(February 8, 2010, ff)
This page details what I hope is almost the final step
toward a DIY flashlamp-pumped dye laser that stores a
very modest amount of energy at extremely high voltage,
and involves a minimum of expensive commercial
components. This design stage involves building
capacitors that can handle 60 kV, and will have (I hope)
reasonably low effective series inductance (ESL).
This laser uses very 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.
In some of its configurations it uses spark gaps that generate powerful acoustical pulses when the laser is cycled; these pulses can (and will!) damage your hearing if you do not use adequate protective gear. (The earmuffs used by shooters should be sufficient; foam earplugs may or may not be.)
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!
(08 February, 2010)
The previous version of this laser, or at least of its flashlamp driver, used a single-stage Marx generator and a single isolation gap. I expect this version to use a single capacitor or an array of capacitors connected in parallel, and a single spark gap for switching. The gap will probably free-run initially, but I hope that eventually I will be able to build a triggered one. If the capacitor does not discharge rapidly enough I will be obliged to add a “peaker” capacitor. The peaker, if it is required, will almost certainly use a liquid dielectric.
(19 February, 2010)
Here is a schematic:
The schematic shows a triggered spark gap, but as I mention above I will initially use a free-running one. It also shows the negative side of the power supply grounded, but I may not bother; it depends on how the driver behaves when it floats.
(08 February, 2010)
Powering a laser of this type with an electrostatic
generator presents an interesting challenge. Instead of
20,000 Volts, we could now be dealing with as much as
perhaps 100,000. Generators are rated in microamps
rather than milliamps, and corona losses interfere with
our ability to generate and store the amounts of energy
we require. Dielectrics must be thick in order to
withstand the supply voltages, and terminals must be
separated by larger distances, both of which increase
the ESL of the capacitors. OTOH, at 60 kV it takes only
about 7 nf to store 12 Joules, which should help a bit
with discharge speed, and at 75 kV, the required
capacitance is even smaller.
(08 February, 2010)
At this point, I am still thinking about ways to
build suitable capacitors. There are several
more-or-less standard methods, all of which seem
bulky, ungainly, and suboptimal. A stack of glass
plates, one obvious method, is also rather heavy.
Large amounts of PVC pipe would be quite expensive.
Although it is at the limit of DIY capability, I am
beginning to think about TiO2 powder with a
calcium boroaluminosilicate glass binder. This material
can, if appropriately processed, have dielectric
constant about equal to that of water and dielectric
strength of about 1200V/mil, which would be viable. The
real questions are whether ordinary TiO2
powder (which I already have) will work; whether I can
come up with an appropriate glass (the composition is
not fully specified in the one reference I’ve been
able to find so far) and whether I can actually make
sheets of the material that are suitable for use in
real-world capacitors. For use at 60 kV they would have
to be about 1/8" thick (figure that some headroom on the
voltage is necessary, and that the dielectric strength
goes down as the thickness increases), and for 7 nf the
capacitor plates would have to be about 7" square, or
about 8" diameter if circular.
Frankly, despite all of the caveats and issues,
that’s considerably better than anything else
I’ve come up with so far, ...if I can bring
it off at all.
(10 February, 2010, late evening, with notes added later)
A day or two back I calculated the capacitance of a
Leyden jar made from a 5-gallon plastic bucket. My
assumptions and guesses were as follows:
The area of the “cylinder” is about 380
square inches, and the area of the bottom is about 82.5
square inches. That gives us total area of about 462.5
square inches. Coupled with the other information this
gives us capacitance of 1906 pf, which we will call 1.9
nf as the starting data were not accurate enough to
warrant more than 1 decimal place.
At 60 kV this stores about 3.4 J, so we would need at
least 3 of the things. The obvious problem is that the
large physical size essentially guarantees somewhat
higher ESL than we would like. OTOH, a 5-gallon bucket
costs very little, and the notion is fairly attractive
for that reason. Mind you, other components would be
required; conductive paint is not cheap, and it would
almost certainly be necessary to plate copper onto the
initial coating in order to get a thick enough
conductive layer. Also, some kind of insulating rim
would be necessary, just under the lowest lip of each
bucket, to help minimize the corona losses. This would
have to be made from a piece of plastic sheet, which
introduces additional expense. Still, it is very DIY,
and that’s good. I think I even have an idea about
how to make decent connections to the plates.
(15 February, 2010, late evening, and 16 February, afternoon)
Jarrod Kinsey has gotten sparks more than 8" long with a
Leyden jar made from a 5-gallon bucket, so I think we
will try for ~75 kV, and I have acquired 2 buckets. My
reasoning is as follows: I am currently using 15 nf. I
am seeing risetimes on the order of 80 nsec, and the
initial peak (the pulses are slightly underdamped) looks
to me like it is roughly 250 nsec across, FWHM. This
corresponds reasonably well to system inductance of
about 400 nh. If I use 2 buckets, each of which is close
to 2 nf, the capacitance will be only 1/4 of what it is
now. This means that I should be able to tolerate system
inductance of about 1600 nh without much change in the
risetime or pulsewidth.
At 75 kV, the two buckets together will have capacitance
of ~3.8 nf, and will store a little more than 10.5 J. If
the system inductance is even as high as, say, 1200 nh,
triple what it is now, that should still be enough
energy to provide adequate peak power in the flashlamp
...if the discharge is not particularly underdamped. It
remains to be seen whether I can achieve critical
damping.
As a preliminary check, I dropped a 2nf 50kV capacitor
directly onto the Wimshurst, and took a scope trace of
the resulting spark. This was difficult, because the
Wim generated a large amount of electrical interference,
but I did eventually get a few traces recorded. Here is
the best of them:
There are some interesting features to be noted here.
First, the risetime is only a little bit more than 50
nsec. That’s pretty decent. Second, the system at
least looks like it is grossly underdamped — there
are at least 7 peaks visible in the trace. (See
below.) My hope is that this will be less of an issue
when I am using a better spark gap and there is a
flashlamp in the circuit. The result is quite
encouraging, despite the underdamping.
(Just by the bye: if we assume that the initial peak is
a fairly clean half-sine about 100 nsec across, and we
ignore the capacitances of the two small Leyden jars
that are built into the generator, this performance
corresponds to system inductance of slightly less than
600 nh. That’s fairly encouraging.)
(18 February, 2010, early am)
I put a flashlamp across the Wimshurst, along with the
2 nf capacitor, and left a small gap so there would be
a slightly higher charge voltage before the lamp
actually lit up. Here is the result:
Superficially, this looks a lot like the previous one,
but as you begin to look more closely you discover a key
difference: the horizontal sweep is at 100 nsec per
division, not 50. I now think that the smallish
“wiggles”, of which there are 13 or 14
visible here, have something to do with the Leyden jars
that are built into the generator, and are not actually
multiple pulses of the lamp.
The risetime is roughly 100 nsec, and the FWHM
pulsewidth appears to be between 350 and 400 nsec. (The
peak is so badly obscured by noise that I am not sure of
the actual height, which makes it difficult to decide
where the half-height is.) If this were actually at
75 kV or so, and if the capacitor were 4 nf instead of
2, it might even be viable. That is, I take it as
a good sign. Still, it is very artificial, and it does
not represent the expected conditions very well.
To get back on track:
You can make an ordinary Leyden jar by gluing aluminum
foil around a bucket, inside and out. This is easy, and
it does work, but there are some issues. Anyplace where
there is air trapped under the foil, you are losing
capacitance. In addition, as you use the capacitor the
trapped air ionizes, and eventually it damages the
dielectric under it. When the damage gets bad enough,
the capacitor fails. Also, although it is certainly
possible to make good electrical contact to aluminum
foil, it is perhaps not the easiest thing in the world.
I thought about that, and decided that I should use a
conductive liquid instead. My first thought was copper
sulfate, which I already know has been used in some
high-voltage resistors, and which should be compatible
with brass shim stock. Reading up on conductive liquids
of common sorts, though, I found that ammonium chloride
has 5X better conductivity. Granted, it is necessary to
use stainless steel shim stock, but that is readily
available. Ammonium chloride has other advantages: for
one thing it is relatively nontoxic, and has been used
in cosmetic and food products. In addition, it is
available and not terribly expensive.
My current thinking is that I will wrap the shim stock
closely around the buckets, and that the liquid will
fill any spaces that would otherwise be air pockets. I
think I know how to make connections to the inside
conductors as well as the outside ones, and I think I
have an idea about how to keep the terminals isolated
from each other. Lisa Peoples, our Range Safety Officer,
has suggested standing the buckets in a plastic tub,
which is good for several reasons, one of which is that
tubs of the required size (about 12 x 24", and about
11 or 12" high at the rim) are readily available.
I have ammonium chloride and some shim stock on order,
and we’ll see what happens when I try to put all
of this into practice.
More as it happens...
Back to the first page of this set,
which covers the initial design and development of the laser.
Back to the second page of this set,
which covers the first set of refinements.
My email address is a@b.com, 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: Thu Jun 23 15:55:48 CDT 2016
2: Capacitor Design and Construction
This work is supported by
the Joss Research Institute
19 Main Street
Laurel MD 20707-4303 USA
Contact Information: