A Joss Research Institute Informal Report:

Construction of a New Roughing Manifold

(09 January, 2010, ff)

I am debugging a hollow-cathode laser head that I recently built, and having some trouble getting it to pump down to a suitable level. A good part of this appears to be an outgassing issue; silicone rubber caulk may not be particularly suitable for vacuum use, even at the minimal vacuum levels involved here. I have constructed a new active section for the laser, with epoxy joints; but in the meanwhile I also wanted to improve the performance of our little vacuum system in a general sense, and to change the manifold from a tangle of polyethylene tubes to something more recognizable. The poly tubing, on its own, pumped down to about 60 mTorr, which is not much of a vacuum even with just a forepump. (I have seen this pump reach 2 mTorr on the bench, running straight into a thermocouple gauge.)

Here are two photos, showing the initial version, before I added a connection to the laser:


(The Baratron tube [grayish blue cylinder with narrow red and silver band] is resting on a gray box that is not part of the system, and can be ignored. The dark brown device at the upper right, across from the Baratron and connected to the same small tee fitting, is a thermocouple gauge tube. The transparent hose at lower left goes to the vacuum pump; the red handle in line with it operates the ball valve that serves as the main shutoff for the system.)

The connection for the laser will be added to the pipe cap that is at the right end of the manifold in the second photo. [NOTE, added later: That’s what I originally thought. See below for the way I changed my mind.]

As of the evening of January 9, 2010, this manifold appears to pump down to 75-80 mTorr, but there is still one largish leak that I am aware of. It is on the part that I will have to remove in order to make a place to connect the laser, so I am not putting epoxy over it [yet], and I hope to put a stop to it when I reinstall the part. There are other leaks, but they seem to be fairly small.

Speaking of which:

Leaks, Part 1:

How to Find Them

(10 January, 2010, morning)

I want to talk about techniques for locating leaks in low-vacuum [roughing] systems. As DIYers, we don’t always get to use fancy parts and fittings (observe the components in the photos above!), and as a result we need to have ways of dealing with circumstances that are probably rare or even nonexistent in fully professional setups.

At the grossest level, a leak is audible. The ear, however, is large, and the leak is small, so it helps to get down to scale. I often use the hands-free attachment for my telephone for this, as the hole of the microphone is only about 1 mm across, and it is in a little plastic housing that is very maneuverable. I have a splendid application called “Spectrogram” on my iPhone, which lets me view the output of the microphone. Here is a photo of the screen (regrettably out of focus, my apologies; I should have done it as a screenshot), showing the microphone moving along part of the head until it encounters a very large leak:

This works until the leaks get small enough that they start screaming at ultrasonic frequencies, and an ordinary microphone will no longer pick them up. At that point, I shift to an ultrasonic translator. (These are loosely called ultrasonic sniffers.) The translator downshifts frequencies that are above the range of human hearing so that you can listen to them on headphones or see an indication on a meter, and thus allows you to find more leaks. I got an old HP Delcron translator on eBay after finding it somewhat difficult to build one, but I will admit that I lost patience fairly rapidly because I wanted my laser to work and I was not particularly interested in other projects at that point. The Delcron has a fairly large sensor head, so I have to be fairly savvy in order to get good use of it, but it definitely works.

Past a certain point either the leaks are howling at frequencies the sniffer doesn’t pick up, or the amplitude of the tiny shrieks is too low for the sniffer to be able to differentiate them from noise, and you have to shift to another technique. One easy method is to put the DUT (“Device Under Test”) in a bucket of water, and push a little air into it. Wherever you see bubbles emerging, there is a leak. This method works even for very large leaks (especially if you don’t mind a little water getting into your device), and I have used it to good effect with several of my nitrogen laser heads, especially before I acquired the sniffer — for some reason, nitrogen lasers seem to ignore modest amounts of water vapor.

The bubble test works quite well for some objects, but not for others, and I suspect that it isn’t very good with truly small leaks. The method I am currently using involves dribbling a volatile liquid (I use 99.85% pure isopropyl alcohol) on places where there could be a leak. When the liquid first reaches the leak it impedes the flow of air, and the pressure in the system goes down a little. As soon as the liquid penetrates the leak, however, and actually gets into the vacuum, it starts to evaporate and the pressure goes back up, often to a level higher than it originally was. This method can probably even take you through the entire “sniffer” range if you are slick about interpreting your results, though it is easier if you have gotten your system down to the point where you can read the pressure with an analog electronic gauge, for example one that uses a thermocouple tube — your eye is sensitive to rather small motions, and the analog gauge is better for this purpose than a digital one. Here is a typical TC gauge, reading 76 or 77 milliTorr:

(10 January, 2010, evening)

Here, if you want to watch it, or here, if you want a significantly smaller download (at the expense of lousy image quality) is a short and somewhat experimental video about this method. It provides a sense of the pace at which the pressure tends to change during the process, though I will admit that this was my second take, and there was some isopropyl alcohol left in the threads from the first take, so the system doesn’t rough down as rapidly as it should. (Fortunately, though, that provides a nice contrast to the speed at which the pressure falls when I put the iso on the joint.)

Meanwhile: because the liquid moves around when you apply it, you have to think carefully about where the leak is actually located. It is also difficult to get the liquid into underhanging locations unless you can pick up the DUT and change its orientation, or squirt the stuff upward. Still, with any luck this method should probably be good enough to take you down to a few mTorr, at which point you can begin to think about the joys and tribulations of actual high vacuum.

...But that’s a different story, for a different page, and we still have territory to cover.

Leaks, Part 2:

How to Plug Them

Clearly, finding a leak is only half of the story; it’s still there until you do something about it. Some leaks are cured by mechanical means: sometimes, for example, you can just turn a fitting until it is tight enough that it stops leaking. If you don’t need anything below a Torr or two, you can happily use silicone rubber caulk, even on some surprisingly large gaps. (I like the aquarium grade.) There is a special grade of epoxy that is intended for vacuum systems, and is good down to 10-9 Torr (“Torr-Seal”, which is carried by Thorlabs), but it is very expensive. I have seen claims on the Web to the effect that Loctite’s Hysol 1C is quite similar, and it is certainly a lot cheaper, but I don’t know how accurate the claims are. As you can see from the gray color in the photos above, I have been using J-B Weld epoxy on this particular system; J-B should certainly be viable down to below 1 mTorr. Teflon tape, carefully applied to clean threads that do not have burrs on them, should likewise be good down to considerably less than 1 mTorr, though it may be advisable to use tape that is thicker than the usual flimsy white kind; the yellow containers that are specified for gas piping may work well, or possibly the pink type that is made for water pipes. (I am not sure of either of these, and we do not yet have a system that goes below 1 micron, so it may be a while before I can confirm this. OTOH, there is probably confirmation out on the Web.)

I will note that the threads on galvanized pipe are filthy (I scrubbed all of the pieces of this manifold rather thoroughly before I attempted to assemble it), and that they are essentially guaranteed to have nicks and burrs. I did not clean these up, and I suspect that this is part of the reason why almost every joint in the manifold was leaky when I first threw it together. (If I build another of these, I will take care to examine the threads and to clean them up with a pattern file.) The interior threads in the tees and some of the other fittings are not at all easy to reshape, of course, but fortunately they are less likely to be damaged.

Back on Track:

(10 January, 2010, late evening)

Although I am not building a new manifold, at least not yet, I am adding a port to this one, so I can connect it to the laser and, later, to other things I will want to evacuate. Rather than drill and tape a hole in the pipe cap, though, I decided to assemble a new set of fittings for that arm of the tee. I did not actually take a file to the threads on the new fittings, for the obvious reason that I put them together with epoxy rather than teflon tape; I did, however, run the threads together a few times, about as tightly as I could get them by hand. (This cleans them up a little bit.) I intend to use pink tef-tape to put this assembly on the tee, as you can see from this photo:

(11 January, 2010, afternoon)

It went right together; I held the new fitting stack in a vise, and rotated the big tee about as far as I could without straining myself unduly. Here’s what it looks like:

It also worked, right off the bat, and I am pleased to report that the thicker teflon tape is viable — within about a minute after I turned on the pump, it was down to just over 30 microns:

It continued to improve after I took the photo, and I have left it for a while to let the pump clear more junk from the interior; we’ll see how far down it gets. This is already much better than my original setup was, and I may be able to improve it slightly: I suspect that there is at least one more minor leak, and there are several places where I used the thin tef-tape, as you can see from the photo.

(early that evening)

It got down to between 20 and 25, but I tried looking for leaks with isopropyl alcohol, and now it seems to be hovering between 25 and 30. The response was slow enough that I am not entirely sure where the leaks are, but I did tighten one joint a bit. I am thinking about redoing the joints with the white teflon tape, but it may not be worth the effort.

I also tried opening the valve to the laser, and was very pleased to note that I can now get the head down to just under 550 mTorr. If I shorten the hose, I may get it below half a Torr. That, however, is something I will discuss on the laser page. (See link, above.)

(Note, added 18 January, 2011)

It has finally occurred to me that the roughing pump on this system is not the one that I’ve seen pulling down to 2 microns on the bench; that pump is one we acquired more recently. I will, as time and tide permit, see what happens when I run this pump directly into the TC tube, and if it isn’t good enough I will switch to the other one.

This work is supported by
the Joss Research Institute
19 Main St.
Laurel  MD  20707-4303   USA

Contact Information:

Email: a@b.com, where you can replace b with joss (as in Joss Research Institute), and a with my first name (no “h”, only 3 letters).

Phone: +1 240 604 4495.

Last modified: Tue Jan 18 15:46:13 EST 2011