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

Re: Possible cheap MMC capacitors



Hi Bryan, all,

Warning: lengthy post ahead.

> Original Poster: "Bryan Steinbach" <warp27-at-juno-dot-com>
> Looking through the Newark catalog (http://www.newark-dot-com) I
>found the 940C product line.  These are self healing polypropylene
>capacitors,and they say "high peak current capability".  What
>caught my eye is the 3kVDC .1uf for $2.40, Newark part number
>48F3953 and CDE part 940C30P1K.  At 2.40 they seem ridicuosly
>cheap, does anybody know anything about these and their
>suitability in a MMC?

What dv/dt rating do they have? What is their AC voltage rating?

Dv/dt tells you the maximum current they are able to process. It
doesn´t make a difference if you use 1 or 1000 in a string. The
amps per string will still be the same. This isn´t easy to see at
first glance, but if you use one in a string, it is simply dv/dt*C.
If you use 1000 per string the equation looks a little different:
Of course, the dv/dt for the whole string is 1000 * (Csingle dv/dt),
BUT the capacitance of the whole string is also only 1/1000* C.
Current handling capabilities would be: 1000*(Csingle dv/dt)* 1/
1000*C. Rearranging that a bit, we get 1000/1000*(Csingle dv/dt)
*C. 1000/1000 is one, so the amperage of the string is still (Csingle
dv/dt)*C(single).

Paralleling strings is what gives you high pulse current capability.
Since 0.1µF is a large cap, you won´t need many strings to
achieve your needed capacitance. As a general rule of thumb: as
the capacitance of a production line (cap series type x) goes up,
the dv/dt goes down. To make it clear, what I mean let us look at
three examples of MMC construction. We want to achieve 50 nF
for our MMC example and let us say our xformer puts out 15kVrms.

Cap #1 (all values are for CSingle)
WIMA
1µF
650Vac
1600Vdc
Dv/dt: 250V/µsec
Max amps per cap = 1µF*250V/µsec or 250A

Cap #2 (all values are for CSingle)
WIMA
33nF
700Vac
6000Vdc
Dv/dt: 5000V/µsec
Max amps per cap = 0.033µF*5000V/µsec or 165A

Cap #3 (all values are for CSingle)
WIMA
68nF
650Vac
1600Vdc
Dv/dt: 3900V/µsec
Max amps per cap = 0.068µF*3900V/µsec or 265.2A

We said we wanted 50nF as the value for our MMC and our
power source is rated at 15kVrms, so:

50nF MMC from Cap #1:
20 caps per string or 50nF per string, so we need 1 string for 50nF.
Total number of caps: 20

50nF MMC from Cap #2:
10 caps per string or 3.3nF per string, so we need 15 strings for 50nF.
Total number of caps: 150

50nF MMC from Cap #3:
18 caps per string or 3.77nF per string, so we need 13 strings for 50nF.
Total number of caps: 234

Looking at these values, one would still say MMC array number
one is the winner. So now let us look at the rest of the MMC data:

50nF MMC from Cap #1:
Vac rated: 20* 650 or 13kVac
Vdc rated: 20* 1600 or 32kVdc
Max amps per MMC: 250A*1 = 250A ;o(((!!

50nF MMC from Cap #2:
Vac rated: 10* 700 or 7kVac ;o(((!
Vdc rated: 10* 6000 or 60kVdc
Max amps per MMC: 165*15 = 2475A

50nF MMC from Cap #3:
Vac rated: 18* 650 or 11.7kVac
Vdc rated: 18* 1600 or 28.8kVdc
Max amps per MMC: 265.5A*13 = 3451.5A

Now we see that the 1µF units are not up to the job, because
they only can take 250A primary current. MMC #2 is a bit
better, but the ratio DCV to ACV is outrageous. Using this MMC
with a 15kV xformer is pushing it quite a bit on the AC rating
(lifetime) and 60kVdc is downright crazy. The caps are factory
tested (and must survive) at 2x VDC rated, which means this
cap will withstand a 120kV peak surge. Your 15kV xformer will
long have arced over before you reach this value. MMC #3 is
best suited for coiling usage. It´s AC rating is near that of our
xformer (15kV rms). The DC rating is pretty good (it will be
able to withstand at LEAST 57kVdc non repetitive) and the
peak current, which the array can withstand, makes it literally
"bullet proof".

While 250A peak primary current SEEMS like a lot, you must also
remember that a 15kVrms xformer that needs a 50nF cap (be it an
LTR or non-LTR coil) IS a big xformer. With 15kVrms and 50nF,
you will be storing 11.25 Joules per Bang. The large Cpri requires a
low Lpri for a certain FRes. Ipeak can be calculated as Ipeak=V*
SQRT(Cpri/Lpri). As you can see, the current in a large Cpri and low
Lpri circuit becomes gigantic! However, for MMC construction, pretty
much the same goes for lower power units (Ipeak being lower). You
will never want to exceed dv/dt. Your caps just won´t like it ;o)) Now
you can also see why a limiting device is a MUST in the safety gap
across the cap. Look at the above equation and now picture L going
towards zero (with no limiting device within the cap SSG). The surge
current goes up into enormous values. Any coiler, who has a SSG
across his cap (and no limiting device installed) will tell you that his
gap fires very LOUD. Not good for your caps ;o(((. I would go for
0.5x-1.5x the primary coil´s impedance for the limiting resistor or
inductance depending on (E)MMC design and Lpri value.

Look for MKP caps with a high dv/dt. If possible try to keep Vac (rated)
as close to Vdc (rated) as possible. In other words don´t look for a
MKP cap with a super high DCV rating, if the ACV rating is very low (as
in our example 6kVdc vs. only 700Vac). If you strive towards building an
EMMC, all the above holds even more true. You are using less caps in
series, which makes the C per string larger, which once again means
you need a much lower number of strings, which then greatly reduces
the Imax, which the EMMC can safely handle.

Some building and cap searching tips:

- You will NEVER want to exceed dv/dt.

- Picture Vac as your "fitting" voltage. You can exceed this, but also
  remember, the lifetime goes down by a power of 15 (!)

- Picture 2xVdc as your one shot failure mode. You want this high enough
  to prevent cap death in case of a strike going the wrong way. However,
  there is NO need to go super high on this value. All manufacturers,
  that I know of, test their caps at 2x Vdc and they MUST survive this.
  Terry did a bit of destructive cap testing a while ago and he found
  out, that his Panasonic caps die at 3 or 4 times the rated voltage. I
  think the other manufacturers out there will have similar values. So
  in the case of our MMC #2, this would be ~180-240kV (for a single
  shot failure mode), which is totally ridiculous and in my favorite
  case (MMC #3), it would still be 86-115kV. As you have safety
  gaps across the xformer and the MMC, this is ample safety in
  my mind.

If I had to chose a selection criteria, the above comments are
in the order from most to least important. Weighting the selection
criteria above, I would say 40-45% for the dv/dt rating, 35-40%
for the Vac rating and only about 15-25% for the Vdc rating.

There are several MMC calculators out there. All can be found
on and around the site http://users.better-dot-org/

1.) Terry´s: http://users.better-dot-org/tfritz/MMCCALC2.ZIP
2.) Stan´s Online Calc: http://tesla.better-dot-org/mmc_form.cfm
3.) Mine: http://users.better-dot-org/rbuchner/download/mmc.zip

Download one of them (or all ;o}) and play around with the different
cap values to get a feel for things.

Hope it was of some help.

Coiler greets from Germany,
Reinhard