[Prev][Next][Index][Thread]
Arrgh! We NEED help
Hello all,
We (the GTL members) have a very hot discussion going on about
MMC´s at the moment. We want to use WIMA caps to build
various different MMCs.
Here are the cap specs:
Cap #1:
WIMA FKP1
33nF
dv/dt = 5000V/µsec
DC rating: 6kV
AC rating: 700V
Cap #2:
WIMA FKP1
100nF
dv/dt = 3500V/µsec
DC rating: 1.6kV
AC rating: 650V
Now, I say we should shoot for the 1.6kV caps and here is why:
Imax per Cap#1 = 0.033µF*5000V/µsec = 165A
Imax per Cap#2 = 0.1µF*3500V/µsec = 350A
(so, the way I see it, Cap #2 is better)
Next:
The difference in AC voltage rating is very low, being only 50V.
However what REALLY BUGS ME, is the discrepancy between
Uac & Udc in these two caps.
Cap #1: Udc->Uac: factor 8.6
Cap #2: Udc->Uac: factor 2.46
I have ABSOLUTELY NO IDEA why the 6kV cap can only withstand
700Vac. That is a very pitiful (de-) rating factor. I say Cap #2 is once
again better, because in TCs, we see large voltage reversals, so the
6kVDC rating is of no interest to us. Using the 6kVDC factor (in any
way) to build a MMC is playing Russian Roulette with the caps.
As I = c* dv/dt, I say it is better to use Cap #2. Here´s why:
Even though Cap #1has a higher dv/dt, it is not as suitable as Cap #2,
because the dv/dt(cap1) to dv/dt(cap2) is 1.42, but Cap# 1 only
has 1/3 of the capacitance of Cap #2.
I say: The dv/dt of the single cap is THE rating for the whole string.
What I mean is, if I use a cap with 3500V/µsec, it doesn´t make a
difference in "Ampacity" if I have 1 cap or 100000 caps in the
string. The maximum current allowed will flow through all caps, so
it doesn´t make a difference (to the dv/dt of the total string)
if I have one cap or more in the SERIES string. To keep the
dv/dt rating per string within the allowable, you have to use
many PARALLED strings.
I say: Imax (per total MMC) = (dv/dt*C)* number of PARALLEL strings.
Imax (per chain)= (C/n)* (dv/dt(sc)*n)
n: number of caps
sc: single cap
Let´s use Cap #2 as an example:
-------------------------------------------------
One Cap:
dv/dt=3500V/µsec
C= 0.1µF
Imax = 3500V/µsec * 0.1µF = 350A max for one cap
Imaginary Cap string:
same data as above, except we make a string
consisting of 1953 caps:
String capacitance is now = 0.1µF/1953 = 5.12*10^-5µF
dv/dt for the string is now 3500V/µsec*1953 = 6835.5Kv/µsec
so this cap is good for:
Imax= C(string)*dv/dt(string)= 5.12*10^-5µF*6835.5kV/µsec = 165A
The Imax per chain is totally independant of the number of caps
that are seriesed in the chain.
The cost of both caps (each) is approx. the same. I (as you can tell)
say go for the 1600V units. Some guys don´t agree, so I would like
to hear some opinions from you guys.
BTW: I think I now pretty much know why my Phillips X2 MMC died.
(it failed within 5 sec):
The caps used were:
Series 336/2 X2 MKP Caps
330nF
250Vac rated
200v/µsec rated
I built a test MMC with these caps. I used 35 caps in series in a single
string. Peak primary current was around 150A. Total capacitance
was 9.47nF. The input voltage was 7500V*sqrt(2)= 10.6kV peak.
Coil data: 200khZ, so following T=1/f, this means 5*10^-6sec.
Of interest is 1/4, so this means the rise time is 1.25µsec.
10606V* pi/2 = 16659V per 1.25µsec
dv/dt = 13327V/µsec/35 = 380V/µsec per Cap
Imax= 0.33µF*200V= 66A.
Now it becomes very clear why this cap (had to) fail(ed):
I exceed the maximum allowable current by a factor of ~2.3
and I exceeded the dv/dt by a factor of: 1.9.
My ERO MMC is still chugging along fine because even
though I run them at Peak ACV = rated DCV, I´m not
exceeding Imax or dv/dt ratings.
So, which WIMA caps should we go for?
a.) 33nf-at-6kvdc/700Vac
b.) 100nF-at-1600Vdc/650Vac
Coiler greets from germany,
Reinhard