Archived - Test of new gas control system

[Doug Coulter]

Which works really well, so worth posting about.

http://www.youtube.com/watch?v=ktNHIqzkjy4

I added a solenoid valve between the turbo and the forepump, turbo going 35% of nominal speed, forepump always on, but only valved in for about 1/2 second every few seconds. As you can see in the movie, this is more than enough to be automateable at this point to keep things on the sweet spot, something that's been pretty elusive so far here -- and you can control the effective speed of the solenoid valve by varying the turbo speed -- I'm looking for the best spot now, this is good for manual control off a pushbutton but obviously I want to make this automatic all the way and might want a different set of time constants for that.

The sound in the movie is the 3He tube going with a theshold set that it takes at least two neutrons at a time (within a couple uS of one another) to click at all to keep the count rate down to "human can hear what's going on" rates.

During this run, I got about a mrem of neuts/minute (on two BTI's), and in two minutes got silver to 278 cpm with a 30 sec delay to the activation counter. This was with the main supply at 50kv and 10ma limit (usually running less than half that) and the supply on the little grid dragged down in current limit (12.5ma) to around 20kv, most of the time.

This is a fairly strange test config otherwise. I have two grids in the tank, a tiny one in the foreground
(which just started out as a way to test new feedthrough designs) and the main one I normally use seen end on, in the back -- the other light source is the uWave ion source. Strangely, this actually works better than either alone, as the little one keeps a nice supply of ions going all the time, even when its tiny HV supply is dragged down below the volts at which it makes neutrons itself. It does cause the central "ray" coming from the main grid to bend toward it, dramatically, which can be altered by how much power is put into the little one. When I run the little one alone, I do get some neutrons when the volts on it are high, but not very many at all.

FWIW, the small grid draws more current and lights off at lower pressure than the large one does, presumably because it's out in the big tank proper (14" ID by a couple feet tall). This also makes the large, main, grid do the same since this ion source is pretty loud compared to the uWave ion source, even though the ion quality (monatomic? Nah.) is no doubt lower with it. The uWave ion source is still needed to get all this lit off at the lower pressures that are giving me higher Q (even in static equilibrium mode, which is shown here).

Pressures obviously varied -- you get a good visual from the pattern of electrons striking the viewport,
(and they are for sure electrons, but that's another post) but in the range of 3-5 e-3 mbar for this run. Normal best for me was 1.9-2.2 e -2 mbar in other lashups -- but this gives far higher Q than that did, roughly 5x, and output scales so far with power input. These pressures read out on a Pfeiffer pkr-251 known to read really high on D, particularly if most of it is ionized already which is the case in some of these pix.

Normally, that's not the case unless the fusor is in what Tyler called a runaway state (I call it divergent or the same thing he does). But with this fine pressure control and setup, it can't run away to the point where I run out of power (or things melt) and I get a lot of ions instead of mostly neutrals. The guess here is that with a lot less gas it can only runaway so much.

First pic is both grids running and cranking out the neutrons at a nifty rate. The other pix are of just the small grid (1.5" by .5") running alone in the middle of the big tank, (or at least out in it, instead of in a sidearm). Grids are on separate power supplies (Spellmans rock! -- more on that later) and ballasts. No pulse mode this run, I was working on gas control hardware, for the moment. I'm sweeping the turbo speed parameter to look for the right speed to make the solenoid forepump connection do the right thing with a dumb opamp/oneshot controller just now, more later.

Nice eye candy, anyhow.

[Steven Sesselmann]

Doug,

Nice work, love the sound of that He3 counter in the background, even if the visual scene is far from becoming a blockbuster...

I would live to add one of your run's to my Q list, but you have not given the radius to the BTI detector, so 1 millirem to where?

Would be nice to know how efficient your rector is running.

Steven

[Doug Coulter]

I don't have good absolute numbers just yet, and in this case, I'm not running my super high Q pulse mode at all, though I am seeing better output (I think) than most other lashups.

I have noticed that the neutrons do NOT seem to be coming from a line or point source here at all, and if fact seem to come out in beams -- as this contradicts nearly all theory, I don't talk about it much till I measure that better.

In this case I was using a BTI standing on end beside the tank end of the grid. EG the cylinder is horizontal, and the BTI is vertical and 4" to the side right at the very end (near the ring). If I move the thing about 3/4" back towards the middle of the grid length -- I see about half the neutrons there, and that is very repeatable. I *think* I even see a non uniform distribution of bubbles in the BTI's too, as if there were fairly narrow beams happening, but to see that I have to run almost to destruction of the BTI (100's of bubbles). In other words, what I *think* I am seeing is non uniform neutron output along the grid length, and also normal to the grid axis as well, as if I had some beams coming out. If that's true, then I have no real good way of measuring total neutron output by measuring in a small area and then extrapolating. In fact, last night during a run I had two silvers in the neutron oven, and the one closer to the grid end (both same distance from grid axis) got significantly hotter than the other one -- even after some moderation which doesn't make a heck of a lot of sense, but it's what I measured.

I am NOT seeing this with detectors that subtend a large angle, which makes sense as that would "smear out" any location sensitivity. For example, the 3He tube in the big (6" by 2 feet) moderator block. So the next test series will be an attempt to look at it with the neutron camera I built -- just like what you drew in the discussion we had with Hendron on that idea, but with a more reasonable sized center hole (1/2") for fast neutrons so I get decent counts. There may have to be design changes to make it really work, this will be first tests. But first I had to get the fusor running right again even in normal mode to have a good N source to test it on!

This plumbing (I'll put up some pix of it tonight) had to go in first so the thing will run stable while I walk away from the controls to do that....like a lot of things, it's all about getting really ready first.

Not only does this gas control method work well, it's cheap and easy to do....I like that part a lot.
Response is far faster than with a needle valve and orifice setup, so it's easier to stay in a small range of gas pressure, and even regulate it off the power supply current so it's all autopilot once set up.

[Doug Coulter]

As promised, here's some more pix of the setup.

The first one shows the new foreline plumbing. There is already a large valve in that line, but it didn't have fine enough control to balance things, and was of course in a not-handy spot. When running the fusor, this large valve is closed, and the little solenoid valve with the smaller bellows tubing on it is used instead, just pulsing it open with a pushbutton. I am still finding the ideal turbo standby speed to make this as sweet as possible, but am getting close at 35% of nominal speed. The minimum, 20% is too slow, and the solenoid valve has to spend too much time open, full speed is way too fast and you can't pulse the solenoid for a short enough time not to way overshoot the desired pressure, which takes a long recovery time if the gas flow is slow, as I prefer. You can see the big turbo and part of the mass spectrometer in this picture as well. The turbo is a 512 L/S nominal turbo-drag pump, and the forepump is an XtraDry teflon ring two stage piston pump. As you can see, for initial tests I just lashed up the solenoid valve to a couple of 12v gel cell batteries and a pushbutton. Questions to be answered yet are the final best turbo speed to get a good time constant for all this, and whether that long bellows tubing should be on the turbo side to give it a little more buffer volume (as it is now), or on the foreline side to help keep the turbo from overshooting when the valve is opened for a short time....luckily with the QF stuff I won't go broke on gaskets trying all that.

Right now, with the numbers above and this lashup, opening the solenoid valve for half a second takes it from the top of the desired gas pressure range to the bottom -- but there's a lag of about two seconds after hitting the button before the lower limit is reached. Then as the gas bleeds in, maybe 10 seconds before needing to do that again. Getting into a nice range for some sort of automated control, in other words, though I'd like to have less lag time.

This is much better than cycling power to the forepump for two reasons. One is that it sometimes backtreams a (pump) cylinder full of gas on the first stroke, and the other is that it has a "graceful" slow start so it's hard to get it to just revolve once -- about what is needed most times. It's so quiet it's hard to hear it start up to do that. The solenoid valve is "right now" -- much easier to control with say a one shot and a FET driver, perhaps hooked to main supply current sense output which is far more sensitive than the pressure gage I have here -- the whole good range is inside its least significant digit.
During that range, the main power current can go from .1 ma to 20 or so...quite a lot of change for the pressure gage to be sitting there reading the same number the entire time, or maybe bobbling the low digit a little.

I have found that once you've run this awhile, you can cut the gas flow way way back and get almost enough for a long run with just the outgassing back into the tank pumping out a little once in awhile as the pressure returns. But the needle valve/orifice arrangement is really slow, always completely overshoots, and is a bit unpredictable, so once set, I tend to leave it alone, and work with the even slower pressure adjustment on the regulator in front of all that, usually running between 2 and 10 psiG, so any D leaks are always "out" and my supply gas stays extra pure. I have now upgraded that plumbing so the little bit that needs to be insulated (for the ion source pusher electrode) is pyrex glued to the copper capillary pipe running from the valve/tank to the ion source.

The other picture is of the back side of the gear (the side I DON'T stand on when power is on, the lead coverage and HV safety isn't complete there). The main grid is attached to the homebrew feedthrough on the right, via a 225w forced air cooled ballast resistor, cooled by the air from an APAP pump (sold for sleep apnea -- nice medium pressure very quiet air source). This cooling isn't usually needed, but the big Spellman hanging from the left side of the rack here can definitely get it very hot in the event of arcing or feedthrough failure. The big Spellman is rated 50kv @ 40 ma, but that's *very* conservative, and in arc recovery puts out perhaps twice that, and so can eat ballasts easily if you don't pay attention to that detail. Even a 225w resistor can't take 4kw for very long...

On the left is the other HV feedthrough and ballast, currently supporting the smaller grid that is in the foreground on the pictures of the original post. This ballast is a bunch of 10k 2W ceramic bulk resistors as suggested by Cliff S from Spellman, and it works very well -- I mounted them up in a pyrex tube as a container, and all these screw together with 10-32 screws (which hole size also conveniently holds a banana plug). The small grid runs off one of the smaller Spellmans sitting on the table, at up to 40kv and 12.5 ma. The smaller Spellman seems to have better "manners" when it comes to frying ballasts, so I may not even need to pot that to get the heat out -- time will tell. So far, the slip on heatshrink tubing hasn't shrunk, a good sign.

I am putting the BTI detector on a piece of plexiglass standing on end, behind the red plastic cup (which is full of lead to hold the plexiglass down) you can see to the left of the sidearm the main grid lives in, right at about the tank/sidearm junction where there are roughly 3 times as many neutrons as there are when it is placed in the middle of the grid length (this experiment has been repeated many times -- something is really going on there but I'm not prepared to interpret what just yet). The big white pipe on end on the right is our 3He detector, moderator, and preamp.

You'll notice the once-pretty tank is now all covered with lead (1/8" thick and overlapped generously in most places).....that's needed for safety of others in the room. I stand behind a quite thick piece of PbO glass that stops just about everything when running, and am set up so all the controls are easy to reach without putting my arm into a hot X ray beam. I prefer the mark I eyeballs with stereo vision and good color rendition to a webcam any day -- but only if it's safe, so I did the extra work for that.

BTW, I just had to replace my first piece of sacrificial 3/8" thick Pyrex I use inside my main window. For one thing it was getting very brown, and since we're trying to look at and take pictures of dim purple/blue stuff, not good. But that's not the real reason. It was starting to craze-crack in kind of a target bullseye pattern, and I feared it might shatter. It gets hit hard with X rays and electrons in some operating conditions (you can see the electron stray beams moving in the video linked above) and in fact gets quite hot sometimes, so I now have a nice desk ornament. I am going to try baking it in my heat treat oven next time I run it and see if I can anneal it back to usable, as some literature suggests. I can't complain though -- it's withstood a few hundred running hours, cheap at the price compared to replacing the glass in the $1k door/window at the manufacturers price for that.

Finally -- almost to the point I can start this and walk away, come back later, and find it still on the sweet spot -- we've managed that a couple of times by very careful valve balancing, but it's the exception -- this will now be the rule.

Now to go play with it all some more!

[Starfire]

Doug - a PID loop should give you the control you seek - nice work - thanks for sharing.

[Doug Coulter]

Yes, a PID loop is just what I had in mind, with minor mods to handle the "bang-bang" valve (perhaps having it drive pwm valve driver, but that has a rate limit how fine an on time I can have). I am simply trying to make it easier for the loop to do a good job first.

[David D Speck MD]

Doug,

Does there exist any sort of neutron sensitive areal detector, sort of an X-ray film for neutrons? If so, you could place some in the proximity of your fusor, expose it for a period, and see if you do indeed have a non-isotropic production of neutrons.

Dave Speck

[Doug Coulter]

Kind of, but that's the rub, such detectors don't really exist at hobbyist price/effort levels unless my one-pixel neutron camera pans out. When it works nice, I'll report complete with how to make your own.

The BTI can do this -- the bubbles only happen where they are hit, but that's over a fairly small area, and not very good resolution at all -- you can see when one is compressed that the only places bubbles will appear no matter what is kind of sparse -- close examination will show where bubbles could appear due to the superheated fluid droplets that aren't that dense in a pixel/inch sense. And I do see hints on the BTI's even with that limitation, but....takes a long time and a lot of different placements and orientations to build a decent picture (which of course means buying a lot more BTI's as they have a limited number of times they work. along with the well known shelf life limit). To really do it with BTI's you first have to have a 100% stable/repeatable fusor -- which is one reason I started this thread -- I am finally getting there on that one. Without a known stable/repeatable source, the rest can only be BS and guessing. Also we have to remember that the rest of the fluid in the BTI can scatter neutrons....and so would smear out the resolution via that mechanism.

There are really (at least) two issues with this -- one is where are the neutrons originating, and obviously, different places would produce different results just due to square law even if the things are pure-isotropic (as theory/lit seem to say). The other is -- is there actual beaming, which is theoretical heresy AFAIK, but seems to be happening nevertheless when tested at longer ranges where the square law effects would be less. -- but the only tests of that I've been able to make so far aren't what I'd call "Good Solid Hard Science" -- yet. No proof from here so far -- I'm working that issue about as hard as I can, though. Because I see too many things that the lit and reports here simply don't explain, and I see them over and over -- nicely repeatable -- I smell that something important is being missed, which for a science guy is kind of interesting to say the very least. Opportunity is knocking here. I figure anything that happens I can't explain is a chance to learn something important, and even if it seems irrelevant at the time, has a tendency to be real useful later on.

I guess I'm hoping to stimulate others to try as well -- the more the merrier -- you might find something I miss. And when I do GSHS tests on other things, I find a lot of discrepancies between what the SIMION non-experimentalists "see", but no one else seems to be doing them on real setups (this on other things about the plasma parameters, but it seems there are some fairly big holes and low hanging fruit to be investigated here).

More when I know more, until then, it's noisy speculations. Hope someone else gets onto this, I don't need all the glory -- or all the work, either.

Till then, what would you think if nearly all the bubbles in a BTI were in a narrow band as though all the neutrons that hit it were in a plane beam cutting across the thing?

In the science canon -- "I did this and I saw that" -- I don't have a defend-able hypothesis, yet. But I do this, see that, and it happens real repeatably, here. More so the more stable I get the basic fusor, which is the main effort right now here -- a little way to go yet, but it's only getting better these days.