VA Tech fusor team visit

[Doug Coulter]

The Va Tech fusor team visited last Friday (1/28/11) to see a fusor run and ask some questions about how to improve the one they are working on now. Hopefully, we helped them! It was quite a full house, since Bill Fain came down for the event, and Joe Jarski was also here for a visit. Always nice to be able to put a real personality to a name on the web!

We managed a somewhat successful run, showing them how we do it, and how we do instrumentation, though time didn't permit the demo of our computer data acquisition stuff on this go -- next time, fellas.

Various pictures were taken, and I'm hoping the Tech team will chime in here to annotate who is who and so on -- and guys, if you like we can create a subforum here to document your project and make a place for any questions/discussion wanted. It's pretty easy with this software, so let me know if you think that's a good idea.

The Tech Crew -- names needed, I stink at that!

Bill Fain took this one outdoors as we were on the tight side indoors to get everyone.
He must be a genius -- how he got a picture of me with my mouth shut I'll never know! I'm a little hoarse from all the talking the last couple of days. Put this one down to humor.

Mouth shut! Stop the presses!

Lee Hall sent me an email with the legend for the pictures:

Names for the first picture, left to right: Mark Pierson, Brad Bersin, Lee Hall, Trey Gebhart, Shawn Murray, Johnny Shelton, Patrick Poitras, and Chris Radford.

The bottom picture, left to right: Mark Pierson, Chris Radford, Brad Bersin, and Patrick Poitras.

So, most of the recent new members are these guys, and I'm hoping they will chime in with names and announce themselves in general, so we really get to know them and can help them better.
Also, they took a good number of pictures I hope they'll send me or post up here. Hint on that -- scale them down as not everyone has a super fast inet connection here. I usually go for 1024 or 1280 width and some compression on jpeg for example.
Videos can be put on youtube and linked from here (see the howto on this site).

[Lee Hall]

Hello,

I am Lee Hall, one of the members of Virginia Tech's Fusion Reactor Project titled "VTec2H Fusion" (Virginia Tech Electrostatic Confinement Deuterium Fusion). This year's project is a continuation of the 2009-2010 Senior Design project that completed the general setup for the fusor. This year, we have 20 student members (it's borderline too many) that are mainly Mechanical Engineers with 1 Electrical Engineer. Our problems so far focus around arcing shutting down our power supply and an improvised gas supply system that requires some updating. Our fusor is an 8 inch diameter 304 Stainless Steel Sphere.

So we know that our power supply (Spellman) has a voltage rating of 40 kV and power rating of 1200W, with built in arc detect, and trip features. Something we have yet to install are ballast resistors to help get us going. We are also beefing up our stalk insulator to hopefully stand off higher voltages. What I'm concerned about most at the moment is our lack of a glowing grid. We consistently get up to 28 kV during our tests and around 10-20 mTorr of deuterium, but we don't see anything with our camera. Part of the reason could be how we are operating it, stalk insulator failures, feedthrough shorts, and our starting vacuum pressure. Currently, our mechanical pump brings us down to around 0.5 mTorr prior to running our test.

What we're attempting to do now is put in a bypass loop around our mechanical pump gate valve to allow for fine tuning on that end. We are also going to try to step down our deuterium leaving the tank from ~5 psig to hopefully something less than atmospheric pressure using a regulator. Then we plan on starting off with a good needle valve and capillary tubing so we can have fine control of what we have going into the fusor. Also, we are in the midst of purchasing a 100 kOhm, 225 W ballast resistor to get us started off and help out our power supply. We are also considering a turbo pump to bring us down to lower pressures as well.

Once we get these things up and running, we are hopeful that we can mess around with the fusor to the point where we find our "sweet spot" and supply some pretty pictures!

[Doug Coulter]

Welcome Lee (and all). What is the current when you see "no glow"? Although it was hard to get rays during my tests of sphere grids, I at least got the grid itself incandescently hot. What I'm trying to determine is if there is something else drawing current that isn't the grid, and so on -- you should hopefully be able to get to vacuum good enough so that no (or very tiny) current is indicated on the supply, and if not, you should look for and fix whatever the problem is.

I'm a big believer in being able to use the mark I human eyeballs -- at low current the X rays are low enough to allow a peek, especially if you do the two layer window thing I suggest -- another piece of pyrex inside the flanged window to take any beating from particles and radiation saves the life of the expensive vacuum seal one. McMaster sells round, thick pieces of pyrex for that. If you can get lead glass, so much the better. At 40kv, you'll not have huge amounts of radiation getting through the tank walls - enough to measure, but probably not do much damage in a quick peek. We also use a mirror inside shielding so we can look for extended periods with no exposure -- the mirror doesn't bounce the X rays very well, really cuts it down but you can still have human eye resolution, rather than some webcam. This is important, mostly at first while you learn what's going on. Then a webcam or similar will allow you to see if things change (and then you revert to eyeballs to see the details again). A full stereo 3-d eyeball view is hard to beat as a "metrology device".

You absolutely need ballast impedance particularly on a new setup, and that's no matter what type of supply you've got. Otherwise, the stored energy in the output capacitors is capable of taking things from glow-discharge all the way to arc (as in arc welding) at low voltages - enough metal is evaporated into the plasma to make a pretty good short circuit at the high peak currents you can get out of just about any supply. The same thing happens with a common neon bulb if you hook it to AC power without a series resistor.

The trick on a new setup (And this is industry wide on high voltage stuff) is to "condition" things before going for anything like full power. This is done via current limiting and very slowly bringing up the voltage -- when it stops sparking at one voltage, bring it up a notch, wait, repeat till you are there. Only then can you use a lower value ballast resistor to avoid voltage drop and input power loss. At first, you'd want to limit peak current to say 1/4 amp under the assumption that the fusor is a dead short and the power supply at fully charged capacitor state at max voltage. You can then go as high as 1 amp limit (eg 40k ohm at 40kv) once things are "fully conditioned". You'll have to do this process at least a little every time you open the tank.

The effect seems to be primarily due to tiny (microscopic) surface imperfections that result in large amounts of concentrated field emission. To get them "burned off" you want to have sparks gentle enough to not make more of these flaws themselves, and most of these power supplies have a few joules of stored energy and can make really high peak currents that will change a microscopic surface projection into a big messy pit - which makes things worse. Once these are smoothed off, you can run much higher inputs safely.

Are you measuring the ID of your sphere, or the OD of the flange (which looks like a standard 8" to me)? It's not a critical number (as far as I know now), except due to Paschen's law effects it will affect the gas pressure vs voltage curve you should expect, as we demo'd here with the two grids, one in a small "tank", one in a larger volume. Remember how easily the 2nd grid out in the big tank would light off at under 40kv, while the one in the smaller space had 50kv on it and would turn off when we turned off the little grid in the big space? I thought that was a pretty dramatic demo of the law in action, as well as the idea that an ion source could let you run at lower pressures and still stay "lit off". In fact, at this point it suggests a test here of simply putting a small electrode at the end of a long insulated pipe coupled into the tank to see how far that can easily be taken for people who don't just have a big tank. It might be the simplest low pressure ion source yet, we'll just have to give it a try here.

[Lee Hall]

Thank you for asking that...the current when we get no glow is...we haven't seen any. This is absolutely giving me fits when I have been thinking about it. From the readout on out power supply, it indicates no current being supplied. Since I do not trust this completely, we are looking to measure the voltage drop across our ballast resistor to determine the current (or if there's a better way we're all for it). To do this we still need to purchase (2)? high voltage probes and a better ohmmeter than what we have since it only goes up to 600 V. But we have not seen the grid heat up or any other sources of light during our tests (even at 34 kV) until we get a discharge. Every once in a while, we'll see the current readout flicker to what it says is 0.01 mA

The large flange we have is 10" OD and 8" ID, but it has been warped a bit during the absolutely terrible welding performed during assembly. This is a nightmare! Right now I have to machine each of our standard 10" OD copper gaskets in order for them to fit.

[chrismb]

Lee, I would suggest you buy yourself a cheap Chinese voltmeter for $3 and run the HV line straight through its current-reading connections. Make sure the voltmeter is fully isolated, sitting on something secure and not-at-all conductive, and it will float with the line. Don't try adjusting the current range whilst the voltage is on. (Did I need to say that? ) Maybe get someone to double-check the safety on that first. I think you really should see a current start up with just 1 kV applied, 2 at tops. It may or may not glow at 1 kV but there should be an inkling of some current there, I think.

You can also use a shunt resistor to ground, though I would prefer to see no resistance in the chamber to ground line due to inductive behaviour, especially whilst you have strange things going on.

Are you sure the power supply is working correctly? Usually, a power supply will show the volts it is about to apply before you flick the 'on' or 'enable' switch - which might be a remote switch or a connection to ground or +10V or whathaveyou in the back somewhere. Just basics to check, so please excuse me if I am covering the simple stuff you have already done but worth looking over again because if the power is really being applied, then you should be seeing something happen!

[Lee Hall]

The simple stuff may actually be the problems we are having! We have a majority of mechanical engineers on our team and only 1 electrical engineer, so he is becoming quite familiar with this system. With our power supply, we are able to set a current limit and use the computer to ramp up voltage based on a voltage step every number of seconds. We know that some sort of voltage is reaching the grid since we have pictures showing an arc from that point. I'm just not seeing the grid heat up at all or a current reading on our power supply. Here's the link to pictures of our setup: viewtopic.php?f=50&t=283

[Doug Coulter]

I believe the trouble then is that your supply is shutting down right away in the initial arc(s) when there's enough gas to light off, so you never see it "up" while it could be drawing current.
The meter has a bit of averaging in it so it will barely flicker during an arc. Did you ever get the schematic/manual? I can ask CliffS for one if you can't. He's a good dude and likes to see Spellman supplies work for us types -- it reflects well on his company after all.

This in turn means a big enough ballast R will get you going. Try 100-200k for starters, and do the conditioning thing -- and expect that to take some pretty serious time, during which you can be watching your feedthrough issues -- you may ruin one set of insulators during that first one -- you should plan on those being easy to replace anyway if my experience means anything. Try enough gas to get a glow down at 20kv or so -- no serious X rays at that point and you can get up close and watch things, then step down the gas and up the voltage, gradually, till all the little arc points are removed. Another possible way (well, you'll always need a ballast) is to start with a ton of gas, say a few millibars, and the power supply set for high volts but low current limit, say 5 ma. Start at say a couple millibars where it will draw current and limit at the few hundred volts range.

Then very gradually pull out gas till it just barely gets unstable and slightly sparky...and let things take their course for awhile at each level. You won't be able to run low ballast R values like my 50k until everything else is just so. Since you really need to get going "at all" so you can use the visuals as diagnostics for the other issues, I'd strongly advise getting a really big ballast for that part -- it won't be power efficient, but it will be far better than what you have now -- nothing at all. I can loan you a resistor or two if that's needed.

You'll have to repeat some of that every time you open the tank, but that first one is the one that takes all the time, sometimes days. After that, it might only take 10 minutes if you've not dinged it up while out of the tank. Just enough to blow off the dust and fingerprints is my experience, which doesn't take long at all. Nicks are another story, though.

I've used cheap harbor freight meters (but prefer the good old analog type) floating way off ground, but this is a school setting and it's all "think of the children" turf -- and that IS a little dangerous. At this point they are not even allowing the students to look into the window at all! You can't really float either the Spellman or the chamber in any practical way to get a sense lead in the ground line, so that's probably out. I'd trust the meter in the Spellman, mine has never lied to me (outside a lack of resolution). Remember, they only started worrying about 30kv X rays when early color TV's had kids sitting close in front of them for hours/day. It's really not that bad till you get up to medical voltages (which is well past your supply ratings anyway). Just remember -- short times, not hours. And also -- there's significant UV in that glow, a lot more than you'd think, so some sort of goggles are a smart move. A couple of us have gotten UV burns learning that one.

Sounds like you need to bring that over for some welding/machining work. If you have some SS scrap, I can teach one or two of you to weld better than that in half an hour (or just do it myself, but then *I* have all the fun!), it's really not hard at all with a good machine (and I've got a really nice one). Warping a big flange like that takes real effort! I've managed to hurt some thin ISO type ones while learning, but that's a thing of the past. I've been known to put a warped one back on the lathe and fix it, too - there's plenty of metal there to work with. I'd have to see it up close to see if there's a way to eliminate the need for gasket machining.

As an aside -- a couple of us have been thinking of designing a tool to re-swage those CF type gaskets for re-use. I must have 50 lbs or more of them waiting for that, and it shouldn't be super hard to manage it since I have a hydraulic press. Just an upfront cost for the die materials, some annealing, and chemical cleaning. Those things get to be real expensive when you use them as a "door gasket" and need to do a lot of opening (like at first). I didn't get to show you the trick I use that lets me use them about 3 times per replacement, but with a warped system that trick doesn't work anyway.

[Doug Coulter]

Oh, note for Chris -- they have a Spellman, the half size version of my big one, so I'm familiar with it. It has an arc trip that shuts it down more or less instantly in arcing, and they were doing this with no ballast or current limit, dumping the whole stored energy in the supply in to an arc (joules), which then turns it off....I simply disabled that on mine, but they are averse to the idea for them (as is Spellman) as that breaks the warranty on the thing, and they paid full price for it. What happens if you turn that off and then abuse it is the minimal series R in the supply (about 5 k ohm) burns up, so the arc trip needs to be there as long as there's any chance of an inexperienced operator using it. I've managed to push mine pretty hard, but then I'm an experience operator, and it's my money...

What happens is, during an arc, joules get dumped. The current limit in the supply doesn't start sensing any current until the load comes back, which means the caps in the supply recharge at full available power, 1200w in their case. Then an arc...then repeat. The cap charge current does go through their series R, and can easily burn it out in that case, in a couple minutes of that.
It's all potted, so burning it out is a big deal to avoid -- they only replace that whole stack/resistor assembly and it's 3/4 the price of the supply. So, it's nice and compact and works well within their (Spellman's) system but if you've got a nasty load, you have to do things outside the supply to keep everything alive. Better to burn up an external ballast R, they are cheap!

[Lee Hall]

So we have started to confuse ourselves when writing the purchase order for our ballast resistor(s). Our adviser wants us to have a connection either the same or similar to what Doug Coulter has on his 50 kOhm ballast resistor. On the Surplus Sales of Nebraska website, we are looking at 50 kOhm tabbed and cartridge type ballast resistors. Are there problems with using the tabbed connection? I seem to remember Doug telling us something about arcing across the tabbed connections. With the cartridge type, what kind of connection would we need to use for both sides if we have HV wire coming in and out?

[Doug Coulter]

The cartridge types are better for only one reason -- no sharp edges to promote corona or arcing to nearby other things. They are also easier to mount, I suppose. I found that fittings for 1" Cu pipe (hardware store or McMaster Carr) fit nicely on these, and with some cleaning and flux, can be soldered on to the ends. I only did that for the end that supports the resistor on my setup, so I can take the pipe adapter off the other end, and fool around with varying the inductance by putting ferrite in there -- things like that (which you won't need to worry about for awhile).

So for the tank end, I used a pipe cap with a 10/32 screw brazed in, to match the end of my HV feedthrough -- you'd use whatever matches yours. For the other end, I used a 1" to 1/2" reducer, a 1/2" pipe cap, and soldered in a banana jack to match the plug on my power supply -- again, you'd do what fits in your setup. I like making it so I can take things apart easy, as thinking you're going to make it never fail, or never want to change it -- is what we call "vanity" in engineer-land and it rarely works out well in science-land.

Don't forget to make provision to blow some air over the stuff, or have a way to watch the temperature at least -- I use a cheap remote IR thermometer from Harbor Freight around here for things like that (and the tank itself). For the tank, I had to add some black spray paint to get accurate numbers, but in general that's a handy tool to have around, and a lot safer than most other ways of seeing how hot something is, since you don't have to touch it - or get close enough to draw an arc to yourself.

And these will get hot because during arcing, the current limiting in the supply doesn't really work -- you dump the whole stored energy in the supply, it charges at full rated power (which in the case of spellman, is usually way conservative, these are rated like the old McIntosh audio amps - -), you arc right away again, repeat -- so the full 1200w or more is turning into heat in that ballast resistor until you stop arcing -- the arc is a pretty good perfect short circuit while it's there. Luckily, these big old 225w wirewounds will take quite a bit of abuse for short times, but you have to watch it, I've melted them (literally). I'd get some of a couple of values. Your final one will probably be about the 50k I use (same sizes more or less). But for conditioning a new grid -- you want to be able to allow very small arcs that blast off the high points on the grid (which can be so tiny they are hard to see under a light microscope) without making it worse by leaving a messy pit there and spraying metal around. I'd suggest more like 200k ohms for that phase of things, but it's not cast in stone or anything, just what works here.