Date: Mon, 28 Aug 1995 12:39:29 -0600
From: Richard Quick <richard.quick@slug.org>
Subject: CAPACITOR INFO

I just posted several times on the TCBOR rolled plastic
capacitor. Since the detail of this particular unit was pretty
well covered, I will focus on other homemade types; the flat
stacked plate type capacitor, a little on the salt water cap, and
a little on capacitor theory as it applies to Tesla coils.

I have seen several types of homemade stacked plate capacitors.
The two types differ as to the orientation of the plate stacks.
Some are stacked vertically, others are stacked horizontally.
Before I go into construction details I should cover some of the
advantages of flat stacked plate caps for use in Tesla coils and
other high voltage applications.

Flat plate caps have very little inductance. Rolled caps contain
two or more plates which are tightly rolled up. Rolled plates
exhibit some properties of coils, they contain a certain degree
of self-inductance. This limits the size of the rolled cap in
Tesla applications. As plates grow in size, the self-inductance
grows, and the caps exhibit self-resonance that will interfere
destructively with the oscillation of the Tesla tank circuit. The
rolled cap that I posted about previously is self resonate at
about 7 megaHertz.

Flat plate caps are better adapted for pulse applications. Rolled
caps have to discharge a long plate. The further away the open
end of the plate is from the high current terminal, the longer it
takes for the cap to discharge. In essence this distance is also
an extension of the tank circuit wiring, as the plate gets longer
performance decreases. As the rolled cap gets larger, efficiency
of pulsing drops off.

Flat plate caps can be constructed to handle higher voltages.
Rolled caps have efficiency limits in individual units as to the
breakdown voltage. A single dielectric is used per plate. If
dielectrics are made thicker, efficiency drops off, if made
thinner efficiency increases, but they break down. Using standard
materials, the rolled cap I posted about is at the edge of this
design limit as well. 

Flat plate caps can be built for larger capacitance. The rolled
cap, because of the design constraints listed above, won't give
you much additional capacitance without increases in losses,
problems with self-resonance, and lowering of the capacitor Q.

The rolled cap that I posted is a good unit. I have built nearly
20 of these caps, and I use them a lot. But do not look to expand
much on this design. It has passed through several improvements
and I really think it is pushing the design limits in all of the
important areas. Next we need to look at the flat plate cap, as
there is much to be done yet, but first look at the dielectric.

The best dielectric for homemade Tesla capacitors is low density
polyethylene plastic. Whether you build rolled, stacked plate, or
salt water caps you should look hard at this plastic before
settling on anything else. It has an extraordinarily low RF
dissipation factor for the cost. The actual "in use" dielectric
constant on homemade caps using this plastic is right around 2.
This is a little lower than the book value, but homemade
applications of this dielectric rarely have the close plate
bonding that are achieved commercially with clean room vacuum
presses.

This dielectric melts at 100 deg. C. But because of the very low
dissipation factor the plastic is subject to very little in-
ductive heating. There is little loss, therefore little heating.
When using this plastic however, it is imperative to cover in
mineral oil to distribute any heat that is formed, suppress
corona and displace air. Plastic caps not covered in oil are
almost guaranteed to fail in seconds. Plates, dielectric, and oil
MUST BE CLEAN!... BTW The cheapest and most common plate material
is aluminum. In the rolled cap, aluminum flashing is available
precut in a perfect plate width, and there are other widths
available. Flat plate caps can use flashing, but it is frequently
more cost effective to use foil.

Now that we have established a few basics, lets talk plate cap
design. The first type of flat stacked plate requires the cap be
pumped down to a pretty hard vacuum to remove air. This is the
horizontal stacked plate capacitor. Typically these are built in
a Tupperware type storage box. Plastic, plate, plastic, plate
etc. are stacked one atop the other to build up the value. The
breakdown voltage is directly related to the dielectric thick-
ness used. 60 mil poly sheet is recommended and will have a
breakdown voltage in the Tesla tank circuit between 11-15 kv
rms input voltage in pulse discharge applications. This of course
depends on the quality of material, and the cleanliness of the
construction.

Once the box is filled, and all parallel plate connections are
made, high current busses are brought through the lid of the
container and sealed airtight with hot glue. Then the lid is
snapped on, and it too is sealed with a bead of hot glue around
the edges. The next part is important:  A single hole is made in
the lid for the vacuum connection. A fitting is hot glued into
the hole and a hose is attached to the vacuum pump. The cap is
pumped down, then the hose is clamped off and disconnected with-
out allowing air back into the cap. Submerge the hose in a bucket
of clean mineral oil and release the clamp. This allows the oil
to backfill the capacitor, and displaces the air that was
removed. Once backfilled to normal pressure, I pump them down a
second time, and repeat the procedure to make sure that all
trapped air between the plates is removed. Air bubbles will form
corona hot spots that will cause dielectric failure. 

The vertical stacked plate capacitor is much like the cap I just
covered. But the vertical cap does not require pumpdown. A tank
is used to hold the vertically stacked plates and dielectrics.
The unit I examined was built in a glass fish tank that employed
no metal in construction. Stiff foam padding was laid in the
bottom of the tank, and wedged in around the sides of the
vertical capacitor stack to cushion it and wedge it in place. The
foam padding also reduced the mineral oil required to cover the
stack. If foam padding is used it is important that "sponge" type
padding NOT be used. This padding can release latent air bubbles
into the capacitor plates. Use a quality "sealed-cell" padding. 

The reason these caps do not require pumpdown is that eventually
the oil will displace the air trapped in the unit. A break in
period of low voltage operation assists the removal of trapped
air, as the pulsing of the cap vibrates the plates and agitates
the air bubbles. The disadvantage of the unit I examined was the 
glass fish tank. I have seen plastic waste cans that could be cut
down for use as a tank in this construction.

Higher Qs, higher voltage, and additional capacitance in stacked
plate capacitors can be easily obtained. The trick is to use
thinner dielectric.

The dielectric strength of polyethylene is given as 1000 volts
per mil, but this is not the case in Tesla coils. The standard
breakdown voltages of a dielectric are calculated using a static
DC voltage. When you run AC across the dielectric, the breakdown
voltage must be divided by two. Then you must figure that the
peak voltage from a AC sine wave is higher than the rms voltage
most people go by. You meter won't see it, but your dielectric
will. Then you have resonate rise in the Tesla tank circuit. To
give you an idea of resonate rise in a tank, think about the
tidal forces that can be created with timed pushes in a bathtub.
It don't take much energy to push water over the side. The same
principal operates in the tank circuit in a coil, especially with
a synchronous gap system. The current pulsing back and forth from
capacitor plate to capacitor plate causes a voltage rise that
appears on the dielectric in the capacitors. The standard 60 mil
poly is supposed to hold up to 60,000 volts per the book. I have
blown holes through 60 mil poly with a 12 kv neon sign xfrmr in a
Tesla tank circuit and my gap wide open. My pinky finger fit
inside the hole.

One of the neatest homemade stack plate caps I have seen was
built by Bill Richards of T.C.B.O.R., the cost was pretty low,
the materials came from his laundry room, the grocery store, and
the drugstore. The only thing required was 56 hours of time in
arranging the plates according to Bill. But he did end up with
.03 uf 15 kv pulse capacitor in a five gallon bucket. It was
quite a performer on his coil at 3600 watts!

He shopped around for one gallon ziplock freezer bags with a 3
mil thickness. With a sharp scissors he cut the ziplocks off of
the tops of the bags. Then he cut aluminum foil squares that fit
inside the bag leaving a 1/2" of space around all four sides of
the plate. So the plate had dielectric borders 1/2" on all sides.

When two bags were stacked on top of one another, there were two
layers of dielectric, for a total of 6 mils. Being practical,
Bill figured correctly that the stacked bags would hold up to at
least 1000 volts rms input in the Tesla tank. He built up stacks
that had a value of about .45 uf each, with each stack rated at
1000 volts. Then he wired stacks in series. 

By squeezing fifteen stacks vertically into a bucket, and
covering the whole thing in about three gallons of mineral oil,
he got the required capacitance at the required voltage. Since
the electrical forces are so well distributed among hundreds of
dielectrics, he had plenty of breakdown safety margin. He gave
the unit a couple of days to rest after construction, topping it
up with oil as required, and gave her the works at 15 kv on a big
coil. The heavy buss wiring never even got warm, and even though
it bubbled out enough air to displace a few more pints of oil, it
did not break down.

It turns out that this is a homemade version of commercial pulse
discharging capacitors. Stacked capacitor sections of very high
value are placed in series until the proper voltage requirement
is met. The cap has a very high Q because all of the plates are
very close together, with a minimum of connections and bussing
required. They deliver a very sharp pulse discharge.

Bill's cap was pretty cramped in the bucket. Because of the
square shape of the bags, a rectangular tank would have made
things easier to fit and wire. But he ran his buss bars through
the side of the bucket (sealed with hot glue) and by snapping on
the lid, he could pick it up by the handle and move it around
with ease.

The novice coiler should think about the capacitor requirements
and experiment some before beginning large scale homemade caps.
Shop for materials; frequently a wholesaler can be found where
bulk products (like mineral oil in 5 gallon pails) can be
purchased for a fraction of the retail cost. But just because
you don't have some big bang pulse caps on line does not mean
that you should wait to begin firing a small coil. Nearly every
beginner gets hir feet wet in salt water capacitors.

Tesla used salt water tanks in Colorado Springs. A tribute to the
genius of the man was his ability to develop his huge peak
powers using low Q saltwater/glass caps. I do not recommend glass
as a dielectric for coiling work. The dielectric constant is much
better than plastic, but the RF dissipation factor is so great
that they can rupture from dielectric heating (even in salt water
the trapped water under the bottles does not circulate) and they
always give a spindly, violet colored spark. Polyethylene again
is the material of choice, and bottles and buckets can be
assembled in a couple of hours that will fire small stuff. I
mentioned he before that I have a friend who is firing 5 kVA
coils, and still using banks of salt water caps to keep his
investment down. As with any homemade capacitor, the salt water
must be covered in oil to suppress surface corona. But the
quality of oil need not be high, and the capacitors need not be
exceptionally clean. A saturated solution of rock salt is all
that is needed for the plates.

I think I have accomplished what I intended to say on this
subject. As always, I am happy to respond on any unclear areas,
the need for additional information, or to note corrections.

Richard Quick

.. If all else fails... Throw another megavolt across it!  
