If you want to, for example, measure the amount of tritium you made as part of the back check on your fusion reactor, or see the evolution as gas is released from things (whether fuel or contaminant), then batch mode might be the one you want. Here, on the big system, we use batch mode - we never got that balancing act fine enough, and didn't want to go through too much deuterium. Our 512 l/s pump doesn't work all that well throttled down to the point where the gas waste isn't an issue. I prefer to keep it spinning fairly fast (say 50% speed) so it will act as a mass sensitive purifier, and run in a batch mode - turbo pumps compress heavier (in this case, those would all be contaminants) preferentially to the light ones - our fuel. So the fuel stays and the bad stuff goes, a nice outcome.
I've personally found that the fusor itself is about the best pressure gage there is, when fed with a current limited power supply - amounts of pressure change that don't even show up on a PKR 251 easily change the operating point 10 or more kV. So the method here has been to set up the power, and adjust gas pressure via inlet and outlet pulses until we are where we want to be at some current and voltage level, and run there. While there is some initial drift as gas is "cleaned up" or evolved - it settles down fairly quickly. This is probably an advantage that mainly accrues from using only a small sidearm on a fairly large tank - there's some effective buffering going on. The rough dimensions of that tank are 14" inside diameter, and 18" tall.
3.1415927*49*18 gives me 2770,8848 cu inches, ignoring the warts and sidearms, which probably add 10% or a little more. Call it 3000 cu in for this purpose, or 49161.19 cc.
Now, I want enough gas to make, say .02 millibar (really .01 since the gage reads 2x high). Since we're back of the envelope here, I'll ignore that a atmospheric pressure isn't exactly 1 bar (where I live, it's a little less, I'm at 2500ft elevation). So, .01 (desired) x 10-3 is 10-5 of the volume to let in if we start with perfect vacuum and want to reach that pressure. Or, around .5 cc at STP is the amount of gas we want to fill the big tank, just to get a sense of scale. The smaller Tee setup up stairs is a little tougher to calculate as the various warts are more of the total volume, but I'm going to make a rough guess that it's equivalent to around 6" ID and 20" long (adding the sidearm to the length with a guesstimate for the warts). So, 566 cu in or 9275 cc. Which would work out to a little less than .1 cc (of STP gas, 0 psig) for a total fill.
That's not much gas! Trying to control that to around the 3rd decimal place is kind of difficult...And at least in the fusor, getting it right has a pretty large effect on Q.
So far, the solution here has been to let in around 50% more gas than the desired amount, and then pull some out in tiny increments via a solenoid valved in the foreline, until we hit the amount of gas we want. The minimal pulse we can let in with what we have, is around 1/6th of the total needed in the big tank (6 pulses of the inlet valve). that's nowhere near the desired resolution. However on the outflow side, we have far better resolution, even though the solenoid valve there is both larger and slower - because the pressure drop across the valve is so much smaller, and the total pressure is higher enough that an orifice inline can cut that flow rate even further. With 100ms pulses - much longer than will operate that valve - we often need 10's of pulses to make a minor change, resolution is more than adequate and a fancier control system that just opened that valve until near the target pressure would do fine - and that's on the way for both systems here, using an arduino clone most likely since I got $25 worth of them recently (9 units).
Still, one needs an inlet valve between the regulator and the tank, that doesn't leak an any possible direction, and is fast enough, even with a tiny orifice, not to let in far too much gas per pulse, for reasonable supply pressures. The big system has a fancy 2 stage regulator and the supply to the valve is around 1/4 STP - but I'd rather not pay for another like that (several hundred bucks) and so will use one from welding and perhaps 1 psig or so supply.
This is the inlet valve I plan to use - substantially the same as on the big system, with a little more tweaking.
- The pieces, partway through the modifications.
This is a surplus piece that was used originally with single acting small air-hydraulic cylinders. A solenoid pulls a disk with a bit of rubber in the middle off of a tiny orifice, connecting one port to the other, or at least that's how it is now. The air cylinder used to be on the side of the body cavity, and that hole in the middle of the solenoid (on the right, above) vented to the atmosphere to let the cylinder spring collapse again. That hole has been plugged for this use, and the normal operation direction reversed, so the supply gas is on the body cavity side, and the vacuum system connected to the other end of that small orifice. The orifice sits a little proud of the body, so the rubber seal can form around the valve seat a little without interfering with the body of the valve.
In making the mods for a new one, I noticed that the rubber seal formed on the disk had not only become hard in storage, but that a permanent indent, a little off center, had formed. This meant that taking the valve apart and putting it back together would pretty much always make it leak, even if it was fine before (leak hunting in vacuum systems is fun, just keep telling yourself that).
So in this modification, I replaced that with a little piece of gooey LDPE glued on instead. The glue isn't strong bonding, but here the joint is under compression from spring and pressure drip almost all the time, and only needs to hold the weight of that tiny bit of plastic when the solenoid pulls it off the seat, so I think that will be fine.
The other issue is leakage across the O ring seal between the top and bottom pieces. It's already got the fattest O ring I could find at McMaster-Carr that would fit, but it seems a little porous. A 10cc reservoir and gage leaks up to atmospheric from vacuum in a few days from that as it is now. I made a PFTE ring to be a gasket there, and since that spot isn't fixed - gap, may have to go to blue permatex silicone to really get both sides leak free - after all, there will be far more gas in even a capillary supply line and whatever cavity on the output side of the regulator than an entire fill..I'd rather not waste it.
The solenoid center post was originally sealed via another O ring, but never did have to handle any real pressure drop and a bit of leakage didn't matter. That white ring is Hysol epoxy.
The blue stuff is "sensor safe" permatex, one of the only silicones that can work in this sort of application - it's not the acetic acid cure type that creates corrosion, and in tests, has negligable vapor pressure once cured - systems I have it in easily get below 10-8 millibar.
I've been using 30 volts or so to drive this 24 volt solenoid on the big system, and around 20 milliseconds is the minimum to get a reasonably uniform opening time. Higher voltage would reduce the required time and would be no issue as regards heating, the resistance is pretty high at 870 ohms or so. One could probably control the pressure when starting from vacuum by controlling that pulse, but it'd be fairly challenging, as gas continues to flow a bit after stopping the pulse for various reasons.
Here's some closeups of the parts. I got a bag of these from Burdin's surplus center for a couple bucks each. The 1/4" orifice solenoids I use on the foreline side of things are everywhere and cheap.
I think the ones I got were originally for refrigerant from heat pumps.
- The spring diaphragm with the new plastic seal
- The valve chamber, orifice is in the middle, the fittings are 10-32 threads.
- The ugly sides. Various leaks plugged with sealants.
