Gas control

[Doug Coulter]

Hopefully, I won't soon be moving this post to "it almost worked", but I figured I'd go ahead and write this up while the sealants harden.

I found some small solenoid air valves at Surplus Center, one of which was tiny enough to be a good candidate for letting deuterium into my fusor.
It's their number 20-1452. Cheap, $6.95. I actually got samples of most of the valves on that page, but this one looked right for the application, with some tweaking. We want a tiny valve, a small "wasted" volume, and something easy to drive. This fits that bill, but has one problem, the fact that the output port is vented to atmosphere when the valve is shut. That turns out not to be real hard to fix, as the vent is through the solenoid pole piece center. I simply drilled it out and tapped it 6-32 (it has a hole already) so I can stick a screw n there and seal it up. Because of the way this is made, for high vacuum work, we switch the direction, eg the port marked "in" is the one that goes to your vacuum tank, and the "out" port gets the supply gas. That's because there's less wasted volume on the "in" side, and more chances of leak on the "out" side, and if I'm going to have a leak, well, let it be an "out" rather than "into the system" leak.

The valve has 10-32 tapped ports, with hose barbs on them. I step drilled one of the barbs .125" and then .128" for a loose fit on the Cu tubing I'm going to solder into it. This doesn't leave a lot of metal, but on a lathe it's easy to drill this precisely enough. For the tank end, I through drilled a 10-32 3/4" long socket head SS screw with a 3/32 bit (broke a 1/16" bit when it seized in there). The head end I fit into a hole step drilled in a 2.75" CF flange (drilled the rest of the way through 1/8") and tig welded the head into the pocket in the flange. If it turns out to be needed, I can later add a little bit of that capillary tubing on the inside side of the flange, to slow down gas flow if I can't get a short enough pulse. I'm not starting out that way, as the valve looks like it's gong to be pretty fast. The coil works directly on a diaphram that has a tiny viton button in the center that seals over a tiny port, so a short pulse should not be too much gas for my large tank.

Valve Kludge

The plan is to put together a little circuit with 556 timers and fet drivers to drive this valve, and the somewhat larger one I use to bypass a valve between the turbo output and the forepump input, and have (for now) a couple buttons on it to control the gas pressure. Assuming things work as hoped, this might wind up being controlled by a microprocessor to hold conditions close to ideal with less effort and attention on my part. The idea behind the timers is to get a uniformly short valve open pulse off a pushbutton. If I wind up using a uP, then it can do that part, but for now, manual should do.

I will seal up my tig weld (might have a pinhole, take no chances) and the vent plug screw and coil interface with Hysol 1-C, since I have no plans of ever removing them. I will use well mixed (which can be a challenge) 5 min epoxy for the other "pipe dope" as I may want to remove the fittings, and that stuff isn't all that strong.

For the other side, I'm using a hard to find Burkert valve, but this valve from Surplus center (20-1463) looks ideal for that if anyone else is doing this, and the price is sure right -- another cheapo.
I wound up buying samples of most of the valves on that page, and these two are the winners for their respective tasks. The bigger valve for foreline use is rated 24v AC, which is kind of a pain, but I've tested it with DC and it works fine at a much lower voltage. It wants about 1 amp, so with it's 6 ohm coil, 5v will do the job fine, or 24v DC with a big series R. Since that would waste a ton of power, I'd go with a 5v supply, myself, or a capacitor coupled pulse if I wanted it super quick. So far, I've not had an issue with speed at that place, as I spin down the turbo anyway to make that less touchy with the valve I'm already using there.

So, snatch up these valves while you can. They have a bunch in stock today, but you should know how it goes with surplus -- gone tomorrow, and these are hard to find new (and to pay for at new prices).

[chrismb]

What's the differential pressure across the valve you are hoping to be running, Doug?

[Doug Coulter]

I am right now fooling with that, just took a break to check other things (for once, the few things I'm long in the markets in are going up, maybe time to get out). Doesn't look like I'll be moving this to the entertaining failure thread just yet, though.

No leaks. Yay! First things first. My mass spec is showing "not fully cured epoxy" at the moment, which was expected. Hysol 1-C takes for doggone ever to finish curing and it hasn't been 24 hours yet.

I am trying various diff pressures now. The valve, even with a tiny orifice (about 1/32"), is too big. Running say 1/3 STP in, and vacuum on the other side, it dumps about 1cc of gas in there when I flick the cliplead over the battery terminal -- probably 50-100 ms energize time. So, I'm upstairs here in the fiddle lab building a 555 timer/fet driver to put in shorter pulses reliably. As of now, it's going from hard vacuum to about 1 millibar in the tank with a short (manual, uncontrolled) pulse with STP on the D feed side. About 100x more than I want, but we'll see how fine I can make the valve controller work. I appear to have zero leakage on the input side too, and I do have a regulator that will go sub-atmospheric, but if I don't have to, I'd rather not run that way, as then any leaks would be "in" and contaminate my gas. Once I purge this (I changed D tanks when I did this -- no reason other than the one on there was getting low) all the way I can test for that better.

It does appear that this would work better with an inch or so of capillary tubing on the output side, which will be easy to add if I can't get short enough pulses reliably. The valve should be pretty fast (inductance limited) because the thing being actuated is a diaphragm with very little mass and a strong spring, but the coil has a lot of turns -- it's about 800 ohms for 24vdc. Nicely tiny current draw, the fet I'll probably use won't notice its loaded.

The idea behind the capillary tubing in this setup (if I add it) and the previous one, is that it slows down the flow a whole lot -- seconds per CC or more. When the valve is pulsed, the volume behind the cap tubing is filled to full supply pressure, then slowly enters the tank. That means you more or less just get that volume per pulse -- whatever "lost volume" was between the valve outlet and the cap tubing. In the case of this valve, used this way (why I' used it backwards), that's a tiny volume, a small fraction of a CC (and about the right amount of gas), so that would make it work even with sloppy input pulses. Of course, I hope to do better than that and make this adjustable via pulse width. Without the cap tubing, gas flows out quick during the whole pulse, so more can come through the valve.

Current mass spec, 2 hours baking after opening the tank and changing things 7e-7 mbar total

Note the water from having the tank at STP for half an hour in damp air, and those lines up at the heavy masses -- epoxy. Tank is at about 120 deg F, baking now. Total pressure is about 7e-7 mbar at this point, and dropping.

[Doug Coulter]

Ok, this is definitely a winner. Here's what the pieces look like installed.

Valve installed

Sorry bout the lousy photography -- night time here and flash stinks. In this pic you can see the 6-32 screw I put in the top of the valve pole piece to block off the "vent", and they Hysol I put on that to make sure it stays that way, along with the flange mount, sealant, and input line.

Control box

Here's the little box I built that holds a 556 dual timer, a 7805 regulator and a couple of IRL-3705 logic level fets (that latter is overkill, it's what I had laying around). I didn't draw a schematic, just wired it up. I used .1uf timing caps, and a 1 meg pot with 10k in series to set times. This gives me down to a few milliseconds for the little valve, and up to a couple hundred for the big one in the foreline. Sadly, the pushbuttons have some bounce, and sometimes I get two pulses. Later, I'll fix that (for now, I'm skilled enough to hit the switch so it doesn't bounce), but probably after some more tuning just make a uP control for it that can look at things like system pressure or main supply voltage/current and regulate things for me. To trigger this, I have the button pull down a 10k resistor tied to +5v, coupled through a .002uf cap (to shorten the pulse) into the 556 trigger pin, which is also pulled up with a 10k. Of course, I used the usual 1n4007 diodes across the outputs to catch the inductive flyback, and for the big valve, an RC snubber at the valve. The whole thing runs off a 24v 600ma wallwart, which is a little wimpy to hold the big valve open for long, but...I don't do that.

As things sit, I'm running both valves right on the edge of the shortest pulse they will reliably "go" with -- and it works fine, but could use slightly finer control. I can spin down the turbo further for that side of things, but would have to add an orifice or capillary tube to the inlet to make that a bit finer.

As it sits, it's usable now, and a lot faster to tune than the way it was, with a plain manual switch on the foreline, and a semi broken needle valve used in on-off mode with a longer line and a capillary orifice.

Hitting the fill button will take me from say 1.4e-2 to 1.6 or 1.7 e-2 (which is where I run mostly), and hitting the exhaust button will take me from 2.2e-2 to 1.9 or thereabouts.
This is with 2 psig (above atmosphere) on the D inlet, which is about where I'd like to be so all leaks are "out" and I get pure gas. The two stage regulator goes all the way to vacuum (not sure how fine though) so I can play with that if it seems a good idea later.

Victory! The system does not leak, this is stable and rock solid just like it is. I probably should have used trimpots so I won't bump the adjustments, but hey, that's what they make glue for, right?

Here's a mass spectrum I took just before this test, after the tank had cooled down to room temperature from this afternoon's baking.

Mass spec showing noise floor

Yeah....cool.

So, these speeds/feeds work OK on the big tank. I'd guess that for those with smaller tanks, running lower input feed pressure and/or using an output orifice would be good. I'm not sure exactly how many liters my tank is, but my gut says it's about 10 gallons or thereabouts. YMMV, as we say. Without an orifice, the response is basically instant, but the chunk size is a little too big, and I think the orifice would be a good trade if we still got there in say, a second but with less gas per pulse. FWIW, the pulse width here is about 5ms for the little guy, it's pretty fast.

[Doug Coulter]

Update -- I've ordered a number of these valves so we will have them when people need them (10 sets). I've worked out a way to cut the flow per pulse dramatically, so they should work even on smaller tanks and systems. Taking the output to the tank out of the bottom of the valves allows a side screw to fill up a lot of wasted volume in there, and if the screw out the bottom is drilled for a capillary tubing that projects pas the end slightly, we'll be in the hundreth of a cc range per pulse, or below -- virtually no wasted volume. The other valve, larger, looks ideal for a foreline valve. Being big, it's not super fast, but is pretty fast. As I mentioned above it was meant for 24v AC, but it works just dandy on DC and doesn't waste much power doing that. I did a test run on this with it near the drop out current (about .7 amp) and it held open for a few hours without getting hot. This is a bonus -- now this doesn't have to be in a sidearm around another manual valve, it can be the only valve, no extra plumbing.

Maybe I can get one of the machinists here to make some braze-in adapters between the 3/8" OD copper tubing and a QF-16 (or other size) QF flange. I've made them here, but on my primitive machines, it's a pain....and Lesker doesn't give them away, quite.

Picture of foreline valve

p20-1463C.jpg (10.04 KiB) Viewed 20396 times

Lesker link for half nipple, brass:
http://www.lesker.com/newweb/flanges/fi ... pgid=half3
stainless:
http://www.lesker.com/newweb/flanges/fi ... pgid=half1
blank flanges:
http://www.lesker.com/newweb/flanges/fl ... id=flange1

None of the nipples match 3/8", but a flange could be drilled. Going to be "fun" to clean off the oxide after brazing so it would seal, though.

Any takers? I've made blanks from the drawings (which is the real reason I put the lesker links up here), but with my tools getting the taper and edge thickness just so (at the same time) is a pain. I'll do it if I must...

[Doug Coulter]

Got motivated to get into the shop and do a little work. Here's a picture of the valve mod parts needed to make this into a super-fine gas inlet control.

Fabbed parts for valve mod

The ss screw has been machined to have a nub on the end that is .092" diameter and .087" long. This fills some excess volume where they step drilled for the original 10-32 thread, in what was the "input" side of the valve. I'm using it as the output side because this is the side you can get to really low wasted volume on -- the other side is hooked to the whole diaphragm cavity.

The brass screw is through-drilled with a #50 drill to allow a little room for solder between the hole and the capillary tubing, and step drilled a little at one end for a solder cavity to solder it in. You have to use solder (or epoxy) here because heating the cap tube to brazing temperatures will cause the hole to close up with oxides. I use some special flux for SS from McMaster, and where there was no flux, no solder will stick, so it's actually a little easier to do it with solder without plugging up the tube.

The cap tube, with electropolished ends (the only way to get the holes open again after cutting one), extends slightly out from the brass screw in use -- filling up a little remaining volume inside the valve. It will almost contact the end of the orifice insert you can see in the middle of the valve body. For this rev, I'm trying a .007" diameter capillary tube, about 1.4" long. By itself, this is just a little too little restriction when feeding gas into my main tank with an "on-off" valve, manually. With the trick solenoid, it ought to take a few seconds to get a tank from hard vacuum to fusor range - and we can pulse this baby with 5ms pulses...so we'll have some resolution...should be nice.

You'd want to be careful about dirt here -- these cap tubes can clog. I blew this one out right after etching with shop air so it wouldn't clog with dried up electrolyte. 100 psi on one side give a little stream of bubbles out the other end in water -- reminiscent of what you see in a beer. But as you can see in the picture, my shop isn't that clean. I always make a clean spot and work fast on this kind of thing, after getting really ready.

Both screws turned to match the drill angle at the bottom of the tapped holes -- roughly 117 degrees. They'll be put in with a little epoxy sealant to fill any gaps and make sure wasted volume is minimal.

The coil has a vented center pole that would leak your supply gas out. This I will drill/tap for 6/32 and put in a screw with epoxy seal. I didn't want to try and plug up the rather delicate looking center pole piece at the business end, as it looks a little dicey and I don't want to mess up how the magnetics work -- excellent as is, and on the gas input side, wasted volume isn't much of an issue anyway.

You can see the little diaphragm in the foreground. It's light, and fast, and leaks just about exactly zero in the one I put on the big tank. I'd bet you could do 50-100 hz on this thing with a uP controller and have a really nice gas controller. OF course, I'll be doing that down the road, and probably using main power supply signals for feedback. A fusor is a more delicate and sensitive gas gage than any of the other ones I have. Heck, with a higher DC supply you could probably go twice as fast (eg half the pulse length to actuate).

The overall idea here is to limit wasted volume on the tank side, before the cap tubing. The cap tubing is essentially a short circuit for gas trying to go fast - like a capacitor to ground. So, the valve operates, fills the wasted volume and closes before any amount of gas makes it down the cap tube. It then trickles into the tank over the next second or so. So it's a low pass filter you can dump a given amount of "charge" into, discretely, with a limit on maximum rate if you just hold the valve open.

With that on the intake, and the big valve on exhaust, and either a couple of buttons or a uP controller, the job is done. Anyone who has tried to do this with manually operated valves will love this - it's a real breakthrough, and not expensive if you do the work to mod this cool little valve. Probably the hardest operation is gun-drilling that brass screw, it took a lot of backing and forthing to clear chips on that one -- and I got tired of breaking drills on SS, so that's why the brass.

I'd bet you could just drill and tap a vacuum system wall and screw this in (add a backing nut and epoxy) and save a flange too. I may do that on my little system.

[johnf]

Doug

sorry been quite awhile
I see @ work that the new GC's (gas Chromatagraphs) use almost impossibly small diameter tubing to space out the fractions-- all of this designed to interface to those very small swagelock fittings ie a very large outside diameter to the 0.18 um inside. These being made of quartz ,silica, etc, don't tend to get the end munted over problem that stainless gets (needing the electropolishing).

Ie some of this tubing then one of your valves, maybe some more tubing and another valve a very fine gas control could be had --I'll enquire @ work where the tubing comes from, I believe its relatively cheap and its an insulator and probably wont support a discharge down it because of the length diameter ratio

PS the new Accel system @ work is moving again ,more pics in eyecandy soon

[Doug Coulter]

It appears to be a time of year when everyone is busy -- hits on the board are down a little (but that could also be partly due to the robots I killed).

I've head of other capillary tubings (silica?) for various uses, but I have what I have, it's a standard thing from McMaster-Carr, cheap. The electropolishing is not terribly hard to do, my total equipment is a 500 ml beaker with a lead cathode, the solution, and my electroplating supply, none of which is terribly critical. It only takes a couple of minutes per end. The real hassle with these (and I'd suspect any small tubing) is that if you get them clogged, there's not much to do about it but replace it. That's another reason to soft-solder it into the system -- easy to change.

For others:
75% phosphoric acid, sold as ice machine cleaner
10% few percent concentrated sulfuric acid (drain cleaner)
the rest glycerin (drugstore grade is fine)

Then add some water, not a lot, and keep the level up by adding more as needed. You'd probably get around the right amount of water by using a little more battery acid (pre diluted as it were). That's reagent grade vs the crummy drain opener (other than the water content), kind of "too good" for this.

Works pretty well, the solution is way not-critical. I suspend the end of the tubing about 3/32" into the liquid from above, using a clip-lead. Put about 5v (also not critical) positive on the piece to be "eaten". For most uses, I just wait till it stops drawing current, which is maybe 5 minutes, then switch ends. I've tried variations on the solution composition, and some with citric acid added worked slightly better. You can get this at some oriental food stores as Limbu Ka Ful, an Indian spice, and it complexes the chromium a little better they say. Any you have left over makes a reasonable lemonade mix.

I'm still on my first 500ml of this stuff. It gets green (chrome and nickel), then brown, but it hasn't stopped working yet. Good for other SS parts, leaves them mirror-shiny, but it does take the current to do that -- I see a couple amps at 5v with just the tip of the tubing in the solution. The main hassle with this tubing is having to make sure you get any solution out of it that crept in before it evaporates and leaves the tubing clogged. Don't know how I survived before I had shop compressed air!

Bill had thought of using two valves, but the wasted volume in the second one limits how little you can dump per pulse. I'll have to try this one (just finished it yesterday) in a tank and see how it works out in my much smaller system -- I think it will be good, but the proof of the pudding...is the tasting of it.

I'd love to see some more accelerator pix, you bet!