Joss Research Institute Web Report #15, part A: Toward an Affordable DIY Dye Laser

TJIIRRS, Report Number 15A:

Toward a Straightforward DIY Flashlamp-Pumped Organic Dye Laser
Step 1, First Proof of Principle: A Minimalist Machine

(Is it possible to threshold a dye laser with only 6 Joules input?)

(December 19, 2009, ff)

This page details the construction of a prototype dye laser that is intended for initial checking of some parameters — for example, I want to know whether a simple design will threshold “easy” dyes with minimal input energy. In principle, the answer is already known to be “Yes”; but in practice it may not be so easy. The laser will be operating close to margins, and any sacrifice of efficiency will be difficult to work around.

This preliminary design uses a commercial capacitor and a commercial spark gap switch, both of which I hope to eliminate in later designs. The machine that I’m working toward will almost certainly use a commercial flashlamp, though, because xenon is the most efficient emitter in the wavelength regions of interest for pumping organic dyes.

It will, very likely, also use commercial laser mirrors. Quite a few dye lasers have been operated with simpler mirrors, which could be homebrewed, but I am not at all certain that these machines will have enough “oomph” to overcome the losses inherent in [e.g., aluminum] mirrors. OTOH, for those who are willing to work at long wavelengths, using red and NIR dyes, sputtered gold mirrors may be viable: a clean coating of gold has reflectance of at least 98% from the far red on out through the IR, and a thinly-sputtered coating could serve as an output coupler.)


This laser uses high voltages, and capacitors that can store lethal amounts of energy. It puts out a laser beam that can damage your eyes and skin, and it uses organic dyes, some of which are known to be quite toxic. It also uses flammable organic solvents.

It is important to take adequate safety precautions and use appropriate safety equipment with any laser; but it is crucially important with lasers that involve high voltages and present a health and/or fire hazard!

Design Rationale

I have built a number of lamp-pumped organic dye lasers in the past, and I have read a number of articles about dye lasers in technical journals. In the course of doing so, I’ve noticed some characteristics that appear to define viability. Reasonably high optical pump power is required, and that is best obtained with a xenon-filled flashlamp. Xenon lamps can have efficiency as high as 50% or more under optimum circumstances. Granted, not all of the light can reach the dye solution, and not all of what reaches the dye solution can be absorbed by it; but those issues make xenon even more attractive as a source.

High pump power is also best obtained by driving the lamp with a short electrical pulse. This is one place where many DIYers go astray, by attempting to run their lasers with large capacitors at relatively low voltages. In order to create a short pulse you need to minimize both the inductance and the capacitance of the circuit that drives the flashlamp; operating at low voltage essentially makes that impossible.

Let’s run through a few numbers here so I can show you what I mean.

If you want to deliver 25 Joules to a flashlamp, and you are using photoflash capacitors that you charge to 400 Volts, you need 312.5 μf to store the energy. We can easily make a rough guess at how fast such a capacitor will discharge; ignoring resistive effects, the FWHM (Full Width at Half of the Maximum Amplitude) pulsewidth, in seconds, can be approximated as π times the square root of the product of the inductance and the capacitance, where the inductance (L) is in henries, and the capacitance (C) is in farads.

In an article on lamp-pumped dye lasers that was published some time ago, the authors determined that it is quite difficult to achieve system inductance of less than 125 nanohenry, so I am going to use 150 nh as the inductance value for this calculation. That’s optimistic; but it is consistent, and will do for now. (Just don’t expect an actual physical lamp driver circuit to meet the number predicted by this calculation; any real circuit probably has higher inductance unless it is extremely well designed, and in addition there are resistive effects that this calculation does not take into account.)

At 312.5 μf the expectable pulsewidth calculates to be about 21.5 microseconds, which corresponds to electrical power of about 1.16 MW. That may seem like a lot, but in fact it isn’t as much as we want. With actual photoflash or other electrolytic capacitors, which have high ESR and ESL (Effective Series Resistance; Effective Series Inductance), the discharge would probably be far slower, with correspondingly lower peak power, not even remotely close to what’s needed here. (Thanks to Jarrod Kinsey for bringing the ESR/ESL issue to my attention, and to Dr. Mark Csele and Dr. Winfield Hill for relevant information. Dr. Csele even has an oscilloscope trace on his Website, [right side, about ¼ of the way down] showing this problem directly.)

Now let’s try the same thing at 20,000 volts.

At 20 kV you need only about 125 nf to store the energy. The expectable pulsewidth is now just over 430 nsec, and the resulting electrical power is a little more than 58 MW. That is considerably more satisfactory. [Side note: if we guess that peak power occurs when the capacitor is at about 2/3 of its initial charging voltage, the peak current is a little over 4,000 Amperes (!).]

In terms of actual circuits, the situation is more extreme than I have presented it: photoflash and other low-voltage pulse capacitors are not designed to have low effective series inductance (ESL), so the system inductance will be larger than the amount I have used in the calculations. The use of wires to connect various parts of the circuit to each other also increases the inductance, slowing the discharge still further.

I trust that this explains my preference for high voltages, in spite of the difficulties and occasional annoyances involved in their use.

Circuit and Physical Layout

The circuit of this machine is quite simple: a capacitor is connected to the flashlamp by a spark gap, which serves as a switch. The other end of the flashlamp is connected to the other end of the capacitor, which is grounded. The one subtlety is that there is a small capacitor across the spark gap; it encourages the rapid formation of a good conduction channel when the gap is triggered. In order to be sure that this small cap gets recharged between pulses, the anode end of the flashlamp is connected to ground through a large-value (you want to use about 1 million ohms, or perhaps a bit more) high-voltage resistor. Here is a schematic diagram:

Wide pieces of brass shim stock are used for all of the high-current connections, to minimize inductance. (The shim stock I used for the initial build is too thin, but it seems to be working. Eventually I will probably replace it with thicker material.)

I am using a commercial spark gap here (an old EG&G GP-70), because I have one; I also have a trigger unit, which is handy. Both of these were acquired on eBay, so you could use similar items, but it is quite possible to build your own. Here is the spark gap:

The physical layout is determined by the relative shapes and sizes of the components, and by the need to avoid flashovers: we want the capacitor to discharge through the lamp, not around it.

My choice of 25 Joules in the example in the previous section was not entirely arbitrary; that’s about the maximum energy that I use in another dye laser that I put together a few years ago. This machine, however, is more “bare-bones” than that; I anticipate using only 30 nf, which stores just 6 Joules at 20 kV. I have several reasons for this, one of which is that the particular capacitor I’m using is a type that is often available on eBay. It looks like this:

Notice that the capacitor itself contributes 20 nh to the system inductance. The ESL of the spark gap is probably within a factor of 2 of this, and my guess is that the flashlamp is even worse. (There are good reasons why it is extremely difficult to build a driver circuit that has less than 125 nh total inductance.)

If someone actually wants to build a laser to this design they should be able to, even though it is not what I am actually aiming for, and despite the fact that there are some reasons why I wouldn’t actually recommend it.

The flashlamp I intend to use, at least for initial testing, is a little over 9" long. As of this writing, it is available from The Electronic Goldmine.

As supplied, the lamp has a trigger wire wrapped around it. I have removed this wire, as I do not intend to trigger the lamp; I will, instead, be firing it by overvolting it as abruptly as I can. (This is a fairly common technique. It is not as good as simmering the lamp, but it’s considerably easier, and will do for now. If this design turns out to be viable, I may attempt to build a simmer circuit later.)

Unfortunately, the lamp is not designed for fast-pulse service; it has skinny little wires out the ends, which are not good for a low-inductance design. It also has one other key shortcoming: the vendor’s listing claims that it is made of some type of borosilicate glass, rather than fused silica. That means it probably doesn’t pass much mid- to short-wavelength UV. For pumping Rhodamine 6G and other dyes that emit at relatively long wavelengths this may not be a problem; but it is likely to prevent the laser from operating blue, indigo, and violet dyes. We will do the best we can with it, and we’ll see whether that’s good enough. If not, we can change over to a different lamp (and/or, if necessary, more stored energy — I have two of the 30-nf caps).

[NOTE, added later: This lamp is significantly fragile. My first one suffered a few too many flashovers, and ceased to work after perhaps a hundred pulses. It is a good idea to provide a little extra insulation at the ends of the lamp, to keep it isolated from the ends of the dye cell; and it is a good idea to keep the aluminum foil close-reflector as narrow as convenient, even though that will lose you a little more light out the ends than is fully optimal.]

The Build

(20 December, 2009, early am)

Here is a view of the lamp driver circuit with the components positioned but not assembled:

I still need to: make the holes that will allow me to connect two of the shims to the capacitor; bolt the parts together; add the little starting capacitor across the spark gap; add charging and bleeder resistors; and clamp the ends of the relevant shims to the wires of the flashlamp. At that point the driver circuit will be done, and I can test it. Assuming that it fires the lamp and does so without exploding it or turning it purple from too much short UV, I can then attach the dye cell and attempt to threshold Rhodamine 6G in isopropanol.

The dye cell should be relatively straightforward, but it will be much longer than it needs to be for this flashlamp. This is because I do not want to cut the tubing — I may need a longer piece in future, especially if this design fails to reach threshold.

I am using compression fittings for the ends of the dye cell. They are convenient, and the tubing I have is a good fit. (I should note that I will be holding the dye cell tubing in with an o-ring, not with the usual parts, which are intended for use with either copper or polypropylene.) Here is a view of one of the fittings, with a hole drilled in it but without an end window:

(20 December, 2009, evening)

I am much closer now to having the lamp driver built and testable. At this point I need to connect charging and bleeder resistors, clamp the shims to the wires from the flashlamp, and provide power and trigger. With some luck, those things will occur later this evening.

Initial Testing: First Lamplight

(20 December, 2009, late evening)

I ran the lamp at various voltages. It appeared to handle all of them reasonably well, which is a good sign. Here is a view of the lamp driver, set up on the bench for testing:

Here are two closer views, first showing the lamp cold, and then showing it being fired, with 18 kV on the capacitor. (The first view is blurred by camera motion; my apologies.)

Although the flash is quite bright, I do not yet know whether it will prove to be bright enough. OTOH, at least I now know that the lamp handles as much energy as I initially intend to put into it.

(21 December, 2009, early evening)

The windows are mounted on the ends of the dye cell; I have the delrin rings and brass inserts I need for connecting the polypropylene tubing that the dye will flow through on its way to and from the cell; the cell ends are being attached to small mounting plates; I have a baseplate on which to mount the parts of the laser. I am now waiting for the epoxy to cure so I can mount the dye cell, set up and align the mirrors, and start testing.

(23 December, 2009, morning)

Two or things become clear to me after a certain amount of thought. First, I have mounted the dye cell ends incorrectly, and I need to fix that. I will be acquiring some half-inch aluminum bar. (Photo forthcoming later, after the new supports are in place.) I also want to bolt the mirror mounts to the baseplate, for stability, rather than simply attaching them with aquarium caulk. In order to do this, I will have to drill holes in the baseplate. Either way, I get to remove some of the paint.

I have also decided that the only source of cheap high-reflectance mirrors I currently know of is scrapped-out HeNe lasers. I even have several sets already on hand. These, unfortunately, are not broadband, so they limit the choice of dye to a very narrow range, but I have some Rhodamine 640, which should be a relatively good match, and will probably be the first dye I test with this laser unless I have some reason to try Rhodamine 6G or Fluorescein first. R640 may have one other advantage: it can probably be assisted by energy transfer from shorter-wavelength dyes, which would permit it to make use of more of the pump light. (This is actually true of various laser dyes, but the effect is only occasionally used. There can be complications under some circumstances, but I think I can avoid those by careful dye choice.)

(25 December, 2009, early afternoon)

After thinking about how to attach things to the baseplate, I decided to remove the small wooden plinths from the ends of the dye cell, and use something larger and metallic. Went to the hardware store to find some half-inch aluminum bar stock, and failed. Instead, I found some connectors for threaded stock. These look like very long hex nuts. I got a few, and eventually settled on the largest ones, which are intended for 7/16" thread. I sanded the irregularities off two faces of each, drilled and tapped two #8-32 holes in one of the faces, and glued the dye cell ends to the other face. Then I marked the baseplate, drilled holes, and bolted the dye cell down. This worked very nicely. I also drilled a hole for each of the mirror mounts, which worked fairly well except that for some reason I don’t quite understand, one of them is about 1/4" off to the side of where I want it. Still viable, though, at least for initial testing. (Photos of this progress are in the camera, and I will post them when I get a chance.)

Once I had everything in place I put a mirror at one end, shined a very small HeNe laser into the other end, aligned the HeNe with the dye cell, and aligned the mirror with the HeNe. Then I filled the dye cell with isopropyl alcohol, realigned the HeNe with the cell, and realigned the mirror. Then I put in the second mirror, aligned that, realigned the HeNe with the dye cell, and realigned both mirrors. (See the FAQ, below, for more information about this, with photos.)

At this point, all I need to do is wrap aluminum foil around the lamp and cell, fill the cell with dye solution, and start testing. It may take quite a few shots before I find the correct dye concentration, and I will probably need ot tweak the mirror alignment, but I am hoping to have a running laser some time this evening. (Fingers crossed; there is still no guarantee that this will actually work, and I may also have to swap out one or both mirrors — I was only able to find one HeNe mirror of a usable size; not sure where my other ones have wandered off to.)

(26 December, 2009, early AM)

I have not yet been successful in getting this thing to lase. I tried several obvious possibilities, none of which worked. My next move is to look for my other Maxwell 30-nf capacitor, and try 12 Joules instead of 6. Adding more capacitance is going to slow down the discharge, so I will get less than double the peak power, but I can still expect some improvement, and if the laser is only moderately below threshold, perhaps that will put it over the bar. OTOH, if there is some more fundamental flaw, adding another capacitor probably won’t help much.

In the meanwhile, here are some photos. First, one of the ends of the dye cell, which was filled with plain 70% isopropanol at the time, and a mirror (it’s the HR from the back end of a dead HeNe laser). Then the other end, with a mirror and an alignment laser.


It is not easy to see in these photos, but the bore of the dye cell does not receive all of the light that hits the outside. Here is a photo that makes this easier to see:

(The tubing appears to be curved, but of course it actually isn’t. I took this photo with my iPhone and a small lens, which accounts for the distortion.)

(26 December, 2009, early afternoon)

I have added the second capacitor —

— but the laser still did not reach threshold. Later today I will switch to Rhodamine 6G, and we’ll see whether that behaves any better.

(27 December, 2009, early AM)

After some thought, I decided that instead of trying R6G it would be a better idea to rebuild the dye cell...


(27 December, 2009, early AM)

Looking through my tubing stash, I found a piece that was just about the right diameter and length, and had a thin wall, so I built a new dye cell from it.

I looked for 1/4"-to-3/8" right-angle adapters, but didn’t find any, so I compromised by getting 3/8" compression to 1/4" MPT right-angle adapters and 1/4" compression to 1/4" FPT straight-through adapters. This is a bit more complex than I wanted to get, but I guess it will do. When I put the adapters together I used the thick pink teflon tape that is intended for waterpipes.

NOTE: If you build a dye cell of this sort, you will have to drill through the brass to provide a path for the dye laser beam. I drilled a small hole from the dye-cell side, because there was already an indentation left from the manufacture of the adapter that served in lieu of a centerpunch. I then turned the adapter around and enlarged the pilot hole from the other side, one or two drill sizes at a time, with the adapter firmly clamped.

I have run out of AR-coated windows for the moment, so I used small pieces of microscope slide instead; they are held on with silicone rubber aquarium caulk. Here are views of the ends:


You can see a tapped mounting hole in each of the “plinths”. It is about a quarter of an inch in from the end.

Because the glass or fused silica tubing is fragile, it is important to take the steps in the right order when you install a cell of this type. First, I will gently remove the tube. (It is present now so I can be sure that everything is in the correct position as I glue the ends to the plinths.) Then, holding onto the fittings (not the plinths), I will attach the polypropylene hoses that the dye runs through, and tighten the caps that hold them. Then I will mount one of the ends on the base. Then, with the fused silica tubing in place but held only loosely, I will bolt the other end into place. Finally, I will hand-tighten the caps that hold the tubing. This procedure worked well with the other cell, and I hope it will work with this one. [[Note, added the next afternoon: it did.]]

I have decided to swap out the baseplate, which I chose primarily because it was long enough for the initial dye cell. That is now out of the picture, and I have found a smaller plate that is much stiffer and will serve better. Here’s a photo, added the following day after I drilled the mounting holes in it:

This plate is brushed aluminum, about 1/8" thick, and it has stiffening ribs on its underside. It is far more stable than the previous base.

I also found my old HeNe output coupler and put it in the second mirror mount. There is some chance that I could use it with its concave side toward the dye cell, as I suspect that its radius of curvature is long enough, but I think that for the moment I will leave it with the AR-coated side facing the dye.

First Light

(27 December, 2009, afternoon)

Here is the laser on the bench, with the new dye cell in place, ready to go...

...and here it is, lasing Rhodamine 640 in 91% isopropanol:

The alignment is perhaps a bit sloppy, but I can deal with that later.

(A Helpful Hint, added some time later: If you use the original bottles from the drugstore, as I often do, you will find that when they are nearly empty they have a tendency to tip over. You can easily prevent this by going to the hardware store and buying two toilet flanges of appropriate size. These are relatively inexpensive, and they turn out to be quite handy for any number of purposes; I have even used them as telescope parts.)

(Later that evening)

I disconnected the second capacitor, to see whether I could threshold R640 with only 6 Joules into the flashlamp. The simple answer is “Yes.” The spot is not very bright, but the dye is definitely lasing:

I then reconnected the second capacitor, rinsed out the dye cell, and tried Rhodamine 6G in 91% isopropanol. My R6G is of low purity, but it clearly works: even a relatively dilute solution lased —

I added more R6G, and got a considerably brighter spot:

There was also a very small amount of output from the opposite end of the laser. It was very difficult to photograph, though, and is just visible here, slightly less than ¼ of the way down from the top of the image and a bit less than 1/6 of the way in from the left edge...

(30 December, 2009, early AM)

Yesterday evening, as I mention above, the lamp ceased to flash. Today I dug through my supplies and found a very robust lamp that I am using until the new ones I’ve ordered arrive from the Electronic Goldmine.

I am working toward thresholding some blue dyes, and learning as I go. This evening I succeeded in lasing Fluorescein in a solution that happened to include some 7-Diethylamino-4-Methyl Coumarin, though I have not yet lased that in this laser. (Note, added 06 December: this has changed; see below.) Oddly, I seemed to get much better performance with the dye flowing toward the output coupler than away from it; I am not yet sure what could cause this.

Alignment was difficult. After I did as well as I could with the HeNe, I was getting rather diffuse lasing. In fact, it took several shots before I really became convinced that there actually was any lasing going on. I kept at it, though, and eventually got a brighter spot. It seemed to be slightly off to one side, so I started tweaking the output coupler, a very small amount at a time. Eventually I got this:

(Clear evidence that the alignment is off.)

Just so you should be aware of the cooling issue, here’s what it looked like when I fired it again before giving the dye enough time to flow through the cell:

After a bit more tweaking, I got things lined up fairly well...

I should note a few things here.

  1. That’s a HeNe output coupler (at the lower left, in the photos), turned around backwards, just as I was using with Rhodamine 640 and Rhodamine 6G. It works passably well with Fluorescein, but is not likely to be good enough for dyes in the blue-green to violet range.

  2. The Fluorescein I have is intended for use as a biological stain; the label proclaims it to be at least 85% pure. (Proper laser-grade Fluorescein is almost certainly more than 99% pure.) Nonetheless, it lases quite nicely.

  3. As I already mentioned, the direction in which the dye is flowing seems to make a difference in the quality of the output from the laser.

  4. Fluorescein does not dissolve well in 99+% isopropanol, and it would be entirely reasonable to use 70% isopropyl rubbing alcohol as the solvent for it.

  5. Fluorescein performs significantly better in a basic solution than in a neutral one, and adding a few drops of strong ammonia to the dye solution is helpful. I have refrained from doing that here, however, because ammonia attacks brass. (I would guess that almost any strong base will attack brass, but I have not yet checked.)

  6. Patience is not just a virtue; it is crucial to this kind of work. (See the “too soon” photo, above, and the “not quite” photo, below..)

(05 January, 2009, evening)


(06 January, 2010)

The next three photos show Dharma Trading Company “Optic Whitener” in 70% isopropyl rubbing alcohol from the drugstore. In the first photo, it is not quite reaching threshold. (It tends to be very frustrating to see that sort of pattern again and again, but careful attention to detail can get you some information even when the laser isn’t working the way you want it to; if you ignore what you see in a pattern like the one in the first photo, even though there isn’t really very much to see, you may miss something you need to know.)

The reasons why the laser failed to reach threshold are not entirely obvious, but I suspect that the mirror alignment was one of them. On the other hand, I have seen it fail to lase when the mirror alignment probably was good enough, so I suspect that other things are also going on.

In the second photo, the dye is just barely starting to reach threshold. You can see the blue area in the middle beginning to get slightly brighter.

In the third photo, it is abundantly clear that the dye is lasing. In fact, it is bright enough to cause all 3 colors in the sensor to max out. (That is, although the bright spot appears to be white in the photo, it is actually blue.) I am extremely pleased, because I have wanted to lase this dye in a lamp-pumped dye laser ever since I first found it, and because I was far from certain whether 12 Joules would be enough to bring it to threshold. In fact, until it lased and I took the photo, I was not really sure whether it would lase at all in a FLP dye laser. (This was the first time I’ve had an opportunity to try it.) It’s a happy-making thing to have a firm benchmark.


There are several reasons why blue/indigo/violet dyes are more difficult to lase than dyes like Rhodamine 6G. For one thing, only some of them have fluorescence quantum efficiency as high as that of R6G. At least equally important, however, is the fact that as the wavelength gets shorter it inherently becomes more difficult to threshold any laser. (This is a good part of why it took so long to develop any X-ray lasers.)

I have not yet received another flashlamp from the Electronic Goldmine, so I’m still using the more robust lamp pictured above, which definitely has a fused-silica envelope. Thus, I don’t know whether this laser in its original state will be able to lase any of the blue dyes. I should also mention the fact that I am currently using a high-quality broadband “Max Ref” flat mirror as the rear reflector, and a helium-cadmium output coupler turned around backwards as the output coupler; these are probably about as good for ordinary blue dyes as HeNe mirrors are for R6G, and almost as good as they are for R640. They are considerably better than enhanced aluminum or protected silver.

(Later that afternoon)

It turns out to be even easier to lase 7-Diethylamino-4-Methyl-Coumarin. Here it is in 91% isopropyl alcohol, also acquired at the drugstore:

It looks to me like the mirrors are not aligned quite as nicely as I would like, and I will probably tweak them in an effort to improve the results a little.

I should perhaps point out that it is necessary to rinse the dye cuvette quite thoroughly when you switch from one dye to another. This is more important when you are going from long wavelengths to shorter; for example, my experience is that even a small amount of Rhodamine B will effectively quench a solution of Fluorescein, preventing it from lasing, while the Fluorescein output photos above are from a mixture of 7-Diethylamino-4-Methyl-Coumarin and Fluorescein, which lases quite happily. Not all such combinations, however, will be successful.

(08 December, 2010, afternoon)

The replacement lamp has arrived, and I have installed it. It does, indeed, appear to be borosilicate glass, as advertised: I have been unable to threshold 7-Diethylamino-4-Methyl-Coumarin with it, which suggests insufficient UV output. There is still some possibility, though, that the problem is partly a matter of pulsewidth rather than a lack of UV, and I will probably have to make some measurements to see which issue is more important.

I added some Fluorescein to the dye solution and was able to get that to lase. It was not as strong with this lamp as it was with the more robust lamp, but there is a different output coupler in the mirror mount now, which may help account for the difference in output.

It is good to know that even with the relatively inexpensive lamp this laser can provide green output.

(NOTE, added 08 January, 2010, evening: if you check the absorption and emission spectra of Fluorescein and the absorption spectrum of Rhodamine 6G, you will find that they are pleasantly compatible. This fact led me to try them together, and I was not at all surprised to find that adding some Fluorescein to a solution of Rhodamine 6G enhances the output.)

Pulsewidth Measurements

(09 January, 2010)

It was suggested to me by Harald Noack, of Graz University of Technology, that I simmer and prepulse the lamp. He referred me to an article, which sounded very interesting. I tried simmering, and found to my dismay that the laser barely worked at all. In order to understand this, I took some measurements of the light output from the flashlamp.

Here is an oscilloscope trace, showing the first obvious problem:

At 2 μsec per division you can see at least three peaks in the trace; my lamp driver circuit is underdamped, and is ringing. This deprives the initial pulse of some of the energy stored in the capacitor, and thus decreases the peak power. I’m considering what to do about this.

(I suspect, btw, that the sustained voltage between the peaks is an artifact; it doesn’t seem likely to me that the lamp would stay that brightly lit without much current going through it. Harald Noack suggests, however, that in fact the lamp does continue to emit a substantial amount of light for some microseconds after the end of an electrical pulse. It doesn’t really matter; the real issue here is not what goes on between the peaks, but the fact that there is more than one.)

I also photographed traces with and without a simmering current. These were taken at a faster sweep rate, and mostly show the first peak. The polarity of the simmer supply is negative, but the difference in performance is not pronounced. Simmer on the left, no simmer on the right. (In the photo on the right, the lower end of the spark gap and its little starting capacitor are connected to ground through a 400 K resistor. If I float them, the performance is not as good.)


If you examine these carefully, you will notice that the photo on the right has a slightly shorter risetime, and a significantly higher peak. These differences are more than enough to explain the fact that the laser barely works at all when I simmer the lamp.

(Later that evening)

For the sake of comparison, I put the more robust lamp (pictured above) back into the laser, and took two scope traces. The first shows the lamp output with the 400 K resistor to ground; for the second, I used a piece of aluminum foil to reflect some R6G output into the photodiode. It is probably instructive to note the longer delay between the triggering event and the beginning of the pulse, and the fact that the pulse itself is narrower than the lamp pulse. Both of these, of course, are entirely expectable.


The lamp pulse is single (I checked at slower sweep rate) and more symmetric, with pulsewidth of 400 or 450 nsec, while the dye laser pulse has a fairly abrupt risetime, begins near the peak of the lamp pulse, and has pulsewidth of about 150 nsec in this photo, though I have seen them as wide as perhaps 200 nsec. Both traces shown here seem very reasonable to me.

I then changed the simmer supply back to positive polarity and connected it to the lamp. I found that I had to use 800 K in series in order to get a sustained discharge, and it became clear that this lamp does not begin to conduct until the voltage across it is far higher than was necessary with the other lamp. This very strongly suggests that it has higher effective impedance, and that fact alone may be enough to account for the single pulse. Fortunately, the circuit does not appear to be overdamped with this lamp in place. Even so, simmering does not work well with it; here is a scope trace:

This clearly shows a slightly longer risetime and a noticeably wider pulse than the one just above, which was taken without any simmer current. I still do not understand why this should be the case, but it obviously is.

I then took a look at the laser output, with and without simmering. You will notice that even though I am just using a piece of aluminum foil as a reflector to get some of the laser light over to the detector, I have had to cut the sensitivity of the vertical amplifier in half because the lasing is significantly brighter with this lamp than with the other one. (It is easy to see the difference by eye when the beam is hitting the paper target.) These two photos were taken in fairly quick succession; the reflector and the photodiode probably didn’t move between the exposures, so the vertical scale should be close to the same in both. The one above, however, was taken at a different time and with things in different positions, so it is not directly comparable.


Both of these are about 200 nsec or a bit more, FWHM. The one without simmering is considerably brighter, which goes hand in hand with the pulsewidth and peak power. (Even if the difference in trace heights is partly an artifact, it matches my visual observations.)

(11 January, 2010, evening)

NOTE: This project continues on the next page, as I am making some changes to the design of the lamp driver.

A Bit of a FAQ

(21 December, 2009)

Here are some questions that people may think to ask, with some answers. Anything that does not have to do specifically and exactly with this particular laser is my opinion, and should be treated with mild skepticism.

  • Why do you include all of the mistakes, and the things that didn’t work?

    The errors and failures are your main pathways to a deeper understanding of the issues you’re facing. I present mine in the hope that they will help you reach a deeper level of understanding more quickly. To some extent they also expose my thought patterns and my approaches to issues and problems, in case those are of any interest or could in some way be helpful to you.

  • Why aren’t you using an elliptical reflector?

    I built my first dye laser in 1970. At that time, I was advised by some very savvy people that close-coupling the lamp to the dye cell would probably work. They also pointed out that wrapping aluminum foil around things is very much easier than trying to make a precise shape; that the interior of an elliptical reflector has to be kept very clean; and that a large amount of light escapes out the ends unless you use reflective end caps. I thought about that, and used aluminum foil. It worked just fine, and I have close-coupled every laser I’ve built since then.

    (I did try a diffuse reflector once, because it was available; but that setup failed to threshold R6G. Then I wrapped foil around the flashlamp and dye cell, and the laser worked very well.)

  • Why are you using a long lamp?

    This is best explained with a diagram. Until I have a chance to make a proper one, I will use ASCII-pictures.

    Here is a laser amplifier. (This is easier to explain with amplifiers than with oscillators, but the principle is exactly the same.) We will pump it with 28 units of energy. I am going to specify that it is operating well over threshold; let’s say it takes 4 units of pump energy to bring it to threshold, so the remaining 24 units are available to produce amplification. This particular laser is capable of amplifying by a factor of 10, so if the input is 1 unit of energy, the output is 10 units of energy...

          X10 amp 1 -> |LASER| -> 10       ^^^^^      28 pump 

    [[Please remember that these numbers have nothing to do with real life; I am just giving you an example here. In fact, I am going to call the length of this laser “five letters” to emphasize that fact.]]

    Now let’s double the length of the laser medium, to 10 letters, without changing the amount of pump energy. For clarity I am going to diagram this as two lasers, each of which is pumped with 14 units of energy. 4 of those units bring it to threshold, leaving 10 units to power the amplification...

           X4 amp          X4 amp 1 -> |LASER| -> 4 -> |LASER| -> 16       ^^^^^           ^^^^^      14 pump         14 pump 

    Notice that each amplifier now has only 10 units of available pump energy instead of 24, so it can only amplify by 10/24 as much. (That’s really almost 4.2X, but let’s call it 4X for simplicity.) Even so, we now get considerably more output. This will, though, be true only if:

    A) The system is running significantly over threshold,


    B) You don’t saturate the gain of the laser medium.

    If the gain is saturated, more length adds to the output arithmetically rather than geometrically, so you quickly lose the advantage. With dye lasers (and with nitrogen lasers) it is unusual to saturate the gain, so we aren’t going to worry about that issue. Ordinarily, all other things being equal, a longer active region will reach threshold more easily and will give you more output for a given amount of input energy, once you are above threshold. (I use a flashlamp with 15" arclength in my larger dye laser, and it works quite well at 20-25 Joules input energy.)

    As Jarrod Kinsey points out, the tradeoff here is that if you have a shorter lamp (and thus a shorter active length of dye), you can compensate by using mirrors with higher reflectance. There are two or three aspects to that; the first is availability and price — can you find mirrors with high reflectance, and are they expensive? The second has to do with the output coupler: if your OC is 99.5% reflective, you are not likely to get much output from the laser. (If you just want to know whether it has reached threshold, that probably isn’t a problem for you.) The third has to do with lamp life: a longer lamp, all other things being equal, has a higher explosion energy. That isn’t likely to be applicable to this laser, though, as we are not putting much energy into the lamp. The bottom line, then, is primarily the issue of whether you can find and afford the mirrors you need.

  • Why use such a high voltage?

    I explain this above, in the text.

    (In a sense, the specific voltage was chosen for me by the fact that I have a small 20 kV power supply, the fact that the capacitor is rated to handle up to 35 kV, and the fact that the GP-70 spark gap works well at 20 kV.)

  • Why all that brass shim stock? Why not just use wires?

    Broad connection paths help keep the inductance down, which is necessary.

    There is also another effect to bear in mind: with DC, a thick wire can carry lots of current, which is what we need to do here — you can figure that the peak current in this laser is on the order of 4,000 Amperes, as I mentioned above. With fast pulses, however, the current travels in a thin layer at the surface of the conductor. (Look up “skin effect”.) The surface area of a cylindrical conductor goes up linearly with the diameter, so it would take a very large wire to carry that much current. It is much easier to get large surface area with a piece of shim stock. (It also lets me get better contact to the long thin wire sticking out the end of the flashlamp, but in general I avoid lamps of this type, so that’s a special case.)

    I should note that the shim stock I used in my initial construction is actually somewhat too thin. I would recommend 5 or 6 mils if you want to build a laser of this type.

  • No Brewster-angle windows?

    It is difficult to cut things at the correct angle without a rotary table, which I do not have, and I don’t need polarized output in any case, so I punted this one. I wanted to use anti-reflection-coated windows on the second dye cell, as I did on the first one, but I didn’t have any more of that type on hand, so I cut pieces from a water-white microscope slide and used those. They are far from optimal, but I guess they’ll do for starters. (I have some AR-coated windows on order, and I may use a pair to replace the pieces of microscope slide at some point.)

    If I decide, in the future, that I want or need polarized output, I will put a Brewster plate in the cavity. That’s considerably easier than trying to find a way to mount the dye cell end windows at Brewster’s angle.

  • Why do you turn the output coupler around backwards?

    If you look at how an ordinary HeNe laser works, you will notice that the rear reflector is flat, and is at the center of curvature of the output coupler. This is a stable configuration, and fairly easy to align, but the active region is roughly conical. That, in turn, means that some of the excited lasing medium is not actually part of the laser, and any energy that goes into it is wasted.

    Moreover, it requires that the distance between the mirrors has to be set fairly precisely. If they are too far apart, only a much more narrow “cone” of medium can lase. If they are too close together the mode structure probably suffers, though that’s not likely with a dye laser of even modest size unless the mirrors came from a really long HeNe.

    I usually turn the OC around backwards because I want to use a larger proportion of the active material, and I also do it when the radius of curvature is too short for the distance between the mirrors.

    The outer surface of an ordinary OC is curved so as to allow the laser to produce a clean Gaussian beam. If you turn it around backwards so that it faces the active region it should act almost like a flat mirror [at least, at the right wavelength], and it may be very slightly easier to align. True, there is a small amount of reflection, but the surface is always AR-coated, so this is not usually an issue.

    There are unstable cavity designs that are more efficient because they use a larger percentage of the excited medium, but I don’t have appropriate mirrors to construct such cavities at visible wavelengths.

  • Why the hugely complicated alignment procedure?

    If you have never built one of these things, you may not have thought through the process of aligning the mirrors; it is unavoidably tweaky.

    Let me give you a runthrough. I hope that will make it easier for you to see why I had to jump through so many hoops, and I also hope it will make it easier for you to do when the time comes.

    First things first: you have to make sure that the path through the laser is unobstructed, and you need to choose an alignment tool that will work well. I use small CW lasers for this. If your alignment laser happens to be at a wavelength that is not reflected well by one or both mirrors you will probably have a hard time doing the alignment; likewise, if it is reflected too well by the mirror that is closer to it, you will have a hard time seeing the return from the mirror that is at the far end of the dye cell. I have several things I can use for performing alignments: a cheap green laser pointer, a variety of little HeNe lasers, a cheap red laser pointer or two, and a small violet diode laser. (I haven’t needed the violet diode for aligning anything yet, but it is handy anyway — I use it to check things for fluorescence.)

    It is also possible to use a borescope. A straightforward design is shown, IIRC, in the Scientific American “Amateur Scientist” column on either the homebrew HeNe laser, the homebrew argon laser, or both. I think it uses a small incandescent bulb, for which you could easily substitute an LED, as its light source; a microscope cover slip as a beamsplitter; an eyepiece from a microscope; and possibly one or two other lenses. If you decide to use a borescope, that column has good directions, so the rest of this discussion will cover the use of a laser.

    One suggestion for your alignment laser: make a paper target with a small hole in its center. (I like to print some convenient pattern of narrow lines on it, either vertical and horizontal bars, a series of rings around the hole, or both.) Tape or glue this target to the front of the alignment laser, positioned so that the beam emerges through the hole. The return is easy to see on the paper, and the lines or rings can make it slightly easier to figure out what is happening as you make adjustments. If you do not have such a target it is hard to tell where the return is and what it looks like, and that makes it much more difficult to achieve alignment.

    Once you have a viable alignment laser:

    1. Line it up with the dye cell. You want the beam to go right down the center from one end to the other. Even doing this can be very difficult, particularly if the laser is long, as this one is. I think it took me about 10 or 15 minutes to get my alignment laser aimed reasonably squarely down the middle of the dye cell, but that is partly because I was not using a good set of mounts for it.

      I find it convenient to put a paper target at the far end of the dye cell, so I can see what the beam looks like as it emerges. The patterns are extremely confusing, and it will take you a while to begin to understand them, but once you have an idea of what you are looking at it becomes somewhat easier to tell when your alignment beam is actually coming through the middle of the cell, and not bouncing off the walls on the way.

      If you darken the room, it is easier to see reflections from the walls of the cell, and you can “walk” the beam down the wall to the far end by making tiny changes in the position of the alignment laser or by moving the laser that you are aligning. My personal experience is that this is about the easiest way to get it lined up. For one thing, it is easy to tell if the alignment laser is aimed slightly up or down, because the beam will hit higher or lower on the wall of the dye cell as you walk it along. Depending on the dye and the wavelength of your alignment laser, you may be able to see the beam inside the cell, even where it isn’t hitting the wall. This can be extremely helpful.

    2. When you have the tool aligned with the laser that you want to set up, you can install the mirror at the far end. Tweak this mirror until it reflects the alignment beam straight back through the dye cell to the target on the front of the alignment laser.

    3. If you have not yet filled the dye cell with liquid, now is the time to do so. You will have to redo both parts of the alignment, but you will have had a little practice, which should make it slightly easier.

    4. Once you have the dye cell filled with liquid and the far mirror aligned, install the near mirror and align it. You will almost certainly find that the far mirror is now out of alignment, and you are also likely to find that the beam is no longer correctly aligned with the dye cell, so you will have to tweak a little; but everything should be very close at this point, so the tweaks should be relatively minor. (Should... a bad word!)

    5. If you can get the dye laser to operate, you may be able to tweak the mirrors again by viewing the output beam on a target some distance away. Assuming that your dye cell has a circular bore, the pattern should have a bright circular center. (See the output photos, above.) There is almost always some noise around this dot; the noise is caused by reflections from the walls of the dye cell. (It may be possible to eliminate or drastically reduce the amount of noise by including an aperture inside the cavity of the laser. I haven’t tried that yet, but I am setting up to do so, and if I get decent results I will post them.)

    If you have followed that description, you will begin to realize just how involved it is. In fact, I would hazard a guess that some people have been unable to get their dye lasers to work because their mirrors were misaligned, rather than from any lack of pump power/energy.

    Here is a set of photos, taken very late in the process. The first one shows what the target looks like when the mirror at the far end of the dye cell is blocked. Notice the bright line in the dye solution; Rhodamine 640 absorbs enough at 633 nm that you can see the alignment beam. (It is very hard to tell from the photo, but the dye emission is much closer to orange than the alignment beam. This is a quantum-mechanical refrigerator, and I should have patented it when I first noticed it, back in 1970 or 1971, with Rhodamine B. [[It was patented, just a few years ago. I grit my teeth.]])

    Notice the bright spot, above and to the right of the hole in the target. That’s the reflection from the front window. It will appear in all of these photos, and we are going to ignore it.

    (The color balance is also off in these images; I tried to figure out how to correct it, but that was not easy.)

    The next few photos show what happens as I adjust the mirror. Notice that the return spot in the first one is not round. This indicates some disturbance in the optical path through the dye. In the last photo, the mirror is approximately aligned. It may look like it’s aligned in the next-to-last photo as well, but if you examine it carefully you will see that it is slightly high and to the right.


    Here, for comparison, are three photos of returns that show serious disturbances in the dye solution. These are direct crops from the originals, so there are no larger versions. As mentioned above, the bright spot to the right and slightly above the central hole is a reflection from one of the end windows of the dye cell, and is not part of the return beam.


Here is a photo showing part of the dye cell, close to the mirror at the far end (away from the alignment laser). You can see two beams in the dye, evidence that the mirror is misaligned. That can also indicate a distorted optical path through the dye, though, so you need to be careful not to jump to conclusions. If you have any doubt, move some dye solution through the cell and watch to see how things change. If you can still see two beams after everything settles, you need to tweak the mirror alignment.

(The large version of this image is only 700 px across. It is a direct crop from the original file; I don’t have anything larger.)


Annals of the New York Academy of Sciences
Volume 168, Issue "Second Conference on the Laser" (February, 1969), Pages 703-714
J. F. Holzrichter, M.S. and A. L. Schawlow, Ph.D.

(Yes, that A. L. Schawlow, the one who got the Nobel Prize.)

On to the second page of this set, in which I try out some possible improvements...

Back to the Index


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

Contact Information:

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

My phone number is +1 240 604 4495.

Last modified: Wed Jan 13 00:15:08 EST 2010

Share this:

  • Facebook
  • Twitter
  • Delicious
  • LinkedIn
  • StumbleUpon
  • Add to favorites
  • Email
  • RSS

I am a Researcher of the Joss Research Institute. I work primarily on lasers and ceramics, with occasional excursions into other areas.

One response to “Joss Research Institute Web Report #15, part A: Toward an Affordable DIY Dye Laser”

  1. Joss Research Institute Web Report #10: Laser Dyes for the DIYer

    [...] in a small dye laser that puts only 12 Joules into its flashlamp. That laser is covered on Page 015a of this series, where there is a photo of the blue output from this [...]

Leave a Reply

WP Socializer Aakash Web