TJIIRRS, Report Number 15B:
Toward a Straightforward DIY Flashlamp-Pumped Organic Dye Laser
Step 2: Improvements
(January 5, 2010, ff)
This page details some things I am trying in an effort to enhance the performance of the laser that I described on the previous page.
!! CAUTION !!
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!
1: Overview and Review of Simmering Results
(11 January, 2010)
Harald Noack, of Graz University of Technology, suggested that I try simmering and pre-pulsing the flashlamp. He cited an article in which the authors obtained 20% improvement in lamp performance just by simmering, and obtained further improvements in laser performance (particularly with blue dyes) and lamp lifetime by prepulsing the lamp.
Simmering involves passing a DC current through the lamp. With lamps that are driven at relatively low voltages this considerably changes the switching and triggering requirements, which ordinarily depend on a lamp that is not initially conducting. With a dye laser like the one here, however, the situation is quite different. A lamp that is already conducting should be, if anything, even easier to work with than a lamp that is not yet started.
Although the authors simmered their lamp at 20 mA, I have seen other papers that listed simmering currents as high as about 100 mA; Harald reports that his lamp works well with 10 mA simmer current, and I gather from this that any particular lamp may have a “sweet spot” that is perhaps best found by trial and error.
It seems to me that as long as the simmer current is present when the laser is fired, it should not be necessary to have it be present at all times. Eventually I will work up a pulsed simmer arrangement in which the simmer current is turned on perhaps 1 millisecond before the main store is discharged through the lamp, but just to get started I think that simply turning the simmer supply on shortly before pulsing the lamp and turning it off shortly thereafter should suffice.
In the meanwhile, however, my early trials with simmering (detailed on the previous page) did not give me the expected results. With either of the lamp types I have used in this laser, simmering impairs the performance. For this reason, I am [temporarily] abandoning it in favor of other techniques.
I should note that the authors of the paper specifically state that the improvement in performance that they saw with simmering resulted from improved imaging of the lamp on the dye cell. I am not imaging anything; my lamp and dye cell are close-coupled, so it comes as no surprise that I did not observe the same improvement in performance that they did. By the same token, however, I still do not understand why there was an impairment.
Another Issue: Damping
With the lamp from The Electronic Goldmine in place, I find that the circuit I’m using is underdamped, and oscillates. With the other lamp, which has very different characteristics, it does not. I mentioned this to Harald, and he suggested that I try two of the Goldmine lamps in series. That would allow me to decrease the inductance of the driver circuit by moving the conductors closer together, which should speed up the pulse a bit; I am hoping to try it, and I will present the results here.
(11 January, 2010, evening, and 12 January, early AM)
I have come up with a slight redesign of the driver circuit, which I built using thicker brass shim stock (6 mils, roughly 0.235 mm). Here are the paper mockups that I used as plans:
I built the actual version, assembled it, added some sheets of plastic to help avoid flashovers (see the photo on the left, below), and attempted to attach the lamps to it. In the process, the thin wire broke off one end of one of them, revealing yet another reason why I strongly dislike this typ of connection. Here are photos of my initial jury-rig:
I expected this to fire with somewhat less oscillation than the original version, and I expected a faster pulse. This is what I got:
It looks fine, very much like the ones I took with the “robust” lamp (see previous page) and with only a tiny hint of a second peak, …until you notice the timescale. In the photos on the other page, the sweep was at 200 nsec per division, while here it is at 500 nsec. The pulse is slow, and I have not been able to obtain any lasing yet.
Thinking about it…
(12 January, 2010, afternoon)
Jarrod Kinsey asked a question in email that caused me to remember a different configuration, one that I first thought about some years back but was unable to build at that time because it requires two capacitors with essentially the same value. I obviously have two such capacitors now, and I am going to rebuild this driver in that configuration. We’ll see how it works.
(Later that evening)
Here is the configuration, a simple voltage-doubling circuit:
(This circuit is similar to the ones often used in nitrogen lasers, where it is usually described as a “Blumlein Circuit”, which is ridiculous: Blumlein’s actual circuit design involves matched transmission lines, a matched load, and an extremely fast switch.)
I expected faster risetime and possibly a more narrow pulse. This is what I found:
We are back to 200 nsec per division, which is good. The pulse rises smoothly, in about 400 nsec, and appears to be a single peak. Occasionally, though, I see the leading edge as it appears in the second photo, with a plateau about halfway up (give or take a bit). Sometimes the plateau is more pronounced than it appears in this photo. Moreover, sometimes the lamp does not flash, and I have not yet seen any trace of lasing. My suspicion is that this driver is not getting enough of its stored energy into the lamp. It was certainly worth a try, however, and I may test it with the other lamp, to see whether there is much difference.
(13 January, 2009, evening) In thinking about this circuit and the results I obtained, I concluded that there was another handy way to get a fairly fast risetime and high voltage: instead of a voltage-doubler circuit, what about a single-stage Marx generator? The difficulty with a Marx generator or Marx bank is that it presents the full power supply voltage at its terminals during charging. This means that unless the load will ignore the power supply, you need another switch of some sort. The easiest way to accomplish this is to put a free-running spark gap in series with the output. I was all set to build one, when I discovered that I had already done so. It isn’t very fancy, and it is a bit smaller in diameter than I had in mind, but it was there, so I used it. (Photos of this setup forthcoming). I suspect that my flashlamp has ceased to be a lamp and is now an inert piece of glass, because it failed to fire, and I got flashovers when I added a close-wrapped aluminum-foil reflector. Nonetheless, the sparks of the flashovers gave me enough light that I was able to get a trace from the oscilloscope. It looks like this:
(13 January, 2009, evening)
In thinking about this circuit and the results I obtained, I concluded that there was another handy way to get a fairly fast risetime and high voltage: instead of a voltage-doubler circuit, what about a single-stage Marx generator? The difficulty with a Marx generator or Marx bank is that it presents the full power supply voltage at its terminals during charging. This means that unless the load will ignore the power supply, you need another switch of some sort. The easiest way to accomplish this is to put a free-running spark gap in series with the output.
I was all set to build one, when I discovered that I had already done so. It isn’t very fancy, and it is a bit smaller in diameter than I had in mind, but it was there, so I used it. (Photos of this setup forthcoming).
I suspect that my flashlamp has ceased to be a lamp and is now an inert piece of glass, because it failed to fire, and I got flashovers when I added a close-wrapped aluminum-foil reflector. Nonetheless, the sparks of the flashovers gave me enough light that I was able to get a trace from the oscilloscope. It looks like this:
The risetime is now about 125 nsec, and the pulse is no more than about 400 nsec FWHM. This looks good, but it is important to remember that a piece of aluminum foil is not exactly the same thing as a flashlamp. I will have to try the “robust” lamp, to see whether I can get a proper flash out of that.
(Later that evening)
It seems that I need more resistance in series with the power supply output; I think the spark gap continued to conduct after one of the flashes, and damaged the 400 K HV resistor that is on one of the caps:
(I have 800 K on the other cap, and that appears to be enough.) The pulse, however, is just fine risetime still less than 200 nsec, pulsewidth just about 400 nsec:
This was just a temporary lashup, however. I need to find a better way to hold the lamp in contact with the brass shims before I will be able to run the laser again. Meanwhile, here is an overview of the Marx setup, without the lamp:
The brass shims on the main spark gap are considerably longer than they need (or want) to be, and eventually I will shorten them. That should give me just a bit more speed.
(14 January, 2010, early evening)
I ended up holding the ends of the lamp this way:
(The large bolt is part of the peaking gap, and is not related to the clamp.) The contact area between the lamp end and the brass shim is considerably smaller than I would like, but it will serve for now. I am thinking about better clamp designs.
It should come as no surprise that this device is a laser.
Unfortunately, the peaking gap no longer holds off the power supply voltage, so I was unable to take a scope trace of the laser output. I will be obliged either to rebuild the gap, or to build a new one. Still, this design has clearly proven itself. I should probably note that the dye solution shown here has been in the laser for several days, and should have been replaced. It worked anyway.
(Later that evening)
When I opened the gap to inspect it, I found a small burned area on the plastic:
That was all it took to unbalanced the field and cause the gap to fire at [relatively] low voltage. To help prevent tracking across the surface, and to decrease the influence of small burns, I am switching to a plastic shell that has internal threads. Here it is with one of the acorn nuts present:
When the RTV has set I will put the other acorn nut in, and I think I will be drilling some small holes in the envelope to allow some airflow. Otherwise, it gets quite hot in there with repeated pulsing.
(15 January, 2010, morning)
I drilled holes around the housing of the gap, sanded the outer faces of the acorn nuts for better contact, and tested the gap. The spacing was not wide enough, and the driver self-flashed several times (I checked, and it did lase), so I turned it off and I’m moving the acorn nuts a bit farther apart.
I am also building another gap, with adjustable spacing. Here’s my idea, in parts:
I have used a hacksaw to remove the two threaded sections from the 5/8″ compression fitting, and I am currently epoxying them into the electrodes; they are just barely long enough, fortunately. I’m holding them in place with the compression nuts from the fitting, which will later hold the machine bushings against the brass shims, compressing them onto the backs of the electrodes. I must now be patient, and wait for the epoxy to cure.
In the meanwhile, the original gap is now wide enough to hold off the HV supply, so I took a look at the laser’s output:
(This is the same tired mixture of Fluorescein and R6G.) Nice steep leading edge, with risetime (if you don’t count the little dip just before the peak) of ~10 nsec; the FWHM pulsewidth looks like it’s on the order of 120 nsec.
I am slightly reluctant to superimpose this trace on the lamp trace. For one thing, the lamp trace was taken without the aluminum foil close-wrap, though I would not actually expect that to make a particularly large difference. In addition, I’m not fully certain about when the trigger events occurred. I am fairly sure, though, that the dye laser pulse (in yellow) cannot reach its peak after the lamp light does, so I have positioned it with that in mind. (The fluorescence lifetime of the dye is quite short, so the dye solution can’t store energy for more than a very few nsec. As the lamp pulse peaks and begins to decrease, the fluorescence [and lasing] will also start to decrease, in a time too short to be visible on the screen of the oscilloscope at this sweep rate.)
(I have brightened the leading edge of the laser pulse here, to make it easier to see.)
Here is an actual lamp pulse trace, with the laser running:
Here’s a version with the vertical scale expanded, to make it easier to see the shape:
The risetime looks like 150-160 nsec, and the pulsewidth is perhaps a little less than 400 nsec. Slightly slower risetime than I got without the reflector, but the pulsewidth doesn’t seem to have changed much.
Here is the R6G output on the wall, roughly 7 feet away from the laser, with the room lights turned out:
Here is the rebuilt spark gap, in place on the lamp driver:
Here are the two electrodes of the new gap, with the threaded sections from the compression fitting epoxied into them; the compression nuts are next to them, and the PVC fittings that I will be sawing apart to make the gap housing are in the background, with the bushings leaning against them:
Here they are again, with the bushings and compression nuts on them:
I have not yet built the housing, but it shouldn’t be too difficult..
(18 January, 2009, early AM)
Here are the halves of the new gap:
Once again I must be patient, and give the epoxy time to cure thoroughly.
Meanwhile, although I was not able to get a trace of the output from “Optic Whitener”, as I do not yet have the dye solution properly adjusted (and the mirrors may not be aligned as well as they should be), I did get a photo of the output on the wall:
This is a bit ragged, and I will be fussing with the solution at some point, to see whether I can get better results. I may also try a different output coupler.
(18 January, 2010, evening)
I assembled and installed the new gap:
You can see a 30-Megohm HV resistor at the upper left; it is 1/3 of the bleeder resistor that is across the flashlamp. I managed to get a few traces of the lamp output without lasing, and one with Optic Whitener lasing. Fortunately, one of the nonlasing lamp traces was just almost the right height to match the laser output trace (they were taken with the detector in the same position, which helps), and I was able to superimpose them. Triggering was unstable, so the divisions on the face of the scope are not in the same places in both images, but the timing of the laser pulse with respect to the lamp pulse should be reasonably accurate the traces overlap nicely where the laser is not running.
The lamp pulse, very handily, has its peak right on one of the vertical lines on the scope face, so it is easy to see that the peak of the laser output actually appears to occur slightly before the peak of the lamp output. I am not sure why this should be, and I will be keeping an eye out during further testing, to see whether it seems to be common. (It could, of course, be an artifact.)
(20 January, 2010, evening)
This afternoon I tried adding a small amount of sodium hydroxide to the dye solution, to see whether Optic Whitener is compatible with it. This is partly because I had a hunch it might improve the output slightly, and partly because I wanted to try mixing Fluorescein into the solution, and Fluorescein has higher fluorescence efficiency in basic solutions. As far as I can tell on a single trial, NaOH does help Optic Whitener a tiny bit, and it also appears that Optic Whitener helps Fluorescein, perhaps partly by energy transfer, but almost certainly by emitting light that can pump the Fluorescein.
(22 January, 2010, evening)
This is just Fluorescein (plus some NaOH); no other dyes are present, at least not intentionally.
(Again, I was fortunate enough to capture a lamp trace that is about the same amplitude as the lamp portion of the laser trace, so I could superimpose them.
(23 January, 2010, evening)
Thinking about the performance of the laser since I put the new gap in, I wondered whether it might be spaced a little too wide. My logic here is based on the fact that the laser ran with the older gap, even when it was too closely spaced and was going off on its own. (The other end of that spectrum would be a gap so wide that it didn’t conduct even at the full 40 kV, which would result in 0 J being delivered to the lamp.) I haven’t really researched this, but my sense is that in the absence of conflicting factors the gap wants to be spaced as closely as possible, consistent with good control of the Marx generator. Having concluded this I removed it from the driver, screwed the insert further in (about 2 turns, I think), and put it back in. Here is a lamp trace, over which I have drawn a smoothed version:
Here is a laser trace. (As before, I have brightened the leading edge, which was so fast that it was rather dim.)
Here is a fairly typical “far-field” pattern, as usual on the wall about 7 feet from the laser:
I think this represents something of an improvement. The lamp risetime is back to ~125 nsec, and the lasing seems brighter. (Note, added 24 January, 2010, evening: I strongly suspect that one reason why the lasing is not as bright with Fluorescein as it is with 7-Diethylamino-4-Methyl-Coumarin is that the output coupler I’m using is optimized for blue, not green. I may have a better OC for green, and if I do I may test this notion.)
(24 January, 2010, late evening, about an hour after I added that note)
This should serve as a reminder that most of the parameters of a FLP dye laser can be optimized. (In fact, many parameters of most projects can be optimized!) Assuming it was labelled correctly, the output coupler currently in place has 6 meters radius of curvature (so I have faced it toward the dye cell rather than away), and transmits 1-3% between 520 and 531 nm. I did a particularly careful alignment when I put it into the mirror mount…
(Sorry about the shadow on the left side; it’s a power cord that was in the way.)
(25 January, 2010, late evening)
The new lamps I ordered from Electronic Goldmine have arrived, so I installed one. I must point out that they are quite fragile; you will need to be extremely careful when you clamp the ends. (Yes, I broke one.) Here is one of the new lamps, as shipped, next to the “good” lamp, with a ruler for scale:
Despite my misgivings about the short arc length (10 cm / 4″), this lamp performs well. Here is a trace of its output, this time at 100 nsec per division. I have drawn a smooth curve (which is regrettably crude, and for which I apologize) over the noisy original.
Notice that the risetime is now about 100 nsec, while the FWHM pulsewidth is still roughly 400 nsec. For whatever reason, this lamp appears to be a good match to the driver circuit.
Here is the laser output, also at 100 nsec per division:
Again, the risetime is quite short; the FWHM is very roughly 120 nsec, which is quite acceptable.
(Notice that this was taken with the vertical sensitivity turned down. When I tried it with the same vertical scale as the lamp output photo, it went right off the top of the screen. Unfortunately, there is a problem with this vertical amp, and I need to get it recalibrated; the pulse should be somewhat taller than it appears here.)
(26 January, 2010, late evening; photos taken early AM)
I decided to check some things. First, here is a reasonably representative photo of the output on the wall, with the new lamp:
I added a small amount of “Optic Whitener”, to see whether it would help. It did not:
Then I emptied and rinsed the dye cell, and set it up again with 7-Diethylamino-4-Methyl-Coumarin (“Coumarin 1”) in 91% isopropanol, to see whether that would work with an output coupler that is optimized for green rather than blue. Answer:
That is, not as well as it would with a mirror optimized for blue, but it certainly isn’t bad.
Please note, btw, that all of these are essentially identical crops from the originals, so they are directly comparable in terms of the size of the pattern. The next one is also comparable.
(29 January, 2010, late evening)
As part of the next phase of this project, I have installed a peaker capacitor, on a temporary basis, as a proof of principle. The details are on the next page, but here, for comparison, is a photo of the output:
This is considerably less pronounced than the similar one above, but there are good reasons for that.
Performance and Utility Issues
(20 January, 2010, evening, updated 04 February, 2010)
I initially said I would be remiss if I failed to mention the fact that this laser is often more finicky than I would like, but as of early February it is behaving itself considerably better. I can pulse it faster than once per second if I have the dye flowing at just the right speed, and when the mirrors are properly aligned it puts out a fairly reasonable beam. I think the lesson here is that if even one key parameter is marginal the machine will be difficult to deal with; and if more than one parameter is out, the machine is guaranteed to be painful if it works at all.
Before I put a pump in, though, I may want to try replacing the end windows on the dye cell, as the nice AR-coated ones have arrived and are waiting. I keep thinking that they should be better than uncoated bits of microscope slide. (I did end up putting coated windows in place, and they certainly aren’t any worse than uncoated ones. My sense is that they are better, but it is very difficult to make the relevant measurements to be entirely certain.)
In its current (Marx generator) configuration it is also quite loud, and I have been wearing earplugs whenever I work on it. If I weren’t making frequent changes, I would probably put it inside a protective box. If you build one of these, please be careful of your hearing; like your vision, it cannot be replaced!
I have already mentioned the fact that the shims on the main switch are longer than they should be, which is at least slightly slowing down the pulse. I am also still using the special flashlamp, but I should be able to swap that out within a few days, as the other ones are on their way. (See above. As of 2010.0125 evening, this has happened. Also see the next page in this set.)
(06 and 11 January, 2010)
In order to make this work I need two power sources, two switches, and a way to prevent the two pulses from interfering with each other. I also need to be able to set the timing of the two pulses with fair precision: they need to be only a few microseconds apart, and I do not yet know the optimal interval, or even whether there is an optimum.
(Note, added 25 January, 2010: Frankly, the performance is now so pleasant that I am reluctant to mess with it.)
More as it happens… Last modified: Thu Feb 4 13:30:55 EST 2010