David F. volunteered to help out Sunday:
We took another crack at running the DAQ in the face of wicked EMFs coming from the fusor.
David picked up a shielded USB cable for the DAQ:
With the DAQ close to the fusor, but not connected we were able to instantly crash it:
Next we put the unconnected DAQ in a makeshift faraday cage made from a small cabinet:
The cage and the lid are grounded:
Surprisingly, we were able to crash the DAQ pretty easily in this configuration! It’s possible that the panels that make up the cabinet are not connected electrically (seems unlikely). Next step is to upgrade the faraday cage to a made for electronics chassis, preferably rack mounted.
The only way we found to keep the DAQ from crashing is to keep it about 2 meters away from the fusor.
David also diamond filed the ceramic tube to fit inside the HV feedthrough ceramic:
Prometheus Fusion Perfection 
Out of curiosity, how are the grounds for the DAQ connected?
If you haven’t already, I would be interested to see what would happen if you disconnected the USB cable shield from the computer ground, and attached it to to your main lab ground instead. Otherwise, you could just be transmitting the noise from the reactor straight into the chassis of your computer, which would explain the crashes.
Anyway, good luck, and keep up the good work!
Interesting idea. So the premise is that the computer USB ground just can’t handle the spike.
A note: it’s not the computer that is crashing it is the DAQ. Without restarting the computer I can unplug / plug the DAQ and it works again.
Hmm, just the DAQ, eh? Still, I’d still like to see what would happen if you grounded the USB shield right at the DAQ end. If you do, you want to disconnect it from the computer end to prevent a ground loop.
Another question. Is the ground line on the DAQ input connected to the USB shield?
Also, if you run your faraday box test again with the USB shield connected to the box, It might work better. I’ve seen noise crawl up lines I thought were well grounded.
An experiment, if you have the time. Build a male->female USB junction so you can probe all the USB lines with your scope while it’s running. Throw it on the computer end, ground your scope to the computer chassis and you can see what kind of noise is making it back up to your computer.
During a project a few years ago, I built a rig with a USB camera running right past a big stepper motor. Same as your setup, running the stepper blew the camera connection away, and I’d have to unplug/replug before the camera would work again. Using the setup I described above, I found out I had ~50V transients at my computer. Not a healthy thing, and after a week the USB port didn’t work very well.
In that case: If I were you I would get some HF toroids to use as Balums.
Ray
Is it possible that it is not EMF, but xrays or other kind of radiation that is zipping through the metal like paper?
A water jacket might eliminate this – putting the DAQ in a plastic container inside another filled with water – or lead sheeting in your faraday cage?
It looks like the cage is steel. I think RF can get though steel due it its relatively high resistance. Perhaps try an aluminium box, or lining it with copper foil.
Hello nophead! I’m a big fan of hydraraptor!
Ive been battling this problem for a while for CE certification. The instrument would lose comms with the pc. swap the usb port to a different one on the computer and it would work again. The fix was to short the usb connector to ground on the instrument (daq to usb gnd) and use the longest cheapest usb cable (they are more lossy for spikes, I used farnell 1201898) and stick ferrite cores (farnell 9415700) on each end of the cable. This used to fail with 200v injected spikes and now I can inject 2kv spikes without losing coms. Not guaranteed but worth a try.