[Prev][Next][Index][Thread]
Re: AC-rating for MMC caps // EMMC vs the GTL-WIMAs
Hi Terry, Coilers,
Terry> I quote from WIMAs data sheets "Furthermore the r.m.s.
voltage derived from the peak voltage shall not be
greater than the normal AC voltage rating of the capacitor
to avoid the ionization inception level: V r.m.s. <<V a.c.
rated."
It's anybody's guess as to what that was supposed to mean...
I guess with 'V a. c.' they mean only sinusodial waveforms. So
I think it means that you should not exceed the rms value
regardless of the waveform. With other words, you could go beyond
1.4 times the rated VAC with your peak voltage if your waveform
is not sinusodial. This applies for example for a 'bipolar
rectangular' signal (+U,0,-U,0,+U,0,-U...) with t_off=3*t_on.
But their formula is not a very reliable thing because it would
allow a far to high voltage for t_off=100*t_on :-(
--------------------
Terry> We use WIMAs at the DC/1.4 level all the time. As long as
they do not get hot they last forever over (15 years of using
them). WIMA would probably just base their answers off our
data anyway. I bet somebody wrote some paper warning of
doom once, or something like that, which caused this concern.
No one I have talked to sees any problem with using them
at the DC peak levels as long as the ESR heating does not
raise the body temperature above 5 degrees C. I doubt WIMA
could explain "ionization inception level". I'll try to
check around for a definition, guidelines, equations, and
data... However, I don't expect a darn thing....
Bert> "Ionization or Corona Inception Level" is an accepted term
in capacitor design, and it refers to the level at which
ionization can begin to occur inside a bubble of entrapped
air or within an air-filled void within the solid dielectric
system. If one had "perfect" dielectrics and could always
exclude any entrapped air, derating for this phenomenon
would not be necessary.
Capacitor vendors will typically estimate the Inception Level
by using Paschen's Law, the anticipated maximum void size
(in the direction of the E-field), the dielectric constant(s)
of the dielectric system, and the thickness of the dielectric
between plates. Basically, this is the threshold at which
ionization within an entrapped void COULD begin during AC or
pulsed stress. Most capacitor vendors multiply this by 2 to
estimate the applied stress level at which progressive damage
WOULD occur due to partial discharges within the dielectric
itself. It's basically a design parameter for the capacitor,
and it permits the manufacturer to estimate how much margin
exists in the presence of manufacturing process or material
defects.
The actual stress at which ionization will occur (IF it occurs)
for a given capacitor is extremely variable, and is very
sensitive to manufacturing process control and incoming material
quality control.
Ultimately, vendors do "torture" testing on large groups of
capacitors out of different manufacturing lots to determine the
actual performance of their products, analyze the statistics,
and then do lifetime projections. Damage due to partial
discharges tends to be gradual, and the smaller the defect size,
the longer it takes to do damage. And, if the vendor maintains
tight materials and processs control, the batch of caps we use
my not even exhibit this failure mode!
Sooo..... as long as we do not exceed the breakdown voltage of
the dielectric system(s), and do not induce damage from
excessively dielectric heating (does not appear to be a problem
for PP), we may see a bit of long-term degradation and reduced
life if we exceed the manufacturers' AC ratings/ionization
inception level. However, the shortened life may not actually
cause any problems for the short run-times and low duty cycles
seen by Tesla Coil application.
Well, today I got a phone call from WIMA and we discussed the things.
The guy said that there ARE partial discharges occuring above a
certain voltage level. This phenomenon takes place ONLY at AC.
Above a certain VERY low frequency (about 1Hz, he guessed), it is
INDEPENDENT of frequency. So we have the same problem at 50/60Hz
as well as in the 100kHz range.
This Corona Inception Level depends on the internal geometry of
the cap. With one floating electrode (= series connection of two
plates like the FKP1-type from WIMA), it is 700VAC. With an internal
series connection of 4 plates, it only raises up to 1200VAC.
I told him that we want the caps to last for 2000 hours and asked
him how hard we could drive them above the rated 700VAC. At first
he said factor '1.5 above', later he only suggested factor 1.25
or with other word 900VAC. BUT: he couldn't tell me how much
of the caps will die in this time (50% or 1% or whatever...). Of
course I nailed him down to tell me that important information in
the next days. I think the number of dead caps will be quite low
as they count failures per 10^9 cap hours. So I think we can go
beyond these 125% AC stress.
Due to the low duty cycle we typically have (<5%), we don't have to
derate the caps HV wise. Only in high breakrate systems we have to
check the temperature (as Terry said).
The max. amperage (pulse risetime) is not as critical as the
Corona Inception Level is, he told me.
----------------------------
Terry> We use WIMAs at the DC/1.4 level all the time. As long as
they do not get hot they last forever over (15 years of using
them)....
How high is the factor Udc/Uac for the caps you mentioned? There
are types where it is pretty low (1000Vdc/600Vac) and where this
should be no problem as the peak voltage (850Vpeak) is nearly the
rated dc voltage (1000Vdc). The caps the GSTCs ('German Speaking
Tesla Coilers) ordered are 6kVdc/700Vac and therefore there is a BIG
difference!
-----------------------------
STK> BTW, did I understand right, that your 'EMMC' is just a new
type of cap used in a configuration where it is stressed
by using 'real peak AC = rated DC'?
Terry> Don't get to excited about my silly acronyms :-))
No, don't worry, I'm used to read Reinhards mails both on the
pupman as on the GTL - looks like you two have gone through the
same school ;-))) (Hi Reinhard, congratulation to your sparklength!)
On our GTL-website, we have a dictionary 'coiler slang <-> German'
which becomes a lookup table for abreviations more and more :)
[:) is the shortcut for :-) <G>]
Terry> I was just jokingly trying to name caps that were used beyond
their manufacturer's rating (with a good degree of
understanding). Manufacturer's rate their caps so anyone
will be able to use the caps within that "rating" without
a problem. The parts can often be pushed much harder.
Terry> My "EEMMC" cap was meant to take the rating to very near
destructive levels. It is right on the edge where my best
guess said it "probably" would survive
What about using the WIMAs in an XEEMMC-arrangement with just
two 6kVDC caps (33nF) in series on an 8kV-neon? This would meet
your 'peak voltage < DC rating'-requirement (peak 11.3kV for
2x 6kVDCr) and will result in an 16.5nF cap for only 1.70$US ;-)
I think I have to perform this test just for fun. The 700VAC
rating compared to their 6kVDC rating is very strange and if
the caps withstand the test for any reasonable time, those
caps seem to be the ultimate TC-caps. I hope they'll do....
(we should get them in KW26)
Terry> And you now see why the manufacturers have so much darn
trouble rating capacitors :-))) It just isn't easy....
Take your best guess as they do....
Yes, that seems to be true. The guy had my data for three weeks
and still couldn't say how high the risk due to the partial
discharges will be. Hope he gives me some more concrete nuumbers
in the next days when he looked over their statistics.
Stefan
\_ _ _/\_________________________________
\_______ _/ `--' |
(_(_|_)_)< ` Reply to: STK.TC-at-gmx-dot-net, please |
_ __/ |_| \_ visit "Stefans Tesla coil pages" |
\_/ ` |_| ` with lots of photographs, formulas |
_/ .__|_|__. and technical infos around TCs at |
/_______\-|-|-/___ http://privat.schlund.de/skluge/toc.htm _|
Famous last words: Oh, a Tesla coil - I heard you can touch
it without getting hurt --> Zzzzzapp ...