Original poster: "D.C. Cox" <resonance@xxxxxxxxxx>
Dr. Resonance
Hi Dmitry and DC,
There are many issues here that should be explained. And yes, it gets messy and complicated ;-))
1st, Many many MMC capacitors are running out there just fine and not blowing up. We continually survey and "ask around" if anyone is having odd problems and none are to be found... Terry has always asked that odd MMC failures be directed to "me" so we can get right on top of any problems and fix them... Most worry more about the 1/2 watts resistors than the caps (they are doing fine too ;-)) Most folks just do not run coils hard enough or long enough for the effects of dielectric ion damage to matter at all. But in DC's case, he makes coils for museums that may run every day for, say, the next 500 years!! Unless DRSSTCs make them obsolete in the next "few years" >:o)), DC's coil's will have to run longer and harder by many orders of magnitude than the rest of our coils. Thus, he has to be very conservative in how he uses the very new MMC technologies.
2nd, DC is just saying what his very knowledgeable friend has "told him", so don't blame "DC" ;-)) We love to have input and hints into this matter and any information is precious to us!!
Onward, Ion inception is where the plates in a poly cap start to "glow blue" with free electrons. That "point" occurs at 320 volts. Our caps are actually two in series, since they have a floating middle plate, so their voltage is doubled to 640 volts.
Free ions tear away at the very long polypropylene molecules and break them up. Eventually, the normally water clear pure 0.0005 inch layer of poly gets cloudy and takes on a rainbow sheen... That actually happens fairly quickly and we see that in periodic destructive inspections... Over a much much longer time, there will develop weak areas in the poly. Oddly though, these weak areas have a lower dielectric constant and then to appear more resistive to AC currents. In effect this shields them from the current and failure... In a very real way, the caps "get stronger" over time since the high ionization areas tend to conduct less current (don't tell nobody you heard that 'secret' from 'me' :o))). Recently, additives have been added to resist the "UV" part of this process, and you did not here that from "me" either.... They are always fiddling with the dielectric recipes and they are making significant improvements as we speak... The MMC caps I buy today are smaller and much better rated than those I bought five years ago by a very significant degree!!!
There is not enough "power" from ion inception "in our case" to "melt" anything... So I would directly disagree with DC's friend there...
Over a very long time though... Those high ion bombardment stress dielectric areas start to breakdown and we have failures. At this point, the cap can recover easily, but the arc overs start to happen at a very high rate. These failures start can be extreme with literally "millions" (!!) of failure sites... The cap may eventually fail "OPEN" in this way... But when arc failures increase dramatically, the cap is near death... I know "much" about "Machine Gun" damaged poly dielectrics %:-)))
"Our" worry is "when". "WHEN" will the caps fail... Today, tomorrow, next week, 100 years, 10,000 years.... If the cap is operated at 640 VAC it will last probably 250,000 hours on average. Then, the failures will probably be more due to "old age" and polymer decay due to age rather than electrical effects. We have all seen "old" poly caps that have turned yellow. Not due to electrical effects but simply due to the fact the polymers breakdown over time... There is an issue with mechanical vibration too at AC frequencies that are low enough to where the plates in the cap "move around"... There are contamination issues from the solder fluxes, built in contaminants, and external junk... So, our poly caps ARE going to fail in say maybe 200 years simply do to their "organic" nature... Of course, "'we' don't care" ;-))
So "how" do you make a MMC last "long enough"...
A. Follow Terry's "essential tips" at:
http://hot-streamer.com/TeslaCoils/MMCInfo/mmcinfo.htm
Even if the cap themselves are wonderful, if you put it together all wrong, it WILL fry...
B. Do NOT exceed the peak rated current!! Even though the specks are "padded" 200%. Too high of peak currents quickly (minutes) destroy the internal connections that over heat and catch fire! Running pulse peak current over speck is a very bad killer of MMCs!!! Garry Freemyer is "THE pioneer" in this area and we will be forever grateful for his work ;-)))
B. Do not run too much RMS current. The caps are slightly resistive due the internal resistance and dissipation factors. When you pump 15 amps RMS thought them, they heat up. The "new" poly (and constructions) are better and can take about 110C as opposed to the 85C of older caps. But "IR^2" rules the day... If the temperature deep inside (well insulated) the cap hits 120C it melts, flows, and blows up... On the outside, if the caps seem to be getting "pretty warm", STOP! and let them cool and consider that you need more strings to share the current. RMS has been a "flying spec". The CD 942 caps are improving everyday and that spec is changing, for the better, all the time. Thus, the old design charts with the new caps yield very robust numbers here!!
But really if you design your cap as we tell and it does not blow in the first ten minutes... Ion damage is the key that will limit its life..........
DC... If your care too, pass these specs to your friend about how we run our "coils on average" "in the wild"...
Run time 2 hours/year
Floating plate 640 volt poly foil/film caps as Cornell Duvalier 942C20P15K:
http://www.cornell-dubilier.com/film/9422000.htm
2500VAC at 60HZ = 2 hours/year VERY "low" currents!!!
2500 VAC 150kHz, fast linear decay, VERY "high" currents!!! (total run time (5% real duty cycle during actual operation) 200 seconds/year...)
Peak and RMS current ratings are NOT violated! (no thermal destruct issues)
I think that "200 seconds" of real high stress is the key to long life... Ion damage is just not going to do much in say "10" hours...
So what are the "unknowns"....
The "equations of life" that are given out by cap scientists and manufacturers fail miserably "in our case". The typical capacitor overstress equations just are not applicable...
We "do not know" what is the average life of a 942C20P15K cap at say 3000VAC 60 Hz.... A fairly simple experiment I wanted to do, but I could not find a "safe area" to contain the flaming failure situation... That would tell us more than anything about the expected long term failures of ion damage... Actually, it should be done at 1500, 2000, 2500, 3000 VAC at 60Hz to graph the effects vs. AC voltage...
I remain "skeptical" that running at 3X VAC is anything more than 33.3% "safer than "4X VAC"... We have lots of ionization,... and I suspect that the 300% over voltage is 75% as bad as 400% overvoltage... Might as well just run them at 1000% over voltage >;-)))
150kHz, 5% duty cycle, 12.5 Arms, 432A peak... Take the CD caps to their limits and just run them there >:-)) We all seem to be doing this experiment all the time... Try to keep track of how many hours. If your caps blow, contact Terry to have him take a look at them to do the autopsy... With DRSSTCs now, this may be a very "experimental thing" now too!!
Another great issue is the frequency.... MMC see mostly 60 Hz... NOT 200kHz... At 60Hz, WIMA suggests that the life is 80% off the DC life... Since MMCs see mostly 60Hz over voltages, The ionization damage might be vastly reduced... Gary Lau had a great, but forgotten, post on this:
http://www.pupman.com/listarchives/1999/March/msg00315.html
I should point out overall... MMCs blowing up is just NOT a problem...
"Getting" MMCs seems to be a problem.... Rell and DC have a minimum order of about 30 caps... The seems to be a demand for a smaller quantity source... I am working that ;-))
Cheers,
Terry
At 09:34 AM 8/19/2005, you wrote:
Even though the puncture thru the dielectric is "self-healing" operating them in the manner you describe will not stop the corona and tracking at the edge of the foil. Excessive operation in this manner will actually melt the dielectric near the edges and "self-healing" will not occur with such large damages.
Do it right the first time and you won't have to spend hours searching for a bad cap.
You can't ignore the pulsed rating. Beau Meskin has been making this type of cap for over 30 years and he says he just shakes his head every time he hears these claims.
But, if you have a lot of time on your hands to make repairs, why not just run them at the DC rating ......?
I was only trying to save the experimenter some time and grief.
Dr. Resonance ----- Original Message ----- From: "Tesla list" <tesla@xxxxxxxxxx> To: <tesla@xxxxxxxxxx> Sent: Friday, August 19, 2005 9:56 AM Subject: Re: capacitor question - rell.com order
Original poster: "Dmitry (father dest)" <dest@xxxxxxxxxxx>
> Original poster: "D.C. Cox" <resonance@xxxxxxxxxx>
>
> No, this is wrong thinking. Penny wise but dollar foolish.
>
> You need to allow at least Erms x 2.5 to 3.0 for safe Erms > performance.
>
> They use Erms x 3.0 to obtain the correct Edc pulse rating for
all > their
> high voltage reversal caps.
"Erms x 3.0" is equal to "Upeak * 2.12" - the same rule is used at derating of commercial pulse caps, which don`t use "self healing", but 942 caps use this technology.
http://www.pupman.com/listarchives/2005/Jul/msg00419.html
"1. Self Healing - MMC capacitors can be driven to twice and sometimes 3 times their rated DC voltage before they arc over internally. Due to their construction, MMC capacitors will self heal and continue to operate even after thousands of over voltage hits. This is pretty important since many coilers have "accidents" and things that arc the caps. Mica caps are destroyed after just "one" hit as are many other types of caps. Although it has been done, it is very difficult to destroy an MMC from over voltage."
"5. Peak voltage - For Tesla coils, we can use the DC rated voltage value and ignore the AC voltage rating. This is due to the short pulsed nature of Tesla coils and the very short clock time that the caps are run. Dielectric ion inception damage is just not a problem for us in the case of Tesla coils"
do you negate all that Terry said?