you lose too much charge to the air itself and you need to use corona rings or something silly.

In short, If the 1µF on the output isn't enough to drown out the ripple, it is likely your measurement technique that is flawed somewhere, or your voltage is leaking. Maybe something is dirty.

As for the meter death, again, that's quite possibly due to stray parasitics. There's a large pulse here with a very sharp leading edge. Fourier series voodoo shows us there're actually a lot of AC components here, and, as stated above, a minute capacitance across your "gig" resistor may very well be enough to pass enough current to damage things.

My advice with the metering is meter once or twice to set the charge level, and disconnect it... at least when firing. The EMP alone is reason enough to disconnect nonessential gear. That's what I do with mine anyway. If you hear a better way (and I'm sure there is somewhere) please inform us. A pair of TVS's or zeners or something of this sort may help.

 

[Anthony P]

Thank you for your input. sir. I highly value your advice.

Disconnecting the meter is exactly what is on the agenda this week. I have researched a variety of HV switches and decided to Frankenstein my own relay. The contact will swing open around 2". Also, I have paralleled a couple of 5401s (opposite polarity) accross the meter for some degree of protection. I will let you know how it goes.

An interesting note on metering: My current configuration is a 200uA digital meter in series with resistors. Mathematically a 100 meg resistor should give full scale at 20KV. In reality, 50 megs calibrates the meter almost perfectly. At low voltage (20Vmax test), the meter obeys Ohm's law and 100k gives full scale at 20v. It seems that high voltage has a mind of its own.

The laser itself is actually fully functional. I just feel that the HV meter is essential for even the slightest level of scientific analysis.