MMC for 24kV: Thoughts on best construction? / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

MMC for 24kV: Thoughts on best construction?

1 2 next

Move Thread

LAN_403

J. Aaron Holmes

Sun Jul 23 2006, 05:18PM

Registered Member #477 Joined: Tue Jun 20 2006, 11:51PM
Location: Seattle, WA
Posts: 546

A couple of months ago, I got my hands on a 200:1 3kVA potential transformer, and have been itching to build a coil with it. The capacitor is a major obstacle at this voltage, however! Together with the fact that I just blew up my Maxwell 37667 in a friend's coil, it's convinced me that the MMC route is the way to go.

I've played with JavaMMC a bit, and it seems to think that, for a total capacitance of 15nF, I could build a pretty darn bulletproof MMC using four strings of 40 CDE 942C's (the .15uF sort) for a grand total of 160 caps (and about $350). It'd give me a DC rating of 80kV. Seems "ok"

But that's a lot of little capacitors! I've seen a lot of MMCs where the caps are sandwiched in between plastic sheets. They look really "cool", however this arrangement would seem to require a lot of "switchbacks" for a long string of (40) caps, and at 24kV, it seems like something that might aggravate corona problems and lead to higher inductance. Wouldn't it? Of course, if corona became a problem, the simplest answer might just be to drop the thing in a tank of oil, but I'd hate to do that! Has anybody done any MMC construction for voltages of this sort and found any particular construction technique to be convenient and serviceable (e.g. if a cap croaks)?

Best Regards,
Aaron, N7OE

...

Sun Jul 23 2006, 05:27PM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

I don't think you really need to worry about corona, just make sure that you don't have both ends of the sting close to eachother...

Also, I think you have about 10x the needed caps in your bank (with the 942's you don't really need to overate the caps, with a saftey gap you should be able to run with strings of 15 caps without overvolting), so the easier method might be to use a single string of 15, and see if they get to hot...

Kolas

Sun Jul 23 2006, 05:49PM

Registered Member #102 Joined: Thu Feb 09 2006, 08:15PM
Location: Raleigh, NC
Posts: 169

Hi, why not use ceramic door knob capacitors?
They can be bought easily rated for 50kV

I've run across the same problem for my coil, and what's the benifit of using MMC type configuration versus using doorknobs in ||?

Kolas

J. Aaron Holmes

Sun Jul 23 2006, 06:06PM

Registered Member #477 Joined: Tue Jun 20 2006, 11:51PM
Location: Seattle, WA
Posts: 546

... wrote ...

I think you have about 10x the needed caps in your bank (with the 942's you don't really need to overate the caps, with a saftey gap you should be able to run with strings of 15 caps without overvolting), so the easier method might be to use a single string of 15, and see if they get to hot...

JavaMMC doesn't like 15x2kV caps for 24kV. Even with three strings of 30 each, the voltage rating is only "Fair". I suppose it is applying the 2xVpeak rule of thumb, though. I've heard that people "get away" with using fewer of these CDE942s, but D.C. Cox recommends *3x* the DC rating regardless, so even with 40 caps, I'm not there. I suppose if I'm going to go to all the trouble to solder all these little caps together, I'd like to "never" have to do it again

Kolas wrote ...

Hi, why not use ceramic door knob capacitors?
They can be bought easily rated for 50kV

I've run across the same problem for my coil, and what's the benifit of using MMC type configuration versus using doorknobs in ||?

That still qualifies as an MMC, I think

They may not be "mini" as voltage goes, but they're "mini" as capacitance goes!

I was thinking of doorknobs too, but it seems relatively little information is generally available about them, and folks say they're often not meant for pulse discharge applications at radio frequencies, but rather are meant for filtering. So they don't have the "voltage reversal" spec. There's a guy selling brand new 1100pF 50kV doorknobs on eBay right now for $18 each. Not bad. The guys is in China, though, and I have relatively little faith that I'd get any adequate sort of datasheet out of him if I asked. I could be wrong, though. Anybody else have any experience with "doorknob MMCs"? Construction would definitely be a breeze!

Aaron

Reaching

Sun Jul 23 2006, 06:10PM

Registered Member #76 Joined: Thu Feb 09 2006, 10:04AM
Location: Hemer, Germany
Posts: 458

doorknobs are the worst choice for a mmc in a spark gap coil. they will consume much power ,will get hot and crack off after a few seconds of runtime. doorknobs are the best for pulsed applications, such as marx generators, but in continious mode they will fail after a short time. there are many people out there who can tell you that.

and for the mmc, why do you want to overrate the mmc in this dimension?, ok i never used the CDE Caps but from what i heard they must be the best caps for this application,and really 350$ for a 15nF cap?for me its like a joke, i build sgtcs with 2kVa transformers and build up a mmc with wima caps (old stock) for 15â‚¬ per 10nF (15kV rating) and they never failed

Sulaiman

Sun Jul 23 2006, 07:47PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

I have been using Arcotronics (Italian) polypropylene film/foil capacitors
at their rated peak voltage in sg coils - no problem.
(If CW etc then major de-rating required)

Manufacturers data sheets usually specify 2.5x over-voltage for testing!

D.C. Cox makes TC's 'professionaly' and needs reliability/safety,
home use TC's don't need such safety margins.
Just do a short run and check for overheating.
(A fan is a good idea for high power/VAR polypropylene capacitors)
as they do not handle high temperatures.
So I'd start with one string of 10 capacitors (with balancing resistors)
If they get too warm then go for more caps in the mmc.

J. Aaron Holmes

Sun Jul 23 2006, 09:04PM

Registered Member #477 Joined: Tue Jun 20 2006, 11:51PM
Location: Seattle, WA
Posts: 546

Reaching wrote ...

for the mmc, why do you want to overrate the mmc in this dimension?

Perhaps then we ought to be asking why it is that programs like JavaMMC and MMCCalc don't like short strings of caps for MMCs? MMCCalc seems a little less adverse to shorter strings, however a "Good" rating from JavaMMC requires 2xVpeak, and an "Excellent" rating requires 2.5xVpeak, which would be even more than 40 series caps in my case!

I suppose I'd assumed that these MMC calculators accounted for the general consensus on MMC construction, but again, perhaps the CDE 942s go by "different rules", as they're called "extremely high dV/dt" caps?

Reaching wrote ...

350$ for a 15nF cap?for me its like a joke, i build sgtcs with 2kVa transformers and build up a mmc with wima caps (old stock) for 15â‚¬ per 10nF (15kV rating) and they never failed

Cost versus cap size misses the point, though. Bang energy varies as the square of voltage, and as I have a stockpile of PTs giving me ample opportunity to experiement with voltages of 24-28kV, I'd rather have a cap that can take the voltage in stride than one with twice the nf's

Sulaiman wrote ...

So I'd start with one string of 10 capacitors (with balancing resistors)
If they get too warm then go for more caps in the mmc.

One string of 10!

They're only good for 8A RMS, according to Terry Fritz's testing. If I crank my RSG up to 360-400 bps (at which point I get 3kVA throughput--the 200:1 PT's rated power), wouldn't I surely cook 'em?

What I seem to be hearing here is that folks don't believe in the "2xVpeak" think for MMCs. Perhaps one ought to simply consider the DC rating to be the peak AC rating? If I round my 33kV up to 40kV, then strings of 20 caps would be "more than enough", in which case four strings would give me 30nF with only half the caps I was planning on using (= half the $$). That's certainly not a bad thing, provided it's not gonna explode on me!

Perhaps Terry will drop by and bless us with some first-hand knowledge of the CDE 942C20P15K's...

Regards,
Aaron, N7OE

HV Enthusiast

Sun Jul 23 2006, 11:50PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

wrote ...

They're only good for 8A RMS, according to Terry Fritz's testing. If I crank my RSG up to 360-400 bps (at which point I get 3kVA throughput--the 200:1 PT's rated power), wouldn't I surely cook 'em?

I've run those capacitors at over 30A RMS with little problems. As long as you keep them cool, they should be fine.

Kolas

Mon Jul 24 2006, 01:27AM

Registered Member #102 Joined: Thu Feb 09 2006, 08:15PM
Location: Raleigh, NC
Posts: 169

I've heard the opposite about ceramic doorknob capacitors. Granted my coil and your coil will be seeing greatly differnt types of pulses, since mine is a tube coil.

I saw the guy in China selling them too, he says, he's the manufacturer of the caps, and I will ask him for a data sheet now.
The reason the door knobs were considered is; they would actually be more cost effective then a bunch of series\parallel caps such as your thinking of.

Kolas

Terry Fritz

Mon Jul 24 2006, 02:25AM

Registered Member #393 Joined: Tue Apr 18 2006, 12:30AM
Location:
Posts: 297

Hi,

To get numbers, you have to give numbers...

Peak voltage?
Peak current?
BPS?
Primary frequency?
Primary capacitance?
Primary inductance?

I hear 200:1 and 24kV and 15nF...

I could assume it is a 120V input 200:1 transformer for 34kV peak output volts. If you need 15nF, then two strings of 20. About $60. Each cap is 432 amps peak at 13.5 amps RMS. you can probably run them to 3000V each but I would not push the current much. They are designed with a 1.4X margin for current from CW destruct.

Each leg is 17kV above ground so corona is not a big deal.

The cap will last about 70 hours.

Don't use oil. Chances are if one caps blows up, they are "all" dead.

Testing and programs from like 4 years ago are probably out of date now since the caps have changed significantly for the better since.

Cheers,

Terry

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob