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

Re: MMC

Hi Ted, Travis,

My comments interspersed.

> Original Poster: Travis Tabbal <bigboss-at-xmission-dot-com>
> The most important thing is that it be pulse rated. They
>are not really very common. Polypropylene pulse caps
>can be had from Digikey and others.

Hmm. I canīt quite agree here. Most PP caps are pulse
duty caps. Ted, what you are looking for are PP caps with
a high dv/dt rate. The higher, the better. Try to get caps
that use metal (and NOT metalized plastic foil) endplates.
The clue to these is the dv/dt value. Anything much above
2000V/ĩsec WILL have metal foil. These stand up much
better to the harsh TC environment.

>Look for values you can make a decent size MMC from.

How about (Ted) posting some specs of your coil. That
makes it easier to find proper caps. See below as to
what *proper* means.

>check the archives at www.pupman-dot-com. There is some
>great data there.

To save you some time, as the archive is gigantic, start
from October 1998, where the MMC was introduced.

>Most caps are rated in DC volts. No big deal. Many are
>using DC=AC for the rating.

No, any cap will be rated in DC and AC voltage. The problem
to what AC values you can use is the corona interception
value. This varies from manufacturer to manufacturer. Going
above a certain AC voltage (for a cap "x" from manufacturer
"y"), will lead to internal corona, a phenomenon which is called
partial discharges. This will lead to treeing that "eats your
dielectric alive". This has nothing to do with "overvoltage" in
the sense of internal flash-over, but rather from very tiny air
pockets, which are present in ANY kind of cap (commercial
or homebrew).

The DC=AC rating was my idea. This did get me a few frowns
in the beginning, that it might NOT work (but it does). However,
I am using caps that have a DCV of 1kV and an ACV of 350V,
so there is *just* a factor of 3 in there. I have another set of
caps from WIMA, which have a 6kvDC rating and *just* 700Vac.
I wouldnīt dream of running these at 6kVac!! Basiclly, Terryīs
EMMC idea uses the fact that commercial caps are overdesigned
and can be run successfully at a much higher voltage than the
stamped rating will let you believe. However, indisputably there
is a trade-off between DC to AC derating and lifetime. The nearer
you get to the DC rating, the lower the life expectancy is. On the
other hand you donīt need a TC cap that has a 10^6hr lifetime.
If you build a cap that has a 1000 hour life expectancy, you could
run your coil for 30 minutes EACH day for the next 5.5 years.
Nobody does that :o). However, also be aware that in calculating
lifetime, there is a power of 15(!) in the equation,which means
there simply IS a limit as to how far down you can go with the
number of seriesed caps in a chain.

Here are some more (important) areas to consider, however:

a.) The FRes of your coil. The lower the FRes, the less
     you stress the caps (more time available).

b.) The breakrate. High BPS stress the caps more.

c.) The peak and rms primary current. The higher the primary
     current, the more you stress the caps.

You will need to build the MMC as an SxP array of seriesed (S)
cap chains in parallel (P). Remember, seriesing IS needed to
divide up the total voltage applied (including worst case). Whereas
paralleling IS needed to divide up the primary current. Donīt try
using just one or two chains in parallel. Most TCīs will simply
exceed the peak current and your caps will die. For this reason
you must consider your TCīs specs (Donīt go for super large
values on the individual caps). Common values for the individual
caps range anywhere from 33nF to 68nF, truly depending on
total coil design. Try to find caps with high dv/dt values and (if
possible) try to find caps, where AC and DC Vrated arenīt super
far apart (like those dratted WIMA caps, I have)

Itīs kind of difficult to pack all the important info from a whole
year into a few sentences, so you will *have to* search the
archives for detailed info. If you post your TC specs (FRes,
bps, primary L, wanted primary C, HV xformer output V & mA),
I could try and suggest some starting values, tho ;o))


Coiler greets from Germany,
Reinhard