2003-01-14
Choke Charging Experiments
I had a little spare time on Monday afternoon, so I tried out some of the ideas expressed in the previous update. I built the testing device I outlined, a quick hack made from a fragment of PCB and some of the 470p 6.3kV capacitors (I have over 400 of them now!):
It allowed me to test various resistor types, and potting materials I had laying around the place. I tested the whole range of resistors I have, 1/8W and 1/4W devices are not surprisingly, incapable of operation under this kind of stress, the arc over immediately. 1/2W devices work fine, all the way to 15kV, when they start arcing around the device, leg to leg. The 1W device shown fails at only 7kV. The encapsulation of these devices is not of the same quality as that on the 1/2W units and punches through.
I tried painting the resistors with polyurethane varnish, several coats dried with the heat of the soldering iron held near the resistor. It did not significantly improve the performance. Heat-shrink did not help, it just promoted arc tracking, providing a carbon rich material to set fire to and stink out the room. I'll have to try liquid electrical tape (which is quite expensive at about $15 for a small bottle if it doesn't work) and epoxy. Bathroom silicone is another material I am yet to test.
Similar to my plasma diode detector project, I wound some wire on the 10M resistors, but this time I used a coat of varnish to hold the turns in place and seal the solder joints. These proved very robust in the tester. No arcing, no burn up, they looked promising, probably because very little voltage was actually dropped across them, and that which was, only appeared for a very short time, their inductance is low.
So, I went to work winding four of these units to built a small Marx generator for testing. My first choke charging Marx generator was born:
The only problem is, it doesn't work very well! I can only assume the inductance is too small. The gap tuning is very difficult, which at first I put down to the low voltages and only three stages, but the more I played with it the more obvious it was that it just wasn't going to work properly.
At first the gaps fired independently of each other, seemingly at random, resulting in no travelling wave erection and no output. After some very fine tweaking of the gaps and the ground return electrode it actually started working properly, but it was a very hit and miss affair, with a spark length only slightly longer than a single stage!
You can really tell when it is working by ear, all the bangs come at once with one big bang, any jitter is easily detectable. The output gap spacing is critical, too close and a constant arc develops, limited by the 20M resistance in the loop to a violet glow on each electrode and a fine purple filament that joins them and writhes around with a crackling sound, typical of a starved arc on the verge of collapse into corona alone. Too far apart and erection just doesn't happen, the gaps fire at random, with the occasional two-stage erection (top gap only) managing to just cross the final gap with a weak crack.
Here is a dark picture of it operating correctly. I did find that the output wire inductance made a big difference. I started with a thin alligator clip lead like usual, but it gave me a nasty nip through the insulation, indicative of significant voltage being developed across it, yet with a piece of de-soldering braid I could hold it in my bare hands near the struck end with no sensation at all. I replaced the braid with a piece of 1.5mm Cu wire from a length of mains power cable. Again, I could safely touch near the end being struck with no sensation. The insulator is there to hold the gap distance constant.
The ground-surge phenomenon that has so often blown my PSU may come from this effect, a huge current is present as can be seen in the shot, the neon/coil widget is glowing brightly, with moderate inductance in the discharge path a huge transient into the ground could be developed as the discharge inductance attempts to recharge the stack in the opposite direction, I am not sure how the gaps behave, it all happens so fast, without better test equipment I can't be sure how much voltage reversal is going on and how many times the electrons slosh back and forwards in the tank circuit before the gaps extinguish. It appears in any case that minimizing the ground path inductance minimizes the damaging transient.
Note that the glow is brighter than in the alligator clip lead shot, and this device has less stages and obvious health problems. I am not sure if that is conclusive or not, but all my future experiments will have heavy ground wiring with minimal inductance. Note also that the glow is on one electrode only, which suggests that the circuit is not ringing much once the inductance of the discharge path is connected with the capacitance in the stack. Still there is obviously high losses in the charging inductances once erection is established. I'll try again with bigger inductors. Perhaps some modelling is in order, but where do I find a model for a spark gap?
Leave a comment on this article.
Parent article: Marx Generator.