Today I took a step back and measured the current going through the coils.
Previously I measured 1,200 amps going through the coils with the capacitors charged to 450V.
My setup is a little different now: There are 2 meters of cabling + feedthrough between the coils and their power supply.
For all of these shots, the capacitors were charged to 450V. The shunt resistor shows 100mV across for 100A through. Multiply the voltage by 1000 to get the current in amps.
Surprisingly I’m seeing significant variation of current for identical conditions.
The most current I saw was 1,095 amps:
But with the same conditions I saw this much lower 344 amp current:
The median current reading was around 734 amps:
Disconnected, the coil shows 0.8 ohm of resistance.
I’m rather surprised by this variation. What could be the cause?
I tried some other conditions.
With the capacitors charged to 200V, I got 560 amps of current:
I also tried charging 5 out of 10 capacitors to 450V. With an average current of 500 amps.
Prometheus Fusion Perfection 
The difference can sometimes be caused by the way the cro leads are arranged (probe/gnd). If the leads are paralell with the feeds from your caps some of the magnetic field (from the current) loops the probe or ground lead inducing a voltage into it. Best to replace with direct (capacitively coupled) coax keeping centre conductor and earth as short as possible and at right angles to the direction of current. Hope this helps.
Dustin.
I agree with Steve about improving the instrumentation and making sure that the grounds are in a star configuration. There are a variety of tests that can be made. You can get a “ground truth” (sorry about the pun) reading by shorting the sense resistor and looking at the residual signal; or just putting the high side of the probe on the ground side of the sense resistor. Both of these give an indication of how good your instrumentation grounding is.
Having said that: I think the problem is probably in your coils. I think they are probably moving.
I guess a good test is to run some tests at a much lower (1/10) current level and look at the consistency; and compare that with the consistency at your nominal level.
Looking at the damage to your coils it wouldn’t surprise me if the loops are banging around and rubbing. You can always check the quality of the winding’s by hooking up a good pulse generator (1-100ns rise time) in place of your cap/thyristor and and comparing the voltage waveform with a “good” waveform from undamaged coils. Be aware that you should be looking at events on a much shorter time scale.
Ray
Ray I think you are onto something with the coils moving.
I know for a fact they can move quite a bit. Guess I should really tie them down tight.
I don’t exactly do a lot of daily circuit analysis (and if I need to I usually cheat and use online simulators like http://www.falstad.com/circuit/ ) but I don’t think those numbers ad up:
The coil should act like an inductance and a resistance in series (ideal case). Correct me if I’m wrong, but shouldn’t the (DC) resistance determine the maximum achievable peak current, with the inductance only acting to widen the pulse (by resisting rapid changes in the current)? If so, then 0.8 ohm coils should be able to get a maximum of 450 volt / 0.8 ohm ~ 560 ampere…
So that is the one of the easiest things you can test, the DC resistance of the coil before and after any shots (if your meter is comfortable going that low?). This shouldn’t even depend on the coil shape and should only change with the temperature of the wire. If it varies a lot then there could be bad connections and/or shorts.
Great point! 560 ampere is the max we should see just with DC resistance!
How much wire has gone into the coils? Using the resistivity of copper (~17nohm m at room temp) for 1mm diameter wire gives 22 mohm / m which could be used to calculate the theoretical resistance. 0.8 ohms would mean almost 40 m of wire – I’m guessing that’s not the case :)
(Also, the >1000 A peak currents you’ve seen imply that the actual resistance is at most half that)
Wiggle the feedthrough connectors and the resistance varies between 0.8Ω and 0.5Ω.
The chamber is closed so I can’t access the vacuum side connectors at the moment.
Kinda disappointed! These connectors are gold plated for good connections!
Lab note: Touching my multimeter’s probes together shows 0.4Ω.
We are at the edge of what this multimeter can measure.
You can get a more accurate resistance measurement by putting an amp or so through the series resistor-coil assembly and calculate the resistance from the ratio of the voltage drops. In practice the coil resistance will increase as the coil warms up in the vacuum, each succesive pulse will have slightly less current and the pulse may get longer (t=rc), you may even be able to use this to calculate the temperature of the coil.
In your photo, the cro ground runs parallel with the resistor shunt for a short length, the magnetic field from the current will induce a voltage in the ground which will offset the probe voltage. If you disconnect the probe altogether and connect the probe just to the cro ground black clip (nothing else) and lay the cro ground lead on top of the shunt, you will still see the induced pulse. All wires have to be twisted and at right angles to the current path to minimise this effect.