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

Re: Another MMC Failure



Original poster: Vardan <vardan01@xxxxxxxxxxxxxxxxxxxxxxx>

Hi Finn,

See this,

http://hot-streamer.com/temp/CD942C20P15K-MMC-Capacitor-Lifetest.pdf

Of course, we don't know the "function", only the limits. At 500V (630 V ion inception voltage- dual series) the plates have no ionization. They moved down to 500V in the ratings since the voltage gradient is higher at the plate edges. But at 500V/cap the life should be "forever". At 2000V they show a decay step at 75 hours. I never tried any other voltages on the tester.

I can't imagine current had anything do do with the failure here. What you describe is voltage failure in many of the caps so it is just not one bad cap but a real problem going on. The unblown areas of the dielectric film are probably white from ion bombardment chewing away at the dielectric.

IMHO - All those life equations that have a "^xx" term in them like V^19 or I^3.54598 are howg wash... I am sure they worked once for the guy making the equation, but I have never seen anyone "repeat" the results. If they try, they just end up making a new "different" equation...

I see the data sheet at:

http://cde.com/catalogs/942C.pdf

Has a "plethora" of AC voltage ratings... They are either 315 or 630 for polypropylene... So I don;t know if they are fudging 630V caps down or 315V caps up. But you answered that when you said the caps have an "intermediate" electrode. So they are dual series types for 630V.

What we need to know is the "AC voltage on time" It does not matter if the AC is 70kHz or 50/60 Hz. Anytime there is above say 30Hz on the caps, the clock is ticking... I wish I had run the above test for a few more data points to predict the "curve"... Since you don't need current, you might consider the 940C20W1K types.

I think DC COX is running at 1/3 the DC peak voltage on advice from experts he knows.

Cheers,

        Terry

At 02:41 PM 7/15/2006, you wrote:
All.

I just had a MMC fail on an OLTC coil that I built a year ago.
The coil resonates at 70kHz , primary cap is 16.2µF and the pri-L is 350µH
With a firing voltage of 1150V this translates into a primary current close to 8000A The primary cap is made up of 18pcs. 942C12P9K so each of them sees 450 peak amps. They are rated at 861Apeak, so it is no wonder that the cap failiure is consistent with what we have come used to see from an overvolted cap.

I am sort of surprised they gave in to over voltage......

The caps show multible punctures of the dielectric that leads to the evaporation deposited layer of the intermediate electrode, with signs of self healing. In a couple of the caps, the intermediate electrode has almost completely lost all of it`s conductive material, so they started to turn resistive, and heated up, although not seriously so.

The coil is in adventure park service, and the duty cycle is low. every 4.5minutes, it puts out 112 bursts of 124 discharges at 500BPS.

This killed the cap in a bit more than a month.

From
http://home.earthlink.net/~jimlux/hv/caplife.htm
I can gather that if I double the voltage rating of the cap, the expected lifetime will increase 180fold and that suits me fine, because I can then expect the caps to last 15 years.

What I wonder is this, should I put 942C12P9K caps in 36 strings with 2 in each string, or should I rather use 942C20P15K and put 108 in parallel.

A Maxwell cap is another option, of course.

Cheers, Finn Hammer