Initial Considerations and Design Issues
(20 March, 2007)
There are only a few viable methods for achieving a clean and sufficiently fast discharge in nitrogen at 1 atm. partial pressure, or in air. All of them involve significant levels of preionization; without this, it is essentially impossible to avoid sparking. The preionization can be present as a side-effect of the design or as a deliberately introduced feature, it can be inherent and automatic, or it can be externally timed and driven; but one way or the other, it must occur. (Externally controlled preionization is not particularly viable for TEA nitrogen lasers because the timing requirements are so tight. Fortunately, it turns out to be quite easy to have the laser preionize itself.)
Higher voltage is hard on dielectrics, but it provides compensating advantages; the amount of energy stored in a given capacitor increases with the square of the applied voltage, while the speed of the discharge (all other things being equal) decreases with the square root of the capacitance. Again, thinner dielectrics have higher capacitance per unit area, allowing the use of smaller plates; and they also have reduced inductance, which helps speed up the discharge.
It is likely that a cylindrical capacitor is the fastest physical configuration, btw; and in fact I have been told that a cylindrical water-cap that was resonantly charged to 120 kV by a 3-stage Marx bank was sufficient to operate a TEA nitrogen laser at well over 10 MW. That level of complexity is outside the scope of this page, though I do anticipate investigating cylindrical capacitors (and liquid dielectrics) in the followup project. I have already constructed a “TERP” nitrogen laser that uses a liquid dielectric in its peaker capacitor.
If you are having serious trouble achieving a clean discharge in air, it can be helpful to use nitrogen, or a mixture of nitrogen and helium in roughly equal proportions (though this will result in a longer pulse). Either of these, however, fails to guide you through the process of learning to produce a clean glow discharge in air, so you should think of it as a set of training wheels, rather than as your main approach. It is much better to learn to lase air first, and only then to make the transition to pure nitrogen.
Dielectric constant and dielectric strength are crucial issues. Water has dielectric constant of about 80, and impressive dielectric strength. It is, moreover, self-healing. Unfortunately, it must be extremely pure in order to work well, and because it is a liquid it tends to leak. For these reasons I am going to bypass it, at least in the early stages of the project. It may find some use when I get to cylindrical capacitors, or to a Charge-Transfer type of circuit, where the capacitor that drives the laser channel is charged only for a few dozen nsec..
I had anticipated switching the first laser for this page with an EG&G GP-20 spark gap, which is quite small and is rated only for voltages up to about 11 kV, but I couldn’t find it, so I used a GP-70 instead. The GP-70 is physically larger, and thus has more inductance (it takes the current-carrying path farther away from the baseplate), which can be a serious issue. In fact, I ended up going to a free-running open spark gap in order to get closer to Jarrod’s designs, as they are known to work and do not involve any expensive components. (There are more details about this below.)
The simplest TEA nitrogen lasers are built as LC-inversion circuits:
(I used a charging resistor instead of a charging inductor; but either, or a combination, can be made to work.) The small capacitor that appears just above the laser channel is one possible method for preionizing the channel. It was suggested by Milan Karakas, as a sliding or surface discharge: a piece of brass shim stock is connected to the anode side of the laser; it extends out on top of the top wall of the channel, which is a thin glass plate, and covers most (but not quite all) of the channel. This shim must be insulated from the cathode so it does not arc over the outside of the wall and thus short-circuit the discharge. In operation, as the voltage across the channel begins to increase, the voltage at the glass surface near the cathode rises very quickly because the brass shim is very close. This causes a surface-corona effect, which injects UV and ions into the channel, preionizing it. I have not built a channel this way yet, however, and so far I have not felt the need, as it seems fairly easy to achieve a glow discharge by other means.
I would like, at some point, to try semiconductor preionization, which has worked well for me in reduced-pressure nitrogen lasers. To do this, I would put a stripe of epoxy cement on the under surface of the glass “roof” of the channel; while the epoxy is freshly mixed and newly applied, I would sprinkle fine-grind (FFF) silicon carbide grit into it. There needs to be a 1-mm-wide gap down the middle, but there are many ways to provide that. Once the epoxy has set and I have shaken off the excess grit, I will use electrically conductive paint to coat the region just outside the epoxy and onto the edge of it, so that the electrodes can supply voltage to the preionizer. (Diagram to follow, if and when I actually get around to trying this method.)
The next issue is the upper capacitor plates. I purchased a 6-inch-wide brass kickplate at the hardware store, and sawed two 8" pieces from it for my initial design. The kickplate was stamped or sheared from a larger piece of brass, so the edges were reasonably straight and smooth. It was slightly curved, so I had to flatten the pieces a little, but a few short sessions of flexing them by hand and observing the results got them flat enough to be usable.
It also helps to have a power supply. I bought an inexpensive Wimshurst machine, as I did not really have time to build one; but I hope to construct either a Voss (Toepler-Holtz) machine, or what amounts to a horizontal Carré machine at some point. Here is the Wimshurst in operation, clearly showing that it can produce more than enough voltage for my needs:
I did make some modifications; the original brushes were
wires cut from pieces of window screen, and they were
tearing up the sectors, which are just aluminum paint,
apparently sprayed onto the disks through a mask. (If
you are building your own generator, btw, you should be
aware that Jarrod and others have discovered that India
ink works quite well.) I pulled out the stiff wire and
substituted Litz wire, which is fine and soft. It has
problems, too, but at least it isn’t destroying
the machine. At Jarrod’s suggestion, I also put a
wooden ball on the end of each collector. I had not done
this when I took the photo, so you can see (dimly,
through the transparent plastic bar and the rotating
disks) the original sharp-cut ends of the collectors,
which are lossy.
To assemble the laser, I took a flat brushed-steel tray
with raised sides, and put a piece of drawer liner in
it. Initially, I did not even bother to remove the thin
blue plastic protective sheets that were on both faces
of the liner. Then I put the capacitor plates down on
top of the liner, and added two lugnut covers as a spark
gap, as Jarrod does, except that I put conductive paint
between each lugnut cover and the metal it sits on, to
improve the connection and reduce sparking.
Here is the base of the laser, with the drawer liner
and the capacitor plates (and a cuvette of Fluorescein)
in place:
Note the resistor string, and the positions of the
lugnuts. (The piece of copper shim stock with the white
cliplead on it is for attaching the EG&G GP-70 spark
gap I mentioned above.)
I tried various items as electrodes, and eventually
settled on one of the two drawer or door handles that
you see at the left in the following picture, along
with one of the 5/32" brass rods at the right.
I bought the drawer pulls at the hardware store (they
are solid stainless steel, about 1/2" thick and just
over 8&5/8" long), and the brass rods at the local
hobby shop. Bending the brass is easy if you have a
propane torch; just be careful to grasp the end with a
pair of pliers or Vise-Grips, and to avoid touching any
hot metal with your hands.
I have carefully rounded and polished the ends of the
steel bars, which were originally just beveled, to
minimize the amount of fussing and sparking that will
occur during the discharge.
(See below for an even simpler version.)
Note the separation between the lugnut covers, which is
8mm or so in the photo (that’s actually a fairly
large separation; 6 or 7 mm is probably enough to begin
with)), and the separation between the capacitor plates,
which is somewhat less (but see below for specifics).
Fire the laser and note the location of the spark at or
near the lugnuts. Chances are that it will be a
“flashover” on the surface of the dielectric
unless you have taken steps to prevent that, as I did on
my next laser (see photos, below). If you only see a few
of these (say, half a dozen or so) before the spark gap
begins to operate, that’s probably okay; if, on
the other hand, you see lots of them, and the spark gap
just doesn’t seem to want to work, you have a few
choices:
When you have the spark gap firing cleanly, observe
what happens between the electrodes. What you want
to see is a lot of small white sparks, distributed
across essentially the entire length of the channel:
(There is a resistor blocking your view of the dye
cuvette, but I can assure you that the laser is
not lasing in this photo.)
If you see a lot more sparks at one end than at the
other, you get to short out the electrostatic generator
so you don’t get a rude surprise, and then
carefully reorient the capacitor plates so the edges are
closer to parallel. (You will find that they are
difficult to move, because charging them up puts rather
strong forces on the capacitor, and that squeezes a lot
of the air out. Take your time, and be gentle.) It is
best to use a shim of uniform thickness between the
plates, to be sure that the facing edges (which should
be straight) are parallel to each other. I have various
pieces of wood and plastic that I use for this purpose.
Notice, though, that there are hardly any sparks in
about the last 3/8" of the channel at the end nearest
the camera in the photo; a difference that small is
probably ignorable.
[I see a lot of photos of TEA nitrogen lasers with huge
weights holding the capacitor plates down to the surface
of the dielectric. If your base is reasonably flat and
your capacitor plates are reasonably flat, that is
completely unnecessary. The electric field will provide
plenty of force. OTOH, it is important to make
sure that the connection between each capacitor plate
and its channel electrode is good, and weights can be
very helpful there.]
Once you have good sparking all the way along the
channel, you can put your electrodes on the laser and
begin to position them. (You can look down on the
brass rod from above and wiggle it a bit until just the
edge of the capacitor plate is visible under it, all the
way across. Then you can edge it just a little further
out into the gap, carefully keeping it parallel to the
edge of the plate. This takes some practice, so give it
time and keep playing with it until you get it right.
Also, see
this page for a reasonably easy way to make small
adjustments.)
Here is what the laser looks like with the smaller rod
in place as its anode, and the larger one as its
cathode:
Note the fact that the leading edge of each of the
electrodes is slightly beyond the edge of the capacitor
plate that it sits on. Both Jarrod and I are still
looking into this; there is at least one suggestion in
the literature (in an article by Ernest E. Bergmann in
Applied Physics Letters, V28N2, 15 January, 1976) that
allowing the edge of one capacitor plate to protrude
slightly can improve preionization, and we need to try
that to see whether we get similar results.
The rods are not likely to be positioned correctly at
this point, and when you fire up the laser you will
probably see something that resembles this, though of
course it won’t be identical:
The purplish glow at the far end is what you want to see
in the entire channel. If you look carefully at the
sparks, you will notice that they are jumping between
the capacitor plates, not between the electrodes;
this is a clear sign that the electrodes are too far
apart where the sparks occur (in this case, at the end
closer to the camera), so I gently tapped the
near end of the steel rod to bring it a fraction of a
millimeter closer to the brass rod. Here is the result:
(Thanks to
Milan Karakas
for tweaking this photo to make it much easier to see
what’s going on.)
The large stainless-steel rod is not a particularly good
electrode for this type of laser; it is very easy to get
a clean discharge with large rods, but harder to get the
laser to reach threshold. Moreover, air is a terrible
lasing medium, and this laser is far from optimized, so
it just barely reached threshold.
You can see a reasonably even purple glow in the channel.
That’s the desired condition. (There are, however,
two bright spots on the left side, which are not
desired.) Please remember that the flat brass bar that
you see on the left side of the channel is just a weight;
the actual anode is the round brass rod underneath it.
You can tell this by the fluorescence on the dye
cuvette, which is a slanted line. (Look to the left of
the white pill-bottle, below and to the left of the
resistor.)
(To get a better view, click the small image and then
change “12c” in the filename to
“22c”. The larger image is my attempt to
perform the same enhancement that Milan performed on
the smaller one; I confess that I was less successful.)
Note that if sparks jump between the electrodes, this
can be a sign that they are too close together. It can,
alternatively, be a sign that the something else is out
of adjustment; I will mention one specific possibility
in my description of the next version of this laser.
I tend to use rods of slightly different sizes as my
electrodes, btw, with the cathode being a bit smaller
than the anode; but Jarrod reports that he gets his best
results when he uses rods that are the same diameter.
(10 April, 2007)
Last night, I tried several things. The first was to
put a commercial power supply on the laser. This
allowed me to get a sense of the operating voltage.
With a spark gap spacing of 8 mm or so, I had to apply
about 22.4 kV to the laser to get it to fire. Because
the gap is free-running, this value varies from shot
to shot, and I am not 100% certain of the calibration
of my variable transformer or my power supply, so
let’s just call it 22.5 kV. (At about 7 mm spacing,
the value decreases to about 20.5 kV.)
I found that I could get some lasing with air, and that
when I put nitrogen in the channel the spot on the
cuvette got a lot brighter. This is expectable.
Then I removed the thin blue protective sheets from the
drawer liner plastic, set the spark gap spacing at about
7 mm, and tried again. The results were very similar.
The increased capacitance of the slightly thinner
dielectric appears to compensate for the slightly lower
voltage.
At that point I put the GP-70 triggered spark gap on the
laser, reduced the voltage to 20 kV (the maximum rated
voltage for the GP-70), and found that although I could
not get a really clean discharge (there were always at
least a few white sparks), and was unable to obtain
lasing with just air in the channel, I seemed to be
getting better performance with nitrogen with the
triggered spark gap than with the free-running gap.
I will, however, have to retest to be sure; other
parameters almost certainly changed, which means that
this result does not constitute a proper test. (For one
thing, the electrode spacing and possibly also the
capacitor plate spacing spacing were different, because
I had to move things around in order to change gaps.)
(03-04 May, 2007)
Because I want to make some measurements, I threw
together another version of the laser. This one has a
base without any walls, and the dielectric overlaps it
almost everywhere. This turns out to be a big help in
avoiding flashovers from the edges and corners of the
capacitor plates, and I have decided that it is an
important design point.
In order to avoid flashovers under the spark gap, which
is the one place where I had to chop a hole in the
dielectric, and where there would be a path in any case,
I put two little ABS or styrene plastic “H”
beams (turned on their sides, so they are really being
“I” beams) on the dielectric. The I-beams
are held in place with “corona dope”,
high-voltage insulating varnish. Here’s a photo
from May 3rd:
(The piece of PVC tube blocks the light from the spark
gap, which would otherwise cause very bright
fluorescence in the dye, and make it hard to see the
laser’s output.)
The capacitor plates of this laser are 6"x10", and I am
using electrodes that are 12" long. I have not yet
tried adding a rear mirror to this laser, but because
the total pulsewidth is not much more than 1 nsec, and
light travels about 1 foot in 1 nsec, a rear mirror
probably won’t do much good. The only way to be
sure, however, is to try it.
The anode is a brass rod, 3/16" in diameter, and the
cathode is a flat brass bar that is 3/32" thick and 1/4"
or 5/16" wide. I have rounded and smoothed the ends of
both of these, but did not do any other special
preparation. (I am taking advantage of the fact that
the bar seems to have been sliced from a sheet of brass,
and has slightly sharp edges.) The electrodes are held
down on the capacitor plates by modest weights, as you
can see in the photos.
I used essentially the same alignment procedure
as before: I put the electrodes in place, fired
the laser, and noticed that because of their shapes
and the fact that the edge of the bar was just about
vertically above the edge of the capacitor plate
the spacing was very wide, so I got lots of long
white sparks. I took the electrodes off, moved the
plates a little closer together (they are now
perhaps 5mm apart), put the electrodes back on,
and fired the laser again.
There were a few white sparks, but the laser lased
anyway.
I then moved the bar a little closer to the rod,
and got this:
My apologies for the small size of the image; it’s
a crop, and the camera lens was not set to full
telephoto. Here is a larger one, from May 4th:
(If you want slightly more detail, click the small image
and then change the filename of the large one to
c0438c.flatround.airlase.c17.jpg — that will give
you the original crop.)
You can see what Jarrod Kinsey refers to as “icicle
sparks” on the cathode side of the channel, which
is on the right in the first photo and on the left in
the second; and you can see a fairly bright spot on the
cuvette, which in the first case is filled with Rhodamine
6G in isopropanol, and in the second is filled with
7-Diethylamino-4-Methyl-Coumarin, also in isopropanol.
(In both photos, the cuvette is out of focus.)
I have, btw, cleaned up some bad pixels so they
don’t distract; but the photos have not been
messed with otherwise, except to scale them.
Here is an end view, taken on May 3rd, showing the
working alignment:
The flat protrudes further past the edge of the
capacitor plate than the rod does. I will note that I
have observed an increased tendency to spark when the
anode (rod) is not far enough out; if you find that you
cannot eliminate white sparks or get a clean discharge
by moving the electrodes closer to and farther from each
other, check the position of the rod above the edge of
its capacitor plate. Also, check to see that the edges
of the plates are parallel to each other, and that the
electrodes are parallel to the plates.
As of May 4th, I started using slightly larger weights
(3/4" bolts, 6" long, in pairs) to hold the electrodes
in firm contact with the capacitor plates, and as of
May 6th I am using a stack of three such bolts on top
of a glass “roof”:
Here is a view into the end of the channel. My apologies
for the poor focus.
Relatively modest voltage (the spark gap spacing is only
5 mm) is enough to pump
7-Diethylamino-4-Methyl-Coumarin, if the laser channel
is filled with nitrogen. Here is a photo showing the gap
spacing:
The blue dye-laser output spot on a paper target is
somewhat diffuse, as one would expect when the dye is
driven to superfluorescence by a relatively short and
wide pump geometry. Here is a photo, taken on May 6th;
the spark gap spacing was increased to 8 mm, there was
nitrogen in the channel, and a cylindrical lens focused
the nitrogen laser’s output onto the dye cuvette:
The output (which is on a tilted paper target, in case
you were wondering about the slant) is left of center,
and the dye cuvette is near the right edge of the image.
You can see a bright spot in the middle of the output;
my best guess is that this is the result of reflections
off at least one of the walls of the cuvette, and the
more diffuse blue line above and below it is probably
superfluorescent lasing. (The blue dot slightly to the
right of the vertical line is less easy to understand,
but may be the result of a multiple bounce off slightly
misaligned surfaces.)
The combination of a flat cathode with a somewhat sharp
edge and a [slightly larger] round anode appears to be a
good one. Also, it is clear that if the electrodes are
long enough to extend about an inch beyond the capacitor
plates at each end, they do not need to have particularly
special shapes. (I used a file to round the ends, and
600-grit sandpaper to smooth them; I did not make any
attempt to polish them.)
(10 May, 2007)
I tried a quarter-inch steel rod instead of the flat,
and got good enough performance that I was able to
lase 7-Diethylamino-4-Methyl-Coumarin with just air
in the channel. This photo was taken from behind the
cuvette:
It is quite difficult to lase dye with an air laser;
you’ll notice that the dye laser’s output
spot on the target at left is small and not extremely
bright.
Here is an overview of the laser, with the weights
removed so you can see the rods:
To achieve 100 Volts per Torr centimeter, which is
approximately the optimum for nitrogen laser operation,
we need to put just under 23 kV across the channel if
the electrodes are 3 mm apart. The impedance of the
channel is changing constantly during the discharge
cycle, and is not really ohmic, but we can think of it
as starting at infinity and decreasing rapidly to a
value that is, at least in typical “TERP”
lasers, at most a few tenths of an ohm.
An LC-inversion circuit puts nearly twice its initial
charging voltage across an open circuit, so if I start
with a charging voltage of 20 kV I should have no
trouble exceeding 23 kV.
(The optimum seems to be fairly broad or ill-defined,
with some researchers reporting values closer to 80
V/T*cm and others reporting values as high as 200 or
so. I chose 100 because it is a convenient number.)
As I report above, I got very decent lasing with a
channel spacing on the order of 3 or 4 mm and initial
charge voltage of ~20 kV. Whether it is well optimized,
however, is another story. Fairly extensive further
testing will be required before I can really answer
that question.
To a set of references to significant articles about nitrogen lasers
To the first page in this set, a general discussion
of the issues involved in designing and building a
high-performance nitrogen laser
To a page about a high-performance nitrogen laser
that is a charge-transfer circuit rather than an
LC-inversion circuit, and uses water as the dielectric
in its peaker capacitor
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: Fri Jun 8 16:22:20 EDT 2007
Adjustment and Operation
More About Alignment; Another Version of the Laser
Operating Parameters and Related Considerations
Links
the Joss Research Institute
19 Main Street
Laurel MD 20707-4303 USA
Contact Information: