# Re: Cap - Puncture Voltage

```
From: 	Barton B. Anderson[SMTP:mopar-at-mn.uswest-dot-net]
Sent: 	Sunday, December 07, 1997 12:41 PM
To: 	Tesla List
Subject: 	Re: Cap - Puncture Voltage

Tesla List wrote:

> Sent:   Saturday, December 06, 1997 10:22 AM
> To:     Tesla List
> Subject:        Re: Cap - Puncture Voltage

> <SNIP>
> I did my own little test, by taking a 30mil piece of LDPE sheet 4" by 4"
> and putting a 2" x 2" aluminum foil square on each side.  I then attached
> this to my 15kV 30mA Transformer (rebuilt and designated test
> transformer), and started cranking up my variac.  The PE punctured at a
> variac setting of 100V exactly.  This works out to be approx 12500 VRMS
> on the capacitor, and 17.7kV peak.   That's 590 V/mil, albeit AC peak,
> not DC.  Strong stuff.  I am using 90 mil in my reconstructed capacitor
> (3 x 30mil) and I am running at 9kV w/resonant charging.  I expect this
> to be fairly stable.
>

Since I started this post, I've come to an understanding that there are a lot
of unknowns as probably the best information comes from the real life
experience of others. Even here one wonders how precise construction
techniques are compared to your own and other variables which would change a
caps stability of stresses.

>From your .090 diel. and 9,000v output, try looking at your (Pv)puncture
voltage as:
Pv =1.414 * 9,000 * 2 = 25,452V needed
which equates to:    25,452 / .090 / 1000 = 282.8V across each mil of
dielectric.
Decrease the dielectric thickness and your potential across each mil of
dielectric increases.
This of course is an average across the dielectric.

Now do you feel comfortable with 282.8v/mil as a peak potential across each
mil? Maybe so. Based on other coilers who have managed 12kv from their caps on
.090, it might very well be a "comfortable build". But then, maybe the coilers
who managed this have done so because of their construction techniques to not
allow defects in materials, materials chosen, cleanliness, and any contruction
flaws including assembly, layered dielectrics, etc..

If you can manage the above, I would think your choice of .090 should be quite
stable.
Bart

```