[Prev][Next][Index][Thread]

Re: Unusual capacitor dielectrics

Jacob and all,

Pure water is an excellent dielectric... sort-of! Unfortunately, since
water's also an excellent solvent, it leaches out ions and organic
contaminants from vitually anything it comes in contact with, so that
even if you started with pure water, it won't stay that way for long! In
order to keep it pure, you'll need to continuously de-gas, filter, and
dionize it. Also, the breakdown strength of water declines with time -
it's only good for short pulse widths (microseconds or less). Subject to
these limitations, water makes for an excellent high voltage dielectric
for pulse work as you've surmised. 

Water is often THE dielectric of choice for making low-impedance
pulsed-power transmission lines for e-beam, fusion, and high power laser
work, since it self-heals after a breakdown, and can safely absorb large
amounts of "left over" power after the high power pulse has come and
gone. It has been used at E-fields in excess of 100-150 kV/cm for short
(usec) pulses, and considerably higher (>500 kV/cm) levels for
nanosecond-width pulses. Unfortunately, the longer high voltage stress
is applied to the water, the greater the probability that it will break
down. Breakdown voltage also declines in the presence of small air
bubbles at the water/electrode surface, or wgen the water contains
contaminants such as ions, organics, or small particles (even
bacteria!). 

Using very pure water, self-breakdown will typically occur within ten's
of microseconds depending upon electrode materials and water purity.
Interestingly enough, as in air, breakdown streamers are more easily
initiated from positive electrodes in water (and oil, for that matter).
A spectacular example of water-dielectric transmission lines, and "post
pulse" discharges, can be seen in the January, 1998 issue of National
Geographic or on Sandia's Z Accelerator site:
http://www.sandia.gov/pulspowr/facilities/zaccelerator.html

The problem of decreasing breakdown voltage with pulse width complicates
recharging a water cap for a Tesla Coil, since you'd need to quickly
recharge and discharge it within 10's of microseconds to prevent
self-breakdown of the water. A typical NST or pig-driven charging
circuit would not do the trick! Even if you overcame this challenge,
water, like many other polar dielectrics, has a relatively high
dielectric loss. At 100 kHz, for example, pure water has a dissipation
factor of 190 e-3. That's about 1.5X lossier than PVC, and > 2000X
lossier than LDPE! So even if you were able to overcome the difficulties
and costs associated with auxilliary deionizing and filtering equipment,
and special charging circuitry, the water-dielectric capacitor you'd end
up with would actually be a "dog" (performance-wise) compared to an LDPE
cap.

The breakdown voltage of barium titanate (BsTiO3) is a function of
formulation, grain size, blend, fabrication, and thickness. Typically an
organic binder or glass is blended with it prior to pressing and firing.
Glass bonded barium titanate in thin layers (as in multi-layer ceramic
caps) can have a dielectric strength approaching 300 kV/cm. As with
other dielectrics, dielectric strength (volts/thickness) is considerably
reduced as you increase thickness. For example, a typical 20 kVDC
ceramic cap typically requires a 1 cm thick dielectric, for a rated
stress of around 20 kV/cm. Being a ceramic, BaTiO3 is brittle.
Dielectric loss is will typically be in the range of 5-10 e-3, which
makes BaTiO3 about 100X as lossy as LDPE. 

Now, if you already have access to the appropriate raw materials, an   
organic binder, silver paste, a hydraulic hot press, and sintering and
annealing oven(s), you certainly could try "rolling your own" high
voltage ceramic caps. Annealing is critical to prevent internal cracking
of the material, so a blowtorch probably won't do. Because of these
challenges, I've not heard of any amateur coilers going this route... as
yet. Keep us posted if you plan to pioneer this area!  :^)

And, safe cappin' to you!

-- Bert --

Some sources for further information: 
1. Vitkovitsky, Ihor, "High Power Switching", Van Nostrand Reinhold,
1987, 304pp, ISBN 0-442-29067-5 

2. Burfoot,  J. C., Taylor, G. W., "Polar Dielectrics and Their
Applications", U of CA Press, 1979, 465pp, ISBN 0-520-03749-9

3. Von Hippel, A. R., "Dielectric Materials and Applications", MIT
Press, 1954


Jacob Roberto wrote:
> 
> I've noticed that pure water has a dielectric constant of 80 (about 40 times
> higher than that of polyethylene). Would it be practical to build capacitors
> that use water as the dielectric? Such a capacitor could probably be
built by
> taking pieces of sheet metal, drilling holes in their corners, and then
bolting
> them together, using nylon washers or other flat insulators to separate the
> plates.
> 
> Guessing that water has a breakdown voltage of 250 volts/mil, A 0.0864 uF
7.5 kV
> capacitor could be built by simply taking two 12x12 inch pieces of sheet
metal
> and using 30 mil seperators. Stacking 9 such plates (making 8 capacitors in
> series) would yield 0.0108 uF at 60 kVDC, suitable for small-to-medium Tesla
> coils. 

<SNIP>
> I don't have any idea what the RF loses of water are; they might prohibit
its
> use in Tesla coils. However, it might still be useful for pulse discharge
> applications.
> 
> Also, what about barium titanate capacitors? Barium tatanate has a *huge*
1200
> dielectric constant. It isn't exactly easy to find, but "barium titanate,
> powder, < 3 micron, 99%" can be purchased from Aldrich for 22.80 for 500g or
> 66.90 for 2kg.
<SNIP>
> 
> Does anyone know the puncture voltage and RF losses of pure water and barium
> titanate???
> 
> --
>         -- Jacob Roberto
> "Some people dream about being famous and having fancy cars. I dream about
> 1000KVA transformers and 1000KJ pulse caps!"