MMC caps in CW use

[Terry Fritz <twftesla-at-uswest-dot-net>]

Hi All,

        I have not quite been following this CW coil MMC thread closely but
maybe I can help out.

        You can easily run MMCs at the full rated "DC" voltage in CW use.
There will be ionization inside the dielectrics which is 'not good', but the
damage will probably not become serious in the limited time the MMC would be
run.  You may not want to push the voltage higher than the DC rating in the
CW case.

        The skinny little leads "may" tend to get hot but they have good
cooling sitting out in the air and all.  In my MMC current testing at 350kHz
CW, lead heating was not an issue at all.  The dV/dT ratings are typically
far above what we would ever be concerned with, so I just ignore that.  Only
if one fully charged and then directly shorted the caps would you run the
risk of burning out the lead connections in pulse rated poly caps.

        Most polypropylene high current caps in the 50nF range can handle
about 4 amps RMS easily.  However, you really need to do a bit of
calculation since the cap value, size, and the particular frequency can be a
problem.  Polypropylene does not like really high frequencies in that the
dielectric starts to get really lossy quickly as the frequency is increased.

So lets say I want a cap with the following specs:

16kV Peak (11.3kV RMS)
20 amps RMS
400 kHz
20nF

Suppose I have a bunch of WIMA 47nF 1600V FKP-1 caps...  I am doing this as
I type so I hope this example works out ;-)

1.      First find the impedance (Xc) of the single cap at 400kHz.

Xc = 1 / (2 x pi x f x C)

In this case, that works out to 8.466 ohms.

2.      Find the current a single cap can handle.  Simply go to WIMA's nice
site and look for the voltage vs. frequency graphs.  The last link is the
one you want.  In general, these graphs are good for other manufacturers
caps of the 'same size' and ratings:

http://www.wima-dot-com/
http://www.wima-dot-com/navig/fkp1.htm
http://www.wima-dot-com/fkp1ue.htm  

So looking up the 47nF 1600V caps at http://www.wima-dot-com/fkp1ue.htm at
400kHz gives a voltage of ~50 volts RMS (note the log scale).  This is the
voltage that will heat the caps 10C and is considered "safe".  Don't push
these ratings since WIMA "really means" them.

3.      To find the current, simply divide the 50 volts by 8.466 ohms to get
5.9 amps RMS per cap at 400kHz.  Note that these are pretty big caps...

4.      We want to have a 20 amp MMC so we can use 4 strings of these.

5.      Now we need to worry about the voltage.  We need at least 10 caps
per string to get to 16000 volts.  So each string will have a capacitance of
47nF / 10 = 4.7nF.  Four strings gives 18.8nF which may be close enough to
the 20nF we wanted.  A 5 row by 12cap/row MMC would have a closer value,
higher voltage, and higher current but would use 60 caps instead of 40...
Such decisions are left to the reader ;-)

If your "feeling lucky", 3 rows by 7 is a possibility with only 21 caps in
the EMMC range ;-)

Of course, a larger number of some smaller (cheaper) caps can work just as
well so there are many possibilities to get the best price and such.  But
these are the math mechanics behind it all.


Cheers,

        Terry