TJIIRRS: Number 19

(28 June, 2008)

Surprisingly many Do-It-Yourselfers have constructed lasers of various sorts. I would hazard a guess that the vast majority of these are nitrogen lasers, which are straightforward. Pulsed organic dye lasers (most commonly pumped by nitrogen lasers or by flashlamps, occasionally pumped by other lasers), which are also relatively easy, are probably second; pulsed lasers that use crystal or glass rods are probably in third place; and I would hazard a guess that metal-vapor lasers may be fourth. I am not counting diode lasers, because the DIYer does not actually make the laser. There are some folks who would not count lasers that use commercial rods, either, especially those that use commercial rods in commercial heads. I vary on this, but just at the moment I think that position may be a bit too extreme. True, it is not very hard to bring up an SSY-1 head; but bringing up the laser rangefinder from an M-60 “Abrams” tank is not trivial, especially if you don’t have the electronics package, and have to design and build one yourself.

There is a large amount of information about DIY lasers in Sam’s Laser FAQ, a huge compendium.

Buried in a subsection of the page about Pulsed Multiple-Gas Lasers is a mention of the Helium-Iodine laser. The writer (Steve Roberts) had access to only one article, which was not entirely representative, so the information in the FAQ is incomplete and perhaps somewhat misleading. That report mentioned using a molecular sieve as an iodine source; the optimum temperature for the sieve was 120-130° Celsius, so the laser was operated inside an oven. This, it turns out, is a lot harder than it needs to be. Other researchers typically used a small amount of elemental iodine, either in a sidearm tube off the main laser tube or in the helium delivery hose, and the optimum temperature for that is 15-20° C, which is basically room temperature if you live in a rather cool room. Moreover, only 5 output lines are listed, and I don’t offhand recall any mention of the fact that this laser will happily operate CW if you can supply sufficient cooling.

My purpose here is not to complain about shortcomings (after all, Steve was doing a gloss of several lasers, only got to read one article about this particular type, and did a good job of reporting what was in it), but rather to indicate some reasons why I think the He:I₂ laser may be a good DIY choice, particularly as an entry-level gas laser of the “ordinary” sort. (The nitrogen laser is a great entry-level gas laser of the fast-discharge type.)

After becoming intrigued about helium-iodine, I found and read several articles about it. These gave me a sense of what people had already tried, and what worked well for them. The single most informative report that I’ve found so far is:

High-Current Characteristics of the Continuous-Wave Hollow-Cathode He-I₂ Laser
James A. Piper and Colin Webb
IEEE Journal of Quantum Electronics, V. QE-12 N. 1 (January, 1976), pages 21-25.

The authors list 25 lines covering the range from 448.9 nm to 887.7 nm, 17 of which are listed as either Medium or Strong. In addition, they discuss both the theoretical issues of how the iodine is excited in the discharge, and the practical concerns that limit output power and efficiency. (In the hollow-cathode design there was no sign of saturation, even at the highest current levels the researchers were able to achieve.) Finally, they give a good list of references.

There are, of course, some downsides to this laser. For one thing, the efficiency is very low; even lower, in fact, than that of the ordinary HeNe. For another, it is not likely to be possible to seal off the tube, which means that the vacuum system will have to run whenever the laser is in use, and there will have to be some sort of trap to catch the iodine that comes through the laser. (More about this later; it turns out to be fairly easy.)

Even so, I strongly suspect that the advantages will outweigh the disadvantages, and that this will prove to be a good project for those who want to begin to explore gas lasers other than nitrogen.

Although He:I₂ can be operated CW, I’ve decided to stay with pulsed operation for now, because a pulsed laser is considerably easier to construct than a CW laser. (The long-pulse lasers discussed in various articles tended to use currents as high as about 8A, with 300V across the discharge; removing a thousand or more Watts of heat from even a 1-meter-long tube is a nontrivial task.) Moreover, the hollow-cathode designs described in these articles involve multiple anodes with individual ballast resistors, which also dissipate a large amount of power when the laser is running. This begins to get into serious (and potentially lethal) power supplies, and nasty electricity bills. Pulsing is much cheaper and easier.

Despite the fact that hollow-cathode designs clearly work better than positive-column discharges, though, I have decided not to try one for my first attempt. I have two reasons for this. The first is that I have built a few devices that are similar to positive-column discharge lasers, so I am familiar with some aspects of such a machine. The second has to do with the fact that all of the papers (the ones I’ve read, at any rate) reported on CW or long-pulse lasers. The briefest feature I recall seeing mentioned in any of them (Piper and Webb, cited above) was 100 microseconds, and that was a peak in the laser output that corresponded with the leading edge of the energizing pulse. The authors conjecture that this peak occurs before the discharge can be depleted of iodine that has not been ionized. This, and the fact that the pulse appears to follow the leading edge of the electrical pump pulse quite closely, combine to suggest to me that a briefer pulse might be viable in a positive-column discharge. Exactly how much briefer, however, I do not yet know. We shall see whether I can find out.

I am used to flashlamp-pumped dye lasers, for which 100 microseconds is an incredibly long time. With a capacitance of a few hundred nanofarads across the tube, I am expecting pulses no more than a few microseconds long; I would have to build a pulse-forming network if I wanted to stretch the duration even to a few dozen microseconds, much less hundreds, so I guess we’ll see how well (or even whether) short pulses work.

[Note, added some days later: It becomes clear that I will actually be using capacitors of only a few nanofarads, at least at first, which will make the pulses even shorter than I had initially expected.]

A particularly fortunate aspect of this laser for the Do-It-Yourselfer is the fact that unlike argon or xenon, which run in the millitorr range, it does not require especially high vacuum. Optimum helium pressures are around 10 to 14 Torr for almost all of the lines, at least in a hollow-cathode discharge. (I hope to find out whether this is also the case for positive-column discharges of short duration.) An ordinary refrigerator compressor can achieve better than 10 Torr when it is used as a vacuum pump, and I will use one if I must; but I am hoping that an aspirator pump filled with RV antifreeze, which is about 50% propylene glycol and 50% water, will do. The vapor pressure of the 50-50 mixture is below 10 Torr at temperatures below about 15° C. My worry is that the viscosity of this mixture is a lot higher than the viscosity of water, and it increases as the temperature goes down, so I doubt that I’ll be able to chill the working fluid as much as I would like to. I will be testing this when I have a chance.

The next step is to design and fabricate a suitable tube. I want this to be as simple as possible, partly because it is a “PoP Machine” [Proof of Principle], intended only to help me decide whether my initial notions are worth pursuing, and partly because I take a certain delight in building apparently complex devices without using machine tools.

My initial tube is a little less than 1 meter long, and has ~5mm bore; with it, I hope to begin to get a sense of the character of the discharge; helium alone is very easy to deal with; but as you start adding things that are strongly electron-attracting, it becomes harder and harder to get a clean discharge. (This issue was resolved very well for transverse discharges, even at high pressures, during the development of excimer lasers. That may turn out to be another argument in favor of a hollow-cathode design; but such a design will have to wait.)

I think I’ve figured out a reasonable way to construct ends for the tube; here is a [very] rough sketch:

I am very pleased with this design, because it uses one part at each end (the brass tee) to accomplish three crucial things for the laser:

1. The tee is the cathode or anode, depending on which end it’s at.
   
   
2. The tee holds the quick-connect that is the gas inlet at one end, and the vacuum outlet at the other.
   
   
3. The tee is part of the mirror mount (and, for that matter, the tube mount).

Notes:

1. I expect the discharge to attack the mirrors, so this is just a stopgap measure for initial testing. If the project proceeds further, I will eventually change over to Brewster-angle windows. That may make the mounts more complex; we’ll just have to see how it goes.
   
   
2. Turning the mirrors around backwards has one small advantage: they are a lot easier to align. I suspect that the resulting resonator will be unstable, but as far as I am aware, that’s fine.
   
   
3. I have shown one adjustment screw on the mirror mount in the drawing. Each mount actually has two tweakers, and I will eventually provide either another drawing or a photograph with the relevant details.
   
   
4. I have also shown only one electrical connection here; each tee actually has two, as you can see in the second photo, below.

Here are the two endpieces with mirrors mounted on them, and a side view of one so you can see the pivot (the ball on the right side, just between the jaws of the clip). The clips probably aren’t in their final positions; I suspect that with them the way you see them here, the adjustments probably won’t work correctly. They aren’t glued on, though, and it will be easy to shift them around.

Note the different colors of the mirrors. When I look straight into them the one on the left is yellow-orange, and the one on the right is red-orange. I expect to get more output at longer wavelengths from the yellow end (because it reflects the shorter wavelengths more), and more output at shorter wavelengths from the red end, which reflects longer wavelengths better. Assuming that I can get the device to lase, we’ll see whether this works out in actual practice. (I like the idea of a laser with a beam of one color coming out one end, and a beam of a different color coming out the other end. This should tend to create a nice cognitive dissonance hit in anyone who does not fully understand how lasers work.)

W/r/t electrical issues: I have selected a pair of high-voltage pulse-rated capacitors with relatively low ESL and nearly the same values (about 725 nf). If I put these caps in series, they will probably handle about 3 kV. I may add a third, though that will further decrease the capacitance of the assembly.

My rationale for this:

1. Although I seriously doubt that it will take even 3 kV to flash the tube at 10 Torr, it is better to be safe than sorry. [Note, added somewhat later: I was just grossly wrong about this. See below.]
   
   
2. The two caps in series will give me a little over 360 nf.
   
   
3. At 3 kV, the stored energy in a 360-nf cap is ~1.6 J.
   
   
4. If the total inductance of the circuit is about 500 nanohenry, the risetime will be [extremely roughly] 1.35 μsec. Let’s figure that the pulsewidth is going to be twice that, 2.7 μsec.
   
   
5. If we pretend that the electrical pulse has a flat top (this is, after all, a product of BotEC Engineering, Unincorporated, which is definitely a flat-topped organization ;o), the average power is then just about 600 kW. I could just as easily pretend that it’s a half-sine shape, in which case the peak is more like 850 kW, but let’s be at least a litle bit conservative here.
   
   
6. If we guess that peak current occurs at about 2/3 of the initial charging voltage (that is, roughly 1/e down from the initial level), we are pushing our 600 kW at about 2 kV, so the current is on the order of 300 A.
   
   
7. The ID of the tube appears to be 5 mm or so, which means that the area is a little less than 0.2 cm^2, so the current density should be on the order of 1500 A/cm^2.

If that is not sufficient to cause some sort of interesting behavior, I will be very surprised. OTOH, it could easily cause various instabilities in the discharge. We’ll see, when the time comes.

There is some chance that the self-flash voltage will be extremely low or high. If either of those turns out to be the case case I expect to use a spark gap to switch about 5 or 10 kV into the caps. If I have to do that, I will use capacitors that can handle significantly higher voltages.

(29 June)

I am assembling a temporary Pyrex™ tube out of two shorter pieces, and I must wait for the epoxy to set. In the meanwhile I have found suitable electrode/endpieces, but have not yet drilled and tapped holes for attaching conductors to them. I have acquired two gallons of RV antifreeze to use in the aspirator pump.

(early AM, 01 July)

I have the bronze washers epoxied onto the ends of the tube, and I have drilled and tapped 4-40 holes into the brass tees, for electrical connections.

(01 July afternoon/evening)

I have drilled (and, where necessary, tapped) the steel washers that will become the mirror mounts, and have epoxied one of them to each of the tees. Now I am waiting for the epoxy to cure.

(01 July, late evening)

I have put quick-connects on the tees, with teflon tape, and I have used aquarium caulk to attach mirrors to the mounts. (See photos, above.) I can’t find my tube of the clear version, which is the better adhesive, but perhaps that isn’t so bad — what I’m looking for here is more a sealant than a glue.

Once the caulk has set, I think the next step will be to begin debugging the laser for vacuum. I’m particularly concerned about the electrical contacts; each tee now has two threaded holes in it with 4-40 screws in them, and these are guaranteed to leak air. I am not yet sure what I will do about that.

(02 July, early afternoon)

I have satisfied myself that I can’t run this device as three separate parts, so I put RTV on the endpieces and stood the whole thing up to cure. I hope I will be able to try again this evening. Once I have the vacuum at least vaguely under control I will try running a discharge, probably with air at first, then helium or helium + nitrogen (there is a nice blue N₂⁺ laser, and I wouldn’t mind seeing it, though the chances are small), and finally helium + iodine if I can actually get all of the bugs ironed out.

(02 July, evening)

I have been unable to get the structure to hold together well enough with RTV, so I gave up and epoxied the Pyrex tube into the end fittings. I would much rather have avoided doing this, but I didn’t see any obvious way around it. At least it will be easy to make a new one if I have to... Unfortunately, I won’t really be able to test the new setup until the morning, because the epoxy wants to set for at least 8 hours.

(03 July, afternoon)

The tube now holds vacuum reasonably well. I connected a 60-nf 35 kV capacitor across it, and was surprised to discover that even at nearly 20 kV I could not get a discharge to occur. There is a metal collar on the tube, where I joined two pieces of Pyrex together, and I may be able to apply a short pulse of HV there to get things going.

...Well, no: I connected an EG&G TM-11A trigger unit to the ground end of the tube and the metal collar, and that did not cause a discharge. I think my next step will be to put a longer trigger wire on the glass of the tube, as if it were an old-style flashlamp.

(Ahem. The tube is now about 3 inches shorter than it was, and the epoxy on the repair is curing, so there will be something of a delay before I can continue. I did, however, get a trigger wire wrapped around the tube before I broke it, so things are in line for the next test. I should point out that I have more tubing on its way, and I will be able to make a new [and, presumably, improved] version of this design in a few days.)

(very early AM, 04 July. Happy Independence Day)

I talked with Jarrod Kinsey about this, and he asked me how much vacuum I had on the tube. I checked (the epoxy has had about enough time to cure), and found that with the refrigerator compressor running, I was getting about 28". This translates to almost 50 Torr, so I gritted my teeth and dragged out one of our real vacuum pumps.

That went down to an indicated 29" in about 2 seconds, ...but I still can’t get a discharge, even at 20 kV, even with a trigger wire wrapped around the tube. I conclude that something is screwy. If the tube were just conducting all the time, I would certainly be able to tell — there would be some glow when I turn off the room lights. Also, when it started to conduct as I raised the voltage, there would be at least a small flash. Neither of these is happening.

I just don’t get this; a neon sign transformer delivers peak voltage of 22.5 kV or so, and it will light many feet of tubing with helium in it. I suppose there’s some chance that the capacitor I’m currently using is shorted out, so I think I will go check it...

(04 July)

It took some diligent searching, but I finally found a reasonable diagram that shows the relationship between pressure, length, and breakdown voltage. I have extended it a little, and I also need to note that I found at least one reference in which the authors pointed out that for long discharges, the breakdown voltage is higher than a simple pd:V graph indicates. I should also note that the Paschen relationship refers to flat parallel electrodes, which is not what I have in this case. Here’s the diagram:

(Click the small image to get one that is 1024px across.)

Let’s just double the number at 36 Torr*cm to account for the added distance (another product of BotEC Engineering). What this means is that a 100-cm-long tube at 0.36 Torr should break down and start conducting at about 520 Volts. (What’s wrong with this picture? The mere fact that I want 10 Torr? Argh.)

It is clear that the Paschen curve rises more steeply as we proceed further to the right (higher pd product), so if we go for higher pressure, we have to increase more than linearly. Instead of a pd of 36, let’s go for 360, which means that instead of 520 Volts, we have to multiply by 10 and then by a further compensating factor, which I will arbitrarily peg at 4.

My initial guess, based on all this handwaving, is that a tube 100 cm long that is operating at 3.6 Torr and is filled with helium should break down at just over 20 kV. That’s probably a bit conservative, but it should be within the general ballpark. It strongly suggests that I should be using a smaller capacitor, though, so I am going to swap out the 30-nF device.

I’m sure that the vacuum pump I am currently using is capable, at least in principle, of taking the tube down to 3 Torr or less, and I am also reasonably sure that most of the gas in the tube was helium. On the other hand, I have a 30-M resistor on the output of the power supply as a bleeder, so I am certainly not getting a full 20 kV from it. (Even so, the trigger wire should have helped things along quite a bit, so I remain somewhat puzzled.) Still, it begins to look like a positive-column discharge 100 cm long is going to require Actual High Voltage, so I think I will switch to the heftier supply at the back of the bench, which can deliver at least 37.5 kV. I will also switch to some 40-kV “doorknob” capacitors, because the 6-nf cap I have in the circuit now is not going to handle enough voltage. I also don’t want to store too much energy in the capacitor; a few hundred kW peak electrical power during the pulse should certainly be enough, at least for now. (2 nf at 35 kV is 1.225 J, not far off from my original figure of 1.6 J, above.)

(A little later, that same afternoon)

It occurred to me that I probably had my vacuum gauge at the wrong point in the system, so I moved it. This provides a fundamental insight: if you want to measure a parameter, it’s definitely a good idea to make sure that you are actually measuring the parameter you intend to measure, and not some other parameter entirely. I was measuring the pump inlet, not the tube, and it turns out that there is a significant difference. The tube leaks, and I will have to find out where; it assuredly isn’t going to break down at any reasonable voltage if it isn’t at a reasonable pressure!

That, however, leads me to...

(05 July, late evening, to 06 July, early AM)

I thought about what I’m attempting to accomplish here, and the following sterling notion occurred to me:

When you are trying to get a vacuum system to work, the place to start your debug process is the [roughing] pump. Work out from there into the maze of twisty little passages, and take it one step at a time.

Accordingly, I disconnected the laser from the pump line, put a thermocouple gauge on the pump (along with a valve so I could repressurize the system if it all went according to plan), and then turned on the pump and the gauge. The gauge is perhaps somewhat suboptimal for this work, as it has no real readings between 2,000 milliTorr and “ATM”, but at a minimum it can provide a “Go/NoGo” signal, and once I get things to less than 2 Torr, it will give me a fair sense of the actual pressure.

[Note, added later: I have purchased a capacitance manometer gauge tube on eBay that is good for 0-100 Torr, and will give me a better sense of what is going on in this sort of laser and in nitrogen lasers. I have also purchased a gauge and readout that are good for 0-1000 Torr. Once one or both of these arrive, I can reserve the 0-2000 mT gauge for argon lasers and the like.]

Plastic quick-connect fittings are probably not going to cut the mustard for this application. (I found that the pressure mostly wouldn’t come down as low as 2 Torr; if I wiggled things just so, it would get to a Torr or even a little less for a few moments; but if I let go of the parts it went right back up, and sometimes it went back up even while I was trying to hold things in place.)

The rule of thumb stated above is The Right Stuff. This is clearly the way to proceed, because if you can’t even connect your vacuum pump to the rest of your system without introducing leaks that prevent you from reaching the desired pressure, you aren’t going to be able to tell whether your laser tube is leaking unless the leaks are huge. (I hope and trust that I have passed that stage with this tube.)

(We Now Return You to Our Regular Scheduled Programming.)

Having learned these things, I tried replacing most of the plastic fittings (the ones I could easily remove) with brass quick-connect fittings that I acquired on eBay a while back. These seemed to be a bit better, and I think there is at least a chance that they’ll do for informal service, but there still appears to be some sort of leak: I can’t get the pressure down to half a Torr, and I keep thinking that this pump, which is a rather decent one, should be capable of less than 0.1 Torr. I have applied RTV caulk to some obvious places, and swapped out even the brass quick-connects for compression fittings; I will take another look when the caulk has had time to cure. I will also check the tightness of the original compression fittings that were already in place on the vacuum pump and its inlet hose, just in case.

(A short time later)

I can now get down to about 350 milliTorr. This is a significant step in the right direction, and should allow me to push the debug out as far as the laser tube.

(06 July, afternoon)

I have found two mediocre valves (I just hope they don’t leak...), and put them into the 1/4" lines that go into and out of the laser tube. I have put 1/4"-3/8" adapters on the lines from the valves. (These are all compression fittings, so they shouldn’t leak too much.) My next step is to tighten everything properly, cover the plastic quick-connects on the tube endpieces with RTV caulk, and then see how badly the tube leaks. (Also, perhaps, how badly the valves leak. If they are lousy, I will find replacements.)

Once I get that straightened out, I get to rework the connection from the helium tank and regulator to the tube; I think I have the requisite hardware, so this should be fairly straightforward — all I really have to do is bypass the existing gas-handling gear and go straight to the 1/4"-3/8" adapter.

When I can get the whole system down to less than 1 Torr, I get to see whether it will conduct electricity at any reasonable voltage when I have it filled with helium at a few Torr. It certainly should!

(late that afternoon)

I decided to eliminate the valves for now, because they would introduce an extra complication. I have not yet applied RTV to the plastic quick-connects on the tube, but with the TC tube at one end and the vacuum pump at the other, I can already pump the system down to less than 1 Torr. This is a very good sign. I may not even have to sludge up the quick-connects with silicone glop.

(late that evening)

The tube lit up very gratifyingly with air, so I put a valve on the helium inlet end. Then I let some helium into it, and I am pleased to report that I now have a helium sign. (Hey: it can’t be a neon sign, because I don’t have any neon.) The next step is to put some I₂ into it (I’m sure I have some around here someplace), and see whether it lases...

...Which it probably won’t do until I align the mirrors.

(a little bit later)

I did not get the mirrors properly aligned before the tube sprang a leak. My early guess is that even 20 or 30 watts is enough to cause significant heating. (The supply is rated to deliver 5 mA at 20 kV, so it probably delivers at least 3 mA at 10 kV, which is about where I had the voltage. I am, if I didn’t already mention it, using a 6-nf capacitor; at 10 kV that gives me stored energy of a few hundred mJ, which I hope will eventually prove to be enough to threshold the laser.)

I have, in any case, learned several important things from this prototype, and I should be able to do a better job next time, whether that involves plugging the leak[s] in this tube or building another one. Also, as of earlier today I have acquired some of the parts I’ll need if I decide to try a hollow-cathode design.

(Monday, 07 July, 2008)

I did my usual “dunk” test (put one end of the tube into a bucket of water; block one vacuum hose and blow into the other one; look for bubbles), and was unable to find any leaks at all, so I am now aligning the mirrors. This is fraught, as I do not have the tube firmly held in place, but I hope can get fairly close. So far, I have the far-end mirror lined up moderately well with the alignment laser, and I am tacking the O-ring and pivot ball to the moving plate of the near end mount, so I can put the mount back together without disturbing the tube any more than is really necessary. I probably ought to make some clamps, even momentary informal ones, to hold the tube firmly in place while I put the near-end mirror back on, and I may very well do so. I think I can safely say that the next version of this laser will be slightly more formal...

(Bit of an aside)

It has occurred to me that if I can get the pressure inside a suitable tube down to well under 1 Torr, I may be able to run it as a pulsed laser with other gases. The first obvious possibility is argon, but argon prefers a narrow bore. (If I recall correctly, deactivation of the lower laser level proceeds at least partly by wall collisions, so you have to keep the gas pressure down to the point at which the mean free path exceeds the distance from the center of the bore to the wall.) Because of this, it wouldn’t be a great candidate for this head.

I have some nice capillary tubing here, though, and there is some temptation to build a little pulsed argon machine as long as I’m fussing around. Argon also takes different mirrors, but I have plenty of them around, so that’s not a problem. The real issue is trying to align the mirrors when they are at opposite ends of two feet of 1-mm-bore tubing. (The pieces I have are 24" long.) That, I’m afraid, is not going to be a trivial exercise. Even just getting an alignment laser lined up with such a tube is extremely difficult, because the beam expands as it travels; even if it is narrower than the bore going in, it will be wider than the bore when it emerges at the other end, and wider still when it returns from the mirror to the target on the front of the alignment laser. This makes for an insanely confusing mass of reflections. I have successfully interpreted these for a 6-foot-long HeNe tube, but only just barely, and only once. That tube had 2 or 3 mm bore, which made it significantly easier.

One possibility is to put the beam through a lens with a very long focal length, so that the beam waist is near or at the far end of the laser tube. (Alternatively, if the mirror you are aligning is the OC, you may be able to get away with a beam waist somewhere inside the tube.) If you can do that, you have at least some chance of avoiding the walls entirely, or at least having less of a mess on the target. This is probably worth a shot, and I will probably go off looking for lenses, assuming I actually decide to try an argon tube at all. (It is nearly irresistible, once one has most or all of the relevant bits lying around, so I probably will.) The one thing I do not have is a glass lathe, which would make construction somewhat easier; but I may be able to fake that for the few moments I would need one, which would be to attach larger tubing to the ends of the capillary, so I could attach electrodes to the laser tube by making glass-to-glass seals. (Argon wants a bit more cleanliness than He:I2, and I have some electrodes from ancient flashlamps, arclamps, and HeNe tubes here, any of which would probably serve. Granted, they are in silica envelopes, but I have two Pyrex-to-silica graded seals, and I might be able to fake a graded seal by other means.)

Argon aside, there are several other lasers than can be operated in such a tube, including triply-ionized Oxygen. We’ll just have to see how it goes, I guess. Helium-mercury is another fine laser, but I am not going to use brass electrodes with it; I would have to switch to stainless steel, which does not form an amalgam with mercury. Although stainless compression tees are not as cheap as brass tees, they are definitely available, and this is a very interesting possibility. The He-Hg laser usually operates at about 567 nm (green) and 615 nm (red-orange), but other wavelengths may be possible under special conditions ...maybe. In any case, the 615-nm wavelength should run with HeNe mirrors.

(End of aside)

(early evening)

The near-end mirror is now on, and I am re-attaching the electrical connections. Because I eliminated the 4-40 screw attachments, which leaked vacuum, I have to use silver-conductive glue, which is not really intended for this type of service; it works well enough for initial testing, though. It is now drying. I took the opportunity to turn on the vacuum pump, which sounded like there were no leaks, so I guess I have another shot at this.

My guess is that it still won’t lase. I doubt that the mirrors are aligned well enough, for one thing. For another, the iodine is warmer than it should be for best operation. (We are at about 30° C here, well above the 15-20° that it wants.) Still, this is being instructive. If it does lase, I will try to take at least one photo of the spot on a piece of paper.

(late that evening)

Although the leak was not detectable by the “dunk” method, it was nonetheless present. I think it’s time to build a new tube. This one will be held firmly in place on an optical rail or an aluminum I-beam: I had forgotten just how flexible glass tubing is.

(12 July, early AM)

I now have several pieces of 9/6 Pyrex tubing (9mm OD, 1.5mm wall thickness, 6mm ID), and I have started setting up an old optical rail to hold the next tube. One of the first things I did was make an insulating post, on which I will put the “hot” tee that is the anode of the laser. One advantage of this setup is that I will be able to connect the tube to the electrodes with RTV caulk, something I did not succeed in doing with the first tube. That will accommodate differential thermal expansion nicely. I have not yet figured out how to make the electrical connections to the tees, but I can always use conductive epoxy if I have to. (I may build clamps that hold the tees on their posts, and simultaneously hold the brass shims on the tees...)

I am expecting to switch the 6-nf capacitor through the tube with a triggered spark gap, which will give me some flexibility in the amount of voltage (and energy) I apply to the laser. The repetition rate will be extremely low, but that’s okay for now. If/when I want more pulses per second, I can always change over to a small thyratron.

I am also continuing to consider a hollow-cathode version, but I am having some difficulty acquiring suitable parts. There is also the issue of ends; a 3-mm-ID square metal tube is not exactly easy to connect round glass tubing to. OTOH, I have some Brewster-angle windows that I can use, assuming I figure out how to build the thing at all. (More about this later, if I get any farther on it.)

(Afternoon, Saturday, 12 July, 2008)

The mirror mounts for tube #2 are going to be slightly different: instead of using a ball bearing as the pivot, I am just going to have three identical screws. These, rather than going directly into threaded holes, will be held by hex-nuts on the back of the baseplate, which is epoxied onto the brass tee. I am using #4-40 flat-head screws, which will give me a bit of pivoting, and having three of them will let me set the compression of the o-ring more easily. (I will try to take a photo at some point to make this clearer.)

I am also going to put RTV on top of the epoxy that holds the mirror-mount baseplates onto the brass tees, so that if there is a thermal expansion issue and the epoxy cracks, there won’t be a leakage issue. The epoxy is now curing, and the tees will be ready for their compression fittings in perhaps 8 more hours; once the compression fittings are on, I can affix the tees to their posts on the optical rail.

I believe I have figured out how to get the tees reasonably well aligned with the Pyrex tube: 3 of the tubes I just bought on eBay broke during shipment, so I have some relatively short pieces. I think I’m going to take one of those, cut the ends off reasonably square, and put it between the tees when I glue them down to their posts. (The posts are on carriers, so I will be able to move them to the ends of the rail after the glue sets.) I haven’t quite decided what I’m going to use for glue; it has to be strong, but it also has to be slightly compliant, because the tees will heat up during operation, and will expand. Most epoxies are a bit stiffer than what I want here, but I may be able to find one that’s a bit more rubbery than most. If not, I will look around for another good candidate.

The main tube will go into the tees with RTV this time, and because the tees are fairly firmly held in place, I shouldn’t need anything else — the tube will not be moved or stressed when I pump down the laser. I may take some steps to keep it centered in the bores of the tees, though, as that makes alignment easier and helps ensure that the excited region faces the central region of the mirror at each end.

(13 July, early AM)

I am now epoxying the brass tees onto their supports. In one case, the support is a hard plastic post that is, in turn, epoxied onto a carrier on the optical rail. In the other, it is an aluminum post that is screwed into another carrier.

I debated installing the compression fittings for the gas inlet and vacuum outlet hoses before gluing the tees to the supports, but upon thinking about it I realized that it was probably going to be difficult to position them in any case, and having random hoses flopping around, or even just extra unbalanced weight, would make it a lot harder. I am just hoping that when the epoxy has set, I will be able to attach the compression fittings without breaking the glue joints. The fact that I can take the carriers off the rail will help.

Once the epoxy has cured and the hose fittings are attached, I have to cut the Pyrex tube to length and RTV it into place. That’s nearly the last construction step, other than attaching the mirrors to the mounts and the hoses to the gas delivery and vacuum pump. Then I get to try aligning the laser, probably with vacuum on it so the conditions are as close to operating conditions as possible. (I will probably start by aligning the far mirror without vacuum, but it might be a good idea to put a flat AR-coated window on the near end, and find out whether the alignment changes much as I pump down the tube...)

(early AM, Bastille Day)

Here is a photo of the tube on the optical rail. You can see part of the power supply (gray can with white insulator out the top). The white object in the lower right foreground is one end of the 6-nf capacitor I used to drive the first tube.

At this point, the mirrors are approximately aligned, and the tube is nearly ready to test. Scott Scidmore has suggested moist Na₂CO₃ as a good scavenger for I2, and says that K₂CO₃, which I have on hand, will work just as well. Now I need to build a small trap to hold some of the stuff, after which I should be ready to test the tube.

(15 July, early AM)

I think I have a fairly good idea about the trap, which will be made of PVC pipe fittings and a sintered brass filter that I have in the junkbox. Unfortunately, the hardware store is not open at this hour, so it will have to wait.

(15 July, early evening)

I have constructed what seems like a reasonable trap, and filled it with what potters refer to as Pearl Ash (sub>2CO₃). The system does not yet want to pump down, but there are quite a few joints in the trap because of the parts that I was able to get, so I am going to put RTV caulk on as many of the obvious possible leaky locations as I can, and when the RTV has had a chance to harden, I will try again.

Here (once I get a chance to download the file) is a photo of the parts for the trap:

[Forthcoming]

(An hour or so later)

I have found and (I hope) plugged one large leak. That RTV now needs to harden, so There Will Be A Brief Delay.

(16 July, 3 AM)

I have now verified that the iodine trap is not leaking. This is a step in the right direction, but it probably means that the leaks (I’m guessing that there is more than one) are in the laser head. I don’t hear anything when I turn off the vacuum pump, and the pump is too loud for me to hear anything when it is running, so I am going to have to resort to bubble-liquid, which is a pain in the neck because it leaves a residue behind when it dries. OTOH, it does work, at least for medium and large leaks, so I should be able to find at least some of the problems fairly quickly.

That, however, won’t happen until tomorrow; it is past my bedtime now.

(16 July, afternoon)

Oddly enough, I found nothing with the bubble liquid. True, I applied more RTV to things last night, and that may have abated more leaks. The pump doesn’t sound quite as loud as it did yesterday. My next step, I think, is to attach the capacitor, pump down the tube, and look at the voltage at which I see breakdown, assuming that I do see breakdown. If it is low enough, I can always try admitting some helium and iodine. If not, I can begin to think about alternative leak-locating techniques.

(later that afternoon)

The vacuum is not good enough to permit a discharge yet, so I am thinking about other ways to figure out where the leaks are, and perhaps to get rid of some of them without necessarily knowing precisely where they are. We’ll see how this goes.

(that evening)

I gritted my teeth, pulled the tube off the bench, and dipped the ends in water. There was a big leak at each end. One of them was fairly easy — the epoxy holding the mirror mount onto the tee had failed, so I cleaned the mount and reattached it. Later, I will cover the joint with RTV, just in case.

At the gas inlet end, however, the leak is at the compression fitting, where the polyethylene tube goes into it. Apparently, it didn’t get tightened enough. I have torqued it, but only a tiny bit, because I don’t want to break the tee loose from its stand. I will retest it, and if it still leaks (which I am sure it will), I will probably apply RTV to it again.

Once all of this has had a chance to cure and harden, I will see whether the tube holds vacuum any better than it did.

(continuing through the evening)

Some progress: my reattachment of the mirror mount appears to have succeeded. Next, we get to see if I can stop the leaks at the compression fitting at the other end of the tube...

(24 July, 2008)

I had bought a nice Millipore vacuum gauge on eBay, and when it arrived I discovered that it needed 4-VCR fittings. I knew it would take a few days for them to arrive, and that seemed to provide a fine opportunity to get the vacuum pump serviced, so I did.

When the dust settled, I connected the gauge to the pump, did a little tweaking, and got a good enough seal for this work. (If it had been a real vacuum system, the leaks would have made it nonviable; but I only need to achieve perhaps a dozen Torr, so if it takes 15 or 20 seconds for the pressure to increase by 1 Torr I am still in good shape.)

Next, I connected the needle valve and the iodine trap. Things are much closer to marginal at this point: I can get down to perhaps 6 Torr, and it takes a few seconds to increase by 1 Torr when I shut the valve. It is possible that some of the residual pressure is water coming out of the K₂CO₃, but that does not account for the speed of the increase when I close the valve.

I am also becoming convinced that I need to build a third version of the laser tube. Fortunately, I have most of the design thought out, and I even have most of the parts on hand already.

(26 July, 2008)

As I continue to learn more about vacuum systems, I feel more and more strongly that it is time to rebuild the head of this laser. I think I have come up with a design that is less likely to leak badly than the previous designs did.

The first important detail is the attachment of the Pyrex tube: I have figured out that a 3/32"-thick o-ring that fits around the Pyrex can be used to replace the compression-seal parts of a 3/8" compression fitting. This should result in a decent o-ring seal, and also makes it really easy to demount the tube.

The second important detail is a redesign of the mirror mounts. I will photograph this when I get one built, as that is probably about the easiest way to convey the information. As an initial verbal description, though, the main difference is that instead of trying to make the holes for the adjustment screws in the washers (see photos, above), I am going to epoxy flat steel reinforcing plates on, as I did for the early versions of my regular bench mount.

(27 July, 2008)

I got some brass washers at the hardware store, and I am epoxying those onto the compression tees that will be the electrodes. The opening in this particular size of washer is just big enough that the end of the tee fits through it, but that’s okay as long as the epoxy is reasonably strong. (J-B Weld claims to have 3960 psi strength, which will do.)

I had to drill a hole at the corner of each of the steel plates, for the pivot point of the mount; this time I drilled all four of them together, in a stack, for convenience. I am currently epoxying the 3-48 hex-nuts onto two of the plates; when this epoxy has set I will put the #3 washers on the front plates (the existing holes are just large enough that the heads of my 3-48 adjustment screws fit through them) and simultaneously epoxy the plates to the brass washers, assuming that I can get it all to fit comfortably together.

After that, I still need to attach the brass shims from the capacitor to the tees; I have enough room for #2-56 screws if I can find short ones and if my tap-drill will actually work — it seems to be rather dull.

Once I get that well in hand (or decide to live without screws, and just use silver-conductive glue to attach the shims), it should be relatively easy to assemble the tube. Then comes the real fun: finding and eliminating the leaks...

(28 July, early AM)

I have some cogent advice about assembling the mirror mounts, should you choose to adopt my design:

1. Polish the facing surfaces of the washers that are going to touch the O-ring. Air can leak through even small scratches.

   [Note, added later: The more I think about this, the more important I suspect it is.]

   Alternatively, if you can be sure that your vacuum grease is compatible with your O-ring material, you can put the thinnest possible coat of grease on the facing surfaces, but do it after you glue things together. Grease and glue don’t go together well.
   
   

2. Attach the “L” plates to the outer washers before you attach the ones that go on the inner washers. It is extremely difficult to position everything if you try to make both attachments at the same time. (I know this from experience: I was in a hurry, and didn’t want to sit around for an extra day, waiting for epoxy to harden. I think mine will work, but I am fairly certain that they would have worked if I had taken the time to follow this piece of advice.)

I will take photos when the epoxy has hardened.

(30 July, 2008)

I have removed the polyethylene tubing from the anode, and I am in the process of gluing a piece of glass to the top of one of the aluminum posts, to serve as an insulator; I will be attaching a short stand to the top surface of the glass plate so I have some room in which to use wrenches, and the anode will go on top of the stand. I think I have successfully attached polyethylene tubing to the cathode, which is now on the optical rail, so this all may proceed nominally if I’m lucky. (If I’m not lucky, the torsion from the tubing will break either the epoxy joint, or the glass.)

(01 August, 2008)

I succeeded in assembling and roughly aligning the tube last night, and during that process it became clear that the mirrors need to be on their plates as parallel as possible, because otherwise the alignment is hard to accomplish. (Remember: because they are facing outward instead of inward, the curvature is working against us here.) It is difficult to place them correctly, because they are not flat; I removed and reattached them this afternoon, and I hope I have them fairly well angled now.

I also discovered that the o-rings I had chosen were not large enough, so I’ve acquired bigger ones, and I am holding them in place with small dabs of RTV so they don’t try to fall out when I assemble the mounts. Actually, I am holding one of them in place; the epoxy on the mirror holder at the other end was stressed by my attempt to align the mirror, and I am regluing it now. I will attach that o-ring after the epoxy has had a chance to harden.

(02 August, 2008, early evening)

I was unable to set the angles of the tees precisely when I glued them to their posts, and as a result it is extremely difficult to get the mirrors into alignment. I am presently re-attaching the anode tee, and we’ll see whether I get it at a slightly better angle.

Here is a photo of the two mirrors on their mounting plates, one facing up and one facing down:

Here is the anode tee being reglued to its insulating mount:

Here it is again, with guidelines (thanks to the Gimp) so you can see that the angle isn’t too far off. The larger image that you will get if you click the small one is a direct crop of the screenshot, and is as many pixels as I have...

(05 August, 2008, early AM)

Here is the anode, after reassembly. You can see that the tube is reasonably clean, and the mirror is in a fairly neutral position. The brass ribbon goes to the positive terminal of the capacitor.

I had no trouble pulling nearly reasonable vacuum on the tube after I put it back together, and I eventually got some discharges (photos forthcoming shortly), but no lasing: I was unable to align the mirrors, because I was using a HeNe. By the time the alignment beam got through the first mirror, it was barely visible; and there was absolutly no way I would be able to see a return from the second mirror with the first one in place. I tried aligning the far (second) mirror first, but these mirrors act as lenses, and when I put the first one back on, it effectively wrecked the alignment of the second.

I am currently working up a stand for my green laser pointer, and I have a green module on order. My hope is that I may be able to see returns from both mirrors if I use an alignment laser that runs at a wavelength that is sufficiently far from the design wavelengths of the mirrors; we’ll see whether this proves to be accurate.

Later in the afternoon I tried running it again, and noticed that I was having trouble pulling the pressure down quite as far. It was also more difficult to get it to pulse — the tube wanted to idle, with a fair amount of current going through it. I eventually managed to get it adjusted so that I could trigger it, but again there was no lasing.

Here is the tube; first idling (more about this in a moment) and then showing several pulses:

When I took a closer look after shutting the laser down, I discovered that there was quite a bit of pale dust in the bore. I removed the mirror from the anode, and sure enough, it was covered with the pale dust, so I cleaned it. Then I removed the mirror from the cathode, and was astonished at the huge quantity of dark dust that fell out. Here is the mirror with some of the dust still on it and some next to it:

I conclude that brass is not compatible with a mixture of air and iodine, and if I rebuild this tube I will use tees made of stainless steel. That may not be entirely exquisite, but it should at least be better than brass.

In the meanwhile I have cleaned up the mirrors, and I will clean the tees as well as I can. I am not going to try to swab out the Pyrex tube, at least for now; it is easier to cut another one.

(A bit later, but still early AM)

That seems to have worked, and the new piece of Pyrex is fine; I am now back to status quo ante — that is, the tube pulses when I fill it with a few Torr of the helium-iodine mixture, ...but it still doesn’t lase. I can, after a fashion, see returns from both mirrors; but each of the mirrors acts like a lens, and the result is so badly scrambled that I have not yet been able to interpret it. At least it does not (at least so far) show any sign of filling up with dust, though I have this awful feeling that the color of the discharge is a bit more purply than it originally was, which suggests an air leak.

I think that discretion is the better part of nearly everything (as my father would have said), and I also think it’s about bedtime. I will deal with this as time and tide permit. (I am obliged to go to Denver, and will be there until the end of the week, so it may be a while before I can post more about this.)

(13 August, 2008, early AM and late afternoon)

In Denver I attended the 66th World Science Fiction Convention. The convention was quite good, and I even got to be on a panel about lasers, with Ben Bova and Jordin Kare and Howard S. Smith. It was not about lasers as they are used in fiction, but rather real-life DIY and space-science uses. We played, as it were, to a full house. I enjoyed the panel immensely, and it looked like the people in the audience were also having a fine time. I am now back; the new windows have not yet arrived, but the green laser module has, and I have constructed an alignment tool from it by turning the power down almost as low as it can go, and gluing the module into a piece of PVC pipe. This should make the alignment process considerably easier. With luck, the windows will arrive within a day or two, and I will be able to make further progress. In the meanwhile, I will be setting up mirrors and mounts.

(15 August, 2008, early AM)

When I put the alignment laser into its mounts and removed the HeNe mirrors from the laser tube, I observed that even after just a modest number of pulses, the inside faces of the mirrors were already cloudy. If I don’t just rebuild the entire head using stainless steel tees, I am going to have to put the windows on extension tubes instead of mounting them directly on the existing flanges. (The windows have arrived; now I need to find some pieces of tubing that are of a usable size.)

(later that day)

Went to the local Ace Hardware and got a piece of CPVC tubing, cut two shorter pieces from it, and cleaned up the ends. I have attached one of these shorter bits to the anode, and the other one to the cathode; when all of the RTV has had a chance to cure, so the attachments are stable, I will put windows on them. I’m using PVC because it is nonconductive and “static-y” — I am hoping that any dust that forms will stick to it, and thus it will help keep the windows more or less unclouded, at least for long enough to permit the tube to lase.

Meanwhile, I continue to think about hollow-cathode designs.

(August 16th, evening)

The windows are on and should be reasonably well sealed, but the tube refuses to pump down. I can only get it to about 11 Torr. I’ve tried adding more RTV to the seals on the new end tubes and windows, and I’ve tried tightening various joints, but nothing seems to make any difference at all. I am retrenching a bit to think about this, while the latest batch of RTV cures.

(August 17th, afternoon)

By the time I went to bed last night, I could pump the head down to 5 Torr; but it still came up to atmospheric pressure within a minute when I shut the valve to the pump. I am thinking about leak-detection techniques. I don’t want to pressurize the tube, because the RTV would probably just come loose at the ends; as of now, I am thinking about dribbling a volatile liquid over each section in turn, and looking for a rapid change in the pressure inside as liquid wicks through the leak and then vaporizes. I have some Fluorinert™ fluid here that should work reasonably well, though I’m a wee bit antsy about whether it is truly compatible with the oil in the vacuum pump. OTOH, it has the advantage of being quite stable, and is more or less inert at room temperature.

(August 22nd, evening)

Last night, I concluded that the arrangement I had was more trouble than I was willing to put up with. Accordingly, I cut two larger pieces of PVC tube, and epoxied a brass washer onto one end of each. This gave me a hole that was bigger than the beam, but smaller than the windows, which let me seal all the way around the edge of each window:

(The windows are attached with rubber cement, which is sticky and which will retain a little bit of flexibility after it dries.)

I used the new green laser module to do a rough alignment of the tube, so that most or all of the beam of the iodine laser should go through the windows when they are in place, and I am now attaching them. When the RTV has hardened at both ends, I will see whether the thing pumps down, and things should proceed from there.

I notice that the plastic tube near the iodine source is no longer as dark as it used to be, which probably means that I have evaporated the few milligrams of iodine I originally put in there. I will have to add more before I try running the laser again.

(24 August, 2008)

I have added more iodine.

Replacing the windows and their tubes was clearly The Right Idea: the head showed no sign of any large leaks, pumped down nicely, and eventually reached 1.2 Torr as indicated by the Millipore gauge. Now I have to figure out where the small leaks are. (I’ve put the thermocouple gauge on the system, which may help at lower pressures because it has an analog meter as a readout; but it has very little sensitivity between 1 and 2 Torr, so it isn’t much help right now.)

Clearly, something like 0.4 Torr is iodine, so I am looking at about 0.8 Torr from leaks. I did see a reasonably whitish glow when I triggered the tube with 2 or 3 Torr of helium in it, which is a good sign, but probably not entirely sufficient. (I should put mirrors into place, though, line them up, and pulse the tube a few times, just to be sure.)

Back to the Index

Home

Contact Information:

My email address is a@b.com, where a is my first name (just jon, only 3 letters, no “h”), and b is joss.

My phone number is +1 240 604 4495.

Last modified: Sun Aug 24 13:33:22 EDT 2008