Gas handling for fusors and similar devices

How to get to vacuum, what the classes are, and what is needed for what job.

Gas handling for fusors and similar devices

Postby Doug Coulter » Sun Nov 21, 2010 12:34 pm

Well, this is not quite vacuum tech, but closely related, so it seems to fit here. When doing fusors, there is a need to have pretty tight control over gas pressure and purity, or you just can't make it go very well, not to mention having too much gas makes things hard on your power supply, and too little makes it go out, and the range of "good" is pretty small in most setups.

Most fusor gas handling schemes so far published count on somehow flowing gas through the system, and trying to balance the input rate with the pump out rate. I did that myself at first, and oh boy is that hard, even with some adjustable throttling on both ends of that equation. It's kind of the same idea as connecting two inputs to a perfect integrator (or maybe a leaky one!) and expecting to be able to balance so perfectly that the output doesn't change. As they say on Slashdot, GoodLuckWithThat. I've heard of one guy doing it with a mass flow controller and some feedback from a pressure gage, but small flow controllers are expensive, and that still wastes a lot of fairly expensive gas.

Now, I'm lucky to have a couple of very tight vacuum systems, with turbo pumps on them. So I've found another way to run hands-free for fairly long periods, and not need anything real expensive or hard to get to do it, so I want to share that here. For this method to work, you need a pretty tight system that's been well out-gassed. Running the fusor a bit and pumping down again gets that done pretty quick if needed. I usually just keep the thing down at near base pressure when not in use, which helps.

I have managed to find a small solenoid valve that can go between my turbo pump and the forepump. I bypass this with a normal valve, so I can just have that operate as normal, but for this, I close the manual valve so the only way for gas to leave the system is through that solenoid valve. ( I should say "BillF managed to find" here). As I have a turbo, it will not do anything evil if I close offf the foreline, just draw more power when there's a lot of gas there. This is easily solved by running it at about 1/4 the nominal speed, enough to be good enough to remove gas at this pressure without drawing a lot of current to spin, and building up only a little pressure in the foreline, so that's all happy.

Here's a picture of that side of things:
PumpPlumbing.jpg
Pump Side plumbing


Yeah, sloppy, clipleads, battery, junky. It's not broke, so I ain't got 'round to fixin' it purty yet, but you can see the AC adapter sitting there to make it nicer. I had to make an RC snubber for across the solenoid to keep from burning out the button I use to pulse it (actually a microswitch I just slap down on the operating desk for an instant to remove a little slug of gas).

I close the big manual valve about the same time I tell the turbo to start spinning down to "standby" speed, which is set lower than normal for that use. Then I start letting in gas, which helps it get to the desired speed a lot faster than just depending on the (minimal) bearing friction to get there. I usually shut off the gas flow when the chamber pressure is about in the right range for running. What actually happens, is that as the turbo goes slower and slower, is that some of this gas is compressed behind it, and manages to get back through it into the tank when it finally hits the slow speed desired. That's fine in this case, and tends to help purify things, as the compression ratio (at any speed) is far higher for air type gases than for hydrogen (of any isotope) so it's kind of a purification step. To enhance that, I generally run a low power glow discharge while this is happening, to blast all the heavier gases off the tank walls.

Once it's all happy and stable, I then turn on the forepump, and pulse the solenoid till things are just right. In my case, I'm using a Pfeiffer PKR 251 ion/pirani gage that doesn't have enough resolution for doing this all that well, so what I really do to know if I'm "there" is look at the current in the glow discharge I'm running, or more accurately, look at the voltage the current liited supply is at, and go for a voltage I know is right for good fusor operation -- this is a much more sensitive way to get the pressure right than reading a 2 digit gage.

Now I can run for quite a long time before feeling the need to do anything about gas supply, and in fact have run long enough to start seeing reaction product lines on my mass spectrometer!
That's pretty cool. Of course, there are times when you need to let in more gas and repeat this process, so now we go to how I do that on the other side of the equation.

Here's a pic of the source end of things:
Dsource.jpg
Source plumbing


The gages are still reading from the fact that I used this Friday. Actually, they'll be reading forever -- this system has NO leaks at all now. The day-glo thing is a handle I made for a tiny needle valve from McMaster, but truly, it's not tiny enough to do the flow through thing anyway, and I now just use it on-off. This is soldered to one end of a 1/8" OD, 1/16" ID piece of copper tubing that goes off to the right, around the back, and is shown in the next picture. At the back of the tank, there is a glass insulating break in this, with pyrex expoxied to both of the Cu tubing's involved. The total volume of this is around .1 cc or so (which is important).

Here's a picture of the other side:
TankEnd.jpg
Tank back end, gas delivery


There's a lot more going on in this picture (some not related, but good to know about). For now, the Cu tubing comes into the screen from the upper left of the picture, goes through the glass inuslator to another piece that winds up going in the end of my microwave ion source and on into the tank. At that end there is a 2" (roughly) piece of .005" ID cappilary tubing to slow gas flow, not needed now, but very badly needed when I was trying to do this the old way. Letting the gas in through the ion source theoretically means the pressure is a little higher there and letting that work down to lower tank pressures, but it works well enough without that feature, and I now do have troubles with the gas in the glass tube lighting up from the ion source "pusher" voltage, that I'm going to go to just putting it in the tank elsewhere.

The point of some of this is it's no longer important just how long the needle valve is open. It more or less instantly pressurized the Cu tubing, and that then slowly leaks through the cap tubing into the tank, so the timing for the valve opening isn't real critical -- a second or so open lets in one standard slug of gas, which does take awhile to make its way into the tank, but not so slow it's a bother. If that turned out to be too much gas (usually), then I resort to the pump-out solenoid till it's just so again. Done! Boy is this ever a relief from trying to "fly" this thing by balancing in and outflows -- it will stay good for many hours this way without any attention at all.

Since it's in the picture, I'll describe some of the other stuff you see. The main HV feed to the main grid is inside that screen, and my ballast R is the brown thing you can see peeking out from the PVC pipe. There is a plastic hose entering that pipe from a sleep-apnea air pump to keep the R cool and keep ozone from building up. The screen is there to give any arcs a place to go that isn't me, and you can see the cavity tuning screw for the microwave ion source poking out through it.

The glass with resistors in it behind this is my ballast for the 2nd grid, which sits out in the main tank. While pumping down, I set the supply on that to 12.5 ma current limit, and 40kv, but I stop removing gas when it hits about 20-25 kv in current limit, that's the magic spot. At this pressure (around 1.6-1.9 e-2 millibar read on the gage) the main grid in the 6" sidearm won't light off by itself with 53 kv on it -- Paschen's law! So I use the 2nd grid as an ion source (right now, the wire to the puller electrode in the microwave ion source is broken inside the tank, and it's not doing too much for me). There are for sure "interesting" things one can do with the 2nd grid and not-DC that I've discussed elsewhere and will have a lot more on later....

There are some more words on this here is you want more detail on the operational steps. The second post gives a step by step.
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: Gas handling for fusors and similar devices

Postby Doug Coulter » Sun Dec 26, 2010 3:05 pm

Ok, so the above post shows a system that works in practice. I recently was asked to run the numbers for another system, and in my (virtuous, in the Perl sense) laziness, I decided to repeat the calculations here for all. Some guy who enjoys algebra could solve this cleaner, but....this will get you pretty close if you use the same assumptions.

Let's assume a cylinder fusor, 6" ID by 1 foot length, no warts that add volume (there will always be some, but that works in a direction to make this easier, so no worries, or you can do the math for the warts you have and add in the volume they have).

So we are trying to get a certain amount of gas into what amounts to a pipe with flat ends, starting with "perfect" vacuum. In practice, e-5 or e-6 mbar is "perfect enough".

This has ~5560 cc or so volume, using the old Pi * r2 * length formula. If you have nipples, flanges, and other warts, calculate them and add that in.

Now, for a "normal fusor" we want to hit around 1.9-2 mbar, or 1.9-5 atm. That's a pretty large expansion ratio (we assume we are starting from perfect vacuum, where in practice we're only starting at say 1/1000th of the desired pressure, but that's good enough for government work and hand grenades). Supposing we want to let in gas that is starting at 1.0 bar. In this case, we'd need 5560 times the amount of gas it would take to get to pressure in a 1cc volume, which would be just that number above 1.9-5 cc at STP input.
This isn't much gas at all, so we're glad we have a decent sized chamber here. For our case, we need 5560 times that much, or 0.10564 cc, still not a heck of a lot! Now, in my systems, I don't run the D out of the requlator at 1 ATM, I run a little pressure, call it 8 psi gage. That's about 1.5 ATM (absolute) so we divide the previous result by that, and we wind up with 0.070426667 cc we need. That's not a lot of gas! How can we possibly accomplish this?

Lets assume we have a near perfect valve (tiny needle valve from McMaster that has a 1/32" needle and almost no wasted volume inside it). We want to hook it to a reservoir we can fill at regulator pressure, that will then leak into the main tank slowly. So we now need a resevoir that has the desired .07 cc volume, which looks tough, but maybe isn't so bad.

Here's some numbers useful for designing your gas inlet system then.
1/8" Cu tubing (which is 1/16" id or a little smaller) has .019 cc/ cm of length.
.020" ID SS cap tubing (the fattest we have) has .002 cc/ cm.
.007" ID SS cap tubing has 0.000248287 cc/cm
.005", skinniest SS we have, is 0.000126677 cc/cm and is what we're using
for the slow final leak.

As luck would have it, these available cap tubing sizes all are 1/16" OD, and soft solder nicely into the Cu tubing with a slight amount of prep and reaming. You must either soft solder or use epoxy or risk oxidizing the inside of the tiny hole to the point of clogging the pipe. We got some "universal" flux from McMaster that has ZnCl, HCl, and HF in it that makes soldering the better choice here (not as brittle as epoxy), else SS is extremely difficult to solder to.

To use any of this cap tubing below .020", you have to cut it, then electropolish the ends to get the hole open again. Trust me, you're not going to pull this off with sandpaper or filing under a microscope, I've tried that, it will not work. We use various electrolytes for this, and it seems fairly uncritical (for this) what proportion of what is used in it. For super polish it gets pickier and the current density is pickier too, here we just want to eat off the end of a pipe, so not such a big deal. We use a mix of phosphoric acid (ice machine cleaner), some sulfuric (not much), and some glycerin (or glycerol if you are a chemist). Sometimes toss in a bit of citric acic which does help and can be bought at some exotic food stores (Indians use it, and call it Limbu-Ka-Ful). It's cheap and a nice way to make fake lemonade too...

Here are some papers on electropolishing to help you find a formula. I've tried most of them, and for this, heck, you could just make a batch of each and randomly mix up a final solution from a little of each and it will work (I did this, actually, once I found out they all worked pretty well and I needed more -- I just mixed them all!).

electropolishSS.pdf
(528.94 KiB) Downloaded 469 times

EpolishCitricglycol.pdf
(248.56 KiB) Downloaded 470 times

epolishglycerol.txt
(40.34 KiB) Downloaded 431 times


Sorry if you find the text file has no hard CR's in your viewer, that's the difference between linux and everybody else sometimes.
These solutions tend to last a long time unless you're eating a lot of grams of metal. They'll turn green real fast from the Nickel in the SS, but actually seem to work better after a little use.

To do this, it's pretty simple and pretty fast. You get a beaker or other similar container, and put a lead cathode around the outside of the inner surface. Nothing special needed here, but lead isn't eaten by the other stuff in the mixture. I just cut a strop longer then the inner circumference of the beaker and put it in there with a wire soldered on it (which will fail due to being eaten eventually, but not that quick). You put solution in there, and then suspend your cap tubing so it's pointing straight down into the solution, maybe 1/16" or a little more under the surface. You then connect a DC power supply so the SS tubing is the positive electrode, the lead the negative and put a few volts on it. This will draw some fairly serious current, maybe an amp or more, and fizz. Just walk away. When it stops drawing current, due to the surface tension and fizz issues, you will have a pointed end on your cap tubing, with a nice clean hole right in the end. Flip it over and do the other end. Then take it to some compressed air and blow out any solution that got in there, else it will solidify in there and block the tubing. Takes only a few minutes (under 10) per end to do this. Further, due to the pointed end, it's now a lot easier to solder into another pipe, as the solder won't tend to wick out into the increased clearance around that cone and plug the end.

Here's what to expect the end of the tubing to look like when you're done. This was actually a gas-dynamic ion source I used to use, with a very tiny gap to ionize gas that hadn't had time to expand to the main tank pressure yet. I modified a 6mm PPC rifle brass shell for the outer conductor.
GasDynHead.jpg
What to expect


So, what we use in our big rig is a tiny valve, a bit of Cu pipe, which is our reservoir volume (and serves double duty to plumb the gas from where it's handy to have the valve to the tank inlet), then about 1.5" of .005" ID capillary to be the final slow leak into the system. This makes the valve-open time a heck of a lot less critical as it takes a couple minutes to effectively dump that tubing reservoir (which fills instantly more or less) into the tank to the point where tank pressure isn't noticeably rising any more. Kind of the opposite of the "short-fat-shiny" plumbing one normally uses around a vacuum system.

Bingo -- this isn't super hard, and now you have a way to dump a known, fixed increment of gas into a system. If you need more, you can repeat the process more than once, or hey, design things so your reservoir is about the right size to get it done in one shot. By adjusting the pressure on the D regulator, you can adjust the effective amount of gas over a decent range besides.

Now, in an ideal world, instead of a slow leak into the tank after the reservoir, you'd have another valve there (on-off) so you could do this faster. You'd also make the resevoir smaller, so you might take a few shots to get to pressure, but have some "resolution" in your system. That's too much work for me, and most valves have too much lost volume to make it practical -- the valve alone would have too much volume for this, the amounts of gas needed are tiny indeed!

Now, to have a truly cool setup, and to be able to fine tune below this resolution as well as deal with real life you need a little more. First part of fusor run heats the gas, blows contaminations off the walls, and generally means you need to do this a couple times to get to perfect pressure and purity at the same time.

So, in essence, we want a way to take out a slug of gas too, leaving the system sealed off otherwise. I show that in the post above. I have a big fat valve in the foreline to open when I just want max pumping (between runs for example) and also a tiny solenoid valve "around" that to let gas out of the turbo output and into the forepump when the big guy is closed, in pulses.
By spinning down the turbo, we can arrange this so we don't take out so much gas per short valve-open pulse. Expressed very crudely, the compression ratio of a turbo at full speed might be 107 or so to one, and when stopped, it's 1::1, which is quite a range. This will be less for a straight turbo than a turbo-drag, but the idea is the same. The compression ratio is very roughly related to about the square of rotational speed, so it's fairly easy to find a number where the valve you can find will work, if you also adjust the effective reservoir on the turbo-to-valve and valve-to-forepump lines as needed. Here, we find we can get that AF-16 valve to cycle in about 100 milliseconds or so (the RC snubber I put across it actually slowed it down to that speed!) which makes this easy to get in range with the turbo going something like 20% of full speed.

I have found that valve for sale new, but it's really pricey - $713 at Kurt Lesker. Since it might not be quite rocket science to make these -- I may start doing that, I can't count on lucky scores off old leak detectors to provide all of these I'll ever need (or y'all either) and no one wants to pay that for a valve. If anyone knows of a valve that will take from a couple atm to zero, no leaks to external air at all, and doesn't care about pressure drop across it to work, and can run continuously (so you don't need the other manual valve at all) -- let me know, else I will put this on my infinite to-do list. This appears to be a hole in the market big enough to drive a profitable product through if you can make money on small volumes. (pun intended :lol: )
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 Vacuum Technique

Who is online

Users browsing this forum: No registered users and 8 guests

cron