Yeah, that's pretty cool. I'm kind of at the limit of what cables I can bury without hitting something (extensive use of metal detectors was already required) - it's busy under the yard these days, as all my other wiring (solare, generators, AC, DC, hardwire controls, battery and water monitoring) and plumbing is also down there somewhere (often under a metal junkyard). Good to know, it may become handy, and I could perhaps buy another switch and spare up one of the already-buried cat5's (2+ 1 cat6). I know, I should have made a map, but it kinda just grew over a few decades.
There is quite the mesh between the 4 buildings on campus here.
On the other hand, this setup
records, in a very tightly shielded Pi, the video, as well as showing it to me in a split-screen preview with other cams (hint, see the Vivaldi browser that lets you tile tabs).
What happens is that the pi cams record full rez and speed, and I send a somewhat reduced rez and speed preview to the op position, and all those cams fit on one monitor. The other 3 do plots, command and control GUIs I've written, and so forth. When a run is done, those external cams are commanded to copy their vids to the main pi, which then copies them, along with the SQL database it's been replicating all along - to the op position for later analysis. Thus I have a bit of redundancy, which in this case seems a good thing. It's just amazing how well you can shield both ends (the source of EMI and these data aq computers) and still crash them. Of course, EMI isn't the only game here - we are making enough neutrons for disruption and transmutation, and all the fusor-local computers are considered "sacrificial" as a result (and backed up out the wazoo remotely for quick replacement). This setup crashes a PC using ddr3 more or less instantly by corrupting its memory in the absence of any EMI my scope can see - just neutrons/gammas knocking electrons around...turns out running out of "rom" or at least larger capacitors in ram is not stupid in this environment.
Before I got real religion around shielding EMI at the source, I fried two PC's that weren't even wired in - just the energy picked up by the mouse cable/power cable used as antennas was enough to fry - completely ruin - a modern PC that wasn't even turned on, of for that matter, wired into any of this.
It's not like you can completely eliminate some EMI when there's an arc (now rare...as I can make it). Things like ground loops are almost unavoidable, and worse, the stray capacity in the HV cables, feedthroughs and such tends to be VERY high Q, so peak currents (at ~~50kv) can be real serious. So even if I use series impedances, I can't do it for the parasitic stuff.
The end result of all this is supposedly going to be (not yet designed or written) data mining software that can show me time-accurate scope data acq, along with the other stuff (about 12 channels at 10hz or so) and room audio and a couple other cameras - in perhaps short (variable length) time-slidable loops on the collected data, which all winds up back at the op positions in near real rime. That way, if something really works well, even if in a burst and the computers over there all crash - I can know just what was going on up to the instant of the failure. If I'd had this before, I'd be done and there'd be no controversy about whether I'd done it (the huge output and Q) in the first place. Once bitten, twice shy. This time it's going to work. If it's overkill, fine. I put up with crap for quite some time while building this, I mean to put a stop to that - or have to go ahead and hang my head in shame (and...those who leaked against my wishes and created the controversy...will get theirs too).
It ain't science if you can't replicate, here and in other labs. The quest now is to do just that, and we'll be starting as soon as Monday.
BTW, the scope does have a data-dump function, but the tiny uP in there that just shares memory with the FPGA that does the real thing is limited to around 9600 baud. So, for this, useless. Since I can afford the bitrate (and resulting storage space), the video is much nicer. What I'm showing here in the preview is just the reduced resolution and framerate mjpeg. What I get in the end is full HD.
- Here's the remote end. Note, you can't see the half pound of ferrites we had to use along with the shielding to keep from violating the "mere" 500v or so common mode limit on ethernet. Can't really use wifi here - too much noise. A wire that long might as well be an antenna...
The led you might notice on the lower right of the display is for time-sync. The grid camera has one too. One problem I had was mere 2n4904's and 2n3906's wouldn't make a low impedance enough drive for my "clapboard" time sync signal such that emi wouldn't "re start" all the timing. I recently went to a 6 amp fet driver chip for that, which so far has been bulletproof.
And now, gee, if you've used a scope any to investigate something, you always want to twiddle the knobs a little, you almost never get them just so on the first go.
If you have to do a long round trip (here that can involve deep snow or mud some times, just to make it more fun) to try incrementing or decrementing one of the many knobs (and you know from experience that it's usually more than one try even when you're right there and the process is going) - this remote control looks like a very nice thing to have.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.