Re: Cap - Puncture Voltage

From: 	Malcolm Watts[SMTP:MALCOLM-at-directorate.wnp.ac.nz]
Sent: 	Wednesday, December 03, 1997 4:08 PM
To: 	Tesla List
Subject: 	Re: Cap - Puncture Voltage

Hi Bart,

> From:   Barton B. Anderson[SMTP:mopar-at-mn.uswest-dot-net]
> Sent:   Monday, December 01, 1997 9:04 PM
> To:     Tesla List
> Subject:    Cap - Puncture Voltage
> Hi everyone,
> I'm trying to design polyethylene caps and am running the numbers for
> both rolled and flat plate caps. Both have their good and bad
> characteristics/variables. In the course of running the numbers, I see
> it is necessary to rely on a "real" value for puncture voltage.
> I have seen info around that vary's polyethylene puncture voltage from
> 240/mil to 1200/mil. All I can say is "WOW", what a variance! I assume
> these to be DC ratings and therefore this would need to be derated for
> AC and especially Tesla Coil use. I have found a local supplier that has
> 30mil 4' x 8' sheets at $13/sheet. What is a "real" assumption I can use
> for puncture voltage? What do you guys use? I know need to over-engineer
> the cap, but going overboard can really screw up the numbers and my
> wallet. It's been stated that a .090 thickness (from a few thin layers
> or just one thick one) can withstand 15,000V. If this is true, then were
> talking 167V/mil as a puncture voltage. I would assume this has been
> derated for AC use? Anyone have a good guess at what I should use for a
> puncture voltage on a per mil basis?

I remember reading in High Power Electronics that the pros don't 
usually go beyond 270V/mil for polypropylene pulse caps. I think PE 
would be much the same dielectric strength-wise. This is for apps 
such as ours where high amplitude reversals are the norm. I 
understand that if the cap is impregnated with oil and there are no 
bubbles and all else is perfect, the next important thing is not to
impress more than about 5kV max across the oil. This may be where
single section HV caps are at risk of failure. Commercial pulse caps 
have a number of high capacitance matched sections in series to 
attain the voltage ratings. 
    The figure of 270V/mil has no bearing on what the caps can stand 
for DC. It is a stress related rating. I will be building a 25nF cap
very soon that will consist of four sections each measuring 100nF and 
each rated (and tested) for intermittent use at 15kVDC. I plan to use 
the composite at about 22kV peak in Tesla service. The dielectric 
thickness I am using is 0.25mm between plates (two thickness of 
0.125mm each). 
The caps turn out to be around a foot long by 3" in diameter each.