Re: Current peak from MMC
If you look at the MKP 375 characteristics, don't forget to have a look to the
a MKP 375 rated for 2 kV has got 30000 V/us dV/dT but only in models up to
pF (yes, pF!). Models up to 0.01 uF have got dV/dT = 10000 V/us and models
0.022 uF drop down to 6700 V/us.
The MMKP 376 (which is MY CHOICE) rated for 2kV is available with all
to 0.033 uF and for ALL values Philips ensures a dV/dT > 10000 V/us.
So the choice is a compromise between the rated voltage, desired capacitance,
space available and pocket size. For instance a 0.022 uF 2 kV MMKP 376 wins the
same MKP 375 for offering a higher dV/dT.
By the way, the formula I sent in my last mail for calculating the substained
current is from Philips and contained in their capacitor Databook.
Another thing to try is the BHC Aerovox capacitors that Farnell sells with code
number 783-110, 783-122, etc. They are "designed to protect IGBT modules
large voltage transients", metalized PP, df 0.1% -at- 1kHz.
For instance the 0.47 uF 1600V is rated 850 V/us 400A peak , 17 A Irms. They
cost each about $12 (Farnell) and a "big": that 0.47 uF 1600V is about 4 x
3 x 4
cm and big terminals.
I purchased myself 5 of them to build a 0.1 uF 5 kV DC capacitor for testing my
capacitor charger switcher. I don't know, will I ever test them as a primary TC
Tesla List <tesla-at-pupman-dot-com> on 12.04.99 00:06:04
cc: (bcc: Marco Denicolai/MARTIS)
Subject: Current peak from MMC
Original Poster: Finn Hammer <f-hammer-at-post5.tele.dk>
One thing that kept me away from considering a MMC to start with, was an
(erroneus) gut feeling, that these small caps would be unable to deliver
enough current. So I tried to plug some data into Excell, and I thought
you might be interested in the results.
The cap of my choise is the Philips KP/MKP 375 series, which is a
polyprop. cap for high pulse applic. like deflection circuits in CRT
monitors. There are datasheets at www.philips-dot-com. and that is where I
got the Dv/Dt values.
Assuming a 0,025 µF cap at 30 KV DC, and using the formula that Marco
posted recently, I came to these results:
(The first row is marco`s example, I included it to check if the formula
was entered correctly.)
Urdc µF dV/dT amp./string Amp. Tot.
0,001 10000 10
1000 0,022 5000 110 3750
This doesn`t seem so bad, and since a cap like this can be built for
something like 300 $, that is the route that I will follow.