Getting high voltage into vacuum RELIABLY.

Things at the limits.

Getting high voltage into vacuum RELIABLY.

Postby Doug Coulter » Sun Jan 11, 2015 3:48 pm

Well, I've put a lot of words up here about this one...most of the attempts at least worked for awhile, but there were two real bad issues. One is that hot D eventually reduces any insulator but either pure alumina (hard to get, impossible to machine, and leaky) or it seems, Born Nitride (hard to get, easy to machine, leaky).

I finally went with BN and pyrex. I sent this in email in response to a private question, but it's the real stuff and it's my fault for not making more noise about here, in public, so here it is:

Ah, sadly, yes, I have all too much experience with things that don't work, and I should have done a better job documenting that and what DID work - it was holding me back for ages to have to keep replacing failed feedthrough parts.
Now it's all perfect (so, I got lazy and didn't write it up much), but you might not be real happy with what I'm going to tell you...it was expensive, but only once and now looks like it's going to last forever, so in the long run, cheaper by far.

Tubing dimensions aren't super-critical, which is good because even the highest quality tubing available isn't quite round. If you're inside a mm, you're good. Maybe a couple mm. It's not the problem, that part works pretty well, though if you use large tubing, you'll have to make a clamp to keep the glass from slowly creeping into the vacuum tank...from an inch (25mm) or so up...you get problems, even if you don't grease the O ring in the tubing coupler (and well, I do, usually).

On the inside of the tank, the main issue revolves around hot deuterium reducing almost everything you can get as gas-tight tubing that is also an insulator. Quartz becomes silicon, arcs then shatter it. Ditto pyrex. You can't get a single tubing thick enough to really hold off all that HV either, and as I found, stacking them (telescoping tubing one inside the other) also does not work as the voltage refuses to divide up in any sane manner, and due to the dielectric constants, this thing stores enough energy to shatter itself when it "punches through" - you've got a capacitor whether you like it or not. AL-99 might work, but it's a little porous, and very very expensive - most alumina is lower purity and contains basically clay to bind it, which has all the same problems as glass does - it reduces and becomes conductive.

You won't need to go as big as I wound up doing unless you want to go over 50kv - since I was paying out the ass anyway, I went big for the 100kv supply (not yet tested, but it should work).

Here's what I wound up with that actually works (I'm using CF, copper gasket flanges since that's what my tank has, but the rest should be the same).

1.5" thickwall pyrex tubing from quartz.com. Pyrex is the *most* conductive of the glasses, which is actually a good thing here. Outside the tank, it helps smooth out the voltage gradients just like a bunch of corona rings, without the hassle of having them. Ditto inside the tank, since where it goes through the tank, it's clamped to ground (literally in this case).

To prevent the mentioned reduction of the various things in glass that reduce to conductive materials and cause failure, this tubing doesn't extend all the way to the grid. Instead (and this is the $$$ part) I bought and machined a 12" long 1.125" diameter BN rod - bored a 1/4" hole in it for a copper conductor (for both electricity and heat) to the grid, which sticks out past the BN about 1/4" for my grid mounting - it's tapped for a screw on the end to hold the grid on.
The BN itself sticks about 2.5" into the tank further than the glass does. This seems to prevent the glass from picking up a charge and attracting D+ ions that destroy it - the BN takes all the pounding, and it isn't decomposed by hot hydrogen one bit - so it looks like it's going to live "forever". After about 100 hrs of running, it picks up some junk sputtered off the grid etc on the end, but being as soft as chalk, it's a one wipe with fine grit sandpaper operation to clean it off. I haven't had to yet.

I have the grid end spaced an additional 1/4-3/8 inch further from the end of the BN. Small changes here seem to make large differences, and I'm still playing around with that. Too close and it both sputters too much onto the FT BN, and doesn't work as well. Too far, and there's a lot of "not grid" exposed and that attracts D+ and wastes energy accelerating D that isn't going to fuse.

A big advantage of this, at least in my setup, is that you can push-pull/rotate the FT if you want to. We've found here that how close to the end of the sidearm our main grid is has a big effect. I step milled some shims to place between the glass clamp and the tubing coupler so I can have a repeatable placement. It was worth doing.

The glass clamp I made out of a 3" diameter, 1" thick piece of HDPE, bored to fit the glass OD. It then took 6 large setscrews, padded with two layers of chinese inner tube rubber to apply enough force to keep the glass from moving without cracking the glass - I used 1/4" diameter hex head screws for that.

Since the tubing coupler flange is a 2.75" CF (which with a little filing and boring will hold a 1.5" hole), and is 2.75" OD, I can and do use a piece of 3" OD schedule 80 (extra thick wall) tubing over this entire mess, held to the flange with setscrews (the thick wall makes that easier) and of course, centered by the 3" glass clamp as well. It helps with safety around the high voltage, and provides a place to wind a secondary - my primary ballast is now just an inductor wound on 1.5" tubing under this. In "breakthrough mode" where we made crazy amounts of fusion, we used this transformer to make a hartley oscillator between the main grid and the smaller ion source grid out in the main tank.

Over that entire mess, I have an 8" ID PVC pipe with copper screen soldered together all around it. When I had a resistor in the ballast inside the 3" pipe, I also forced air through it to keep it cool, but with the total lack of arcs (and most of the ones I had weren't very visible) - I don't need any extra ballast anymore. That's great, as it was a big power waster and a voltage drop I had to account for in my plotting math etc. So, my entire HV feed is in "coax" with no ground loops and this is not only a lot safer for me - it's tons less EMI to mess up my other measuring gear. After losing a $2000 scope to that...well, you become more careful.

Boron nitride rod is super expensive. I got the lowest grade, which is slightly hygroscopic, and thought that would be an issue. It isn't in real life. A week or so under pumping and it's all gone, and doesn't come back with short tank open periods. I got mine at McMaster-Carr, which sadly does no export business. It's looking like a very good investment at this point, though. A whole lot of problems just went away...seemingly for all time. It's also very thermally conductive which helps get heat out of the grid...came out very nice. That foot long drill to bore it wasn't cheap either, but...you gotta do what you gotta do. I'd suspect that 1" glass and a correspondingly smaller piece of BN would do for most fusors. I had to turn mine down to fit the glass tightly, which was also a problem as no glass is quite perfectly round...I had to kind of "screw it in" to avoid lengthwise arcs, shown on my youtube channel for an earlier attempt with too-small BN diameter. But it looks like I'll never have to take this one apart again, so...I don't care.

Doug


On Sun, Jan 11, 2015 at 2:44 PM, Philipp Windischhofer <philipp.windischhofer@student.tuwien.ac.at> wrote:
Hi Doug!

I’m currently pondering around with a decent HV-feedthrough for my fusor setup. I’ve come across those CF Quick-Connect Flanges / Tubing Couplers — aren’t you using a similar system (based on what I saw on the forums)?

The only problem that keeps me from pulling ahead: I don’t know anything (and the guys at the German vacuum supplier where I get most of my stuff couldn’t tell me either) about the tolerances the glass tube has to comply with in order to still get a good seal. Do you have any practical experience in this respect? What kind of tubes did you use — standard quality or high precision ones?

Any input would be highly appreciated!

Philipp



--
Why guess when you can know. Measure!


I'll add some pix to this ASAP. THIS IS WHAT WORKS. A whole bunch of little other issues "just went away" like near-invisible arcs that required a big ballast resistor, shattered glass in the tank (danger, turbo present) and so on, and now it all just works...which is the goal of good engineering. What's funny is that reading Miley's book in a private writing of Farnsworth, he predicted what I now find is the best solution ever. While I don't agree with all of his work (we've measured some different things than what he was forced to guess were going on due to lack of metrology - we have better tools now) - he was dead on about using BN for the main insulator.
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.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA

Re: Getting high voltage into vacuum RELIABLY.

Postby Doug Coulter » Sun Jan 11, 2015 4:25 pm

Sadly, the protective front SS screen and my dumb camera's autofocus prevented me from getting a "full frontal" on the feedtrough, but I was able to get a picture from above using the raspi project. It sadly does not show the end of the glass, which doesn't get in there as far as the BN does, which actually is why this works...but you can get an idea of the overall scale of things from this. The main grid is in that sidearm which is 5.9" ID.
FusorInside.jpg
Fusor innards from above

At the lower left we see the end of a 2.4 ghz ground plane antenna which we thought might be interesting. Fail, but it makes a good faraday probel In the lower right is the end of the ion-source grid, basically another fusor type grid, but without the accuracy, and much smaller (to match the power I have for that). That one runs at fairly low voltage, and so far, the dual layer hand-blown quartz over anodized aluminum is holding up. Since it's in a much larger space, it only takes 5-10kv to "light off" when even 50kv won't do the same for the main grid in its smaller confines (Paschen's law). This gives us a plasma "tube" (or you could model it as a high threshold super high voltage high power gain enhancement mode p mosfet) that has a lot of gain and some interesting possibilities, though initially we just used this to be able to run lower gas pressures so as to reduce scattering pre-focus and increase the net Q of the fusor. Turns out there's more you can do, but I'm not ready to report much on that yet - it was enough of an increase that I have to go to remote control to run in that mode...and I'm doing the boring work to remote it over this winter.

At least you can see the end of the fat BN rod I bored as the main insulator here. Funny how after years of trying things that "should work" the "overnight sensation" came via a rather boring experiment (pun intended) ;) .

Here's a short video from the outside:
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.
User avatar
Doug Coulter
 
Posts: 3515
Joined: Wed Jul 14, 2010 7:05 pm
Location: Floyd county, VA, USA


Return to High Voltage, High Power

Who is online

Users browsing this forum: No registered users and 1 guest