GE 42L cap blow out / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

GE 42L cap blow out

1 2 next

Move Thread

LAN_403

radioman

Sun Oct 03 2010, 01:17PM

Registered Member #3026 Joined: Fri Jul 23 2010, 02:46PM
Location: Los Angeles, California, U.S.A.
Posts: 54

Hi everyone!

I am still trying to perfect my first tesla coil of the static spark gap variety. My neon sign transformer is a Franceformer, 7,500 volts at 60 ma. My MMC is the G.E. 42L4473 which is .047 mf @ 3,000 vdc. I am using 18 of these arranged in six banks of three caps each connected in parallel. The six banks are then connected in series. I figure the total capacity to be .0078 mf. And 3,000 times six gives me a voltage handling capacity of 18,000 volts. Am I right?? Trouble is the caps keep blowing out small holes in the sides through which a tarry black substance emerges! (fire, too) And, of course, coil performance stops. What am I doing wrong? What is the very best MMC to use

Thanks everyone!

Mads Barnkob

Sun Oct 03 2010, 02:40PM

Registered Member #1403 Joined: Tue Mar 18 2008, 06:05PM
Location: Denmark, Odense C
Posts: 1968

As far as I have read the G.E. 42L series capacitors will fail in tesla coil duty if they are pushed as hard as CDE capacitors normally are...

Look into getting some Cornell-Dublier (CDE) 942C series capactiors instead or get even more GE capacitors to reach a proper voltage rating.

Herr Zapp

Sun Oct 03 2010, 04:12PM

Registered Member #480 Joined: Thu Jul 06 2006, 07:08PM
Location: North America
Posts: 644

Radioman -

You've got a number of problems here; we'll address them one at a time:

1. How did you calculate your total MMC capacitance value to be .0078uF? If I understand your connection scheme, each of your "banks" consists of three .047uF caps in parallel; each "bank" therefore yields .141uF. Then you have connected six of these "banks" in series, which should yield a total MMC capacitance of ~.023uF, not .0078uF.

2. The "resonant" capacitor value for a 7.5kV, 60ma NST is around .021uF, which is very close to your calculated .023uF MMC value. With a "resonant" value capacitor, the voltage in the tank circuit can "ring up" to extremely high values that can very quickly destroy both the MMC and the NST. For a static-gap AC Tesla coil, you want to use a "Larger Than Resonant" tank cap value. The capacitor failure mode you described ("small holes in the sides") sounds like overvoltage failure, not overcurrent failure. With metallized film caps like the GE 43L series, overvoltage will usually puncture the dielectric in the center 50% of the capacitor's body length, whereas overcurrent will almost always cause failure at the ends of the cap, where the lead wire is connected to the end of the capacitor "roll". You need to increase your MMC capacitance value to around 1.5X the resonant value (1.5 X .021= .032uF) to reduce the danger of excessively high voltage caused by resonant rise. (See

3. From a failure-tolerance aspect, the way you have interconnected the capacitors in "banks" in your MMC is not ideal. Let's say that a single cap in one of your "banks" experiences an internal short. Now, that bank is effectively removed from the MMC, so all the individual caps in the rest of your banks now see a greater peak voltage. If the caps in several banks fail shorted, you may end up overvolting ALL the caps in your MMC. Conversely, if you assemble your MMC as "strings" of six series-connected caps, with the "strings" connected in parallel, one or more shorted capacitors in a "string" may overvolt the rest of the capacitors in that individual string, but won't affect the other strings. Electrically, the total capacitance value is the same, but the failure mode is more benign. This is more of an issue in large MMCs that have dozens or scores of individual caps, but is valid across the board.

The GE 42L caps will work fine IF you run them within their voltage and current ratings. However, the Cornell-Dubilier 942C series film-foil caps have higher peak current ratings than almost any other rolled-film caps, and have a "self-healing" design feature that help them to survive overvoltage conditions that will kill other caps. Obviously, a properly-designed MMC will not subject the caps to either current or voltage peaks greater than the cap's ratings.

Herr Zapp

Dr. Dark Current

Sun Oct 03 2010, 05:30PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

How can the voltage ring up to dangerous values with a static gap coil if the gap is at a fixed distance and won't allow the caps to charge more than to its breakdown voltage?

Steve Conner

Sun Oct 03 2010, 06:00PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Because when you get your new Tesla coil finished, you always go "MOAR POWARRR" and wind the gap open as far as it'll go.

With LTR, the gap just stops sparking before you blow anything, but with a resonant capacitor, no matter how wide you open the gap, the voltage just rings up until it sparks, and you can destroy the transformer or cap bank in no time.

radioman

Sun Oct 03 2010, 11:20PM

Registered Member #3026 Joined: Fri Jul 23 2010, 02:46PM
Location: Los Angeles, California, U.S.A.
Posts: 54

Herr Zapp

Thank you for your excellent, clear explanation! Yes, my math is a bit off, I should have said .0235 mf. I will try again without the banks of three. Your technical knowledge is very impressive. I canâ€™t wonder if you are either a scientist or an engineer of some kind. Thanks again.

Herr Zapp

Mon Oct 04 2010, 03:11AM

Registered Member #480 Joined: Thu Jul 06 2006, 07:08PM
Location: North America
Posts: 644

Dr. Kilovolt -

As Mr. Conner stated, once you get your coil running, an irrational "quest-for-more-output" frequently takes over. You cautiously open the gap just a little bit, the coil delivers obviously better performance, and nothing "bad" happens. Before you know it, you're readjusting the gap again, and the coil runs even better! After a while, the little man inside your head is whispering "Oh, go ahead, do it! Don't be a sissy! Nothing bad happened the last time you opened up the gap, and every time you do it, the streamers get longer!" You give in, and this time the coil is absolutely spectacular, much better than before, when suddenly .......silence (or maybe a dull humming from the NST or the MMC). With great trepidation you power it off, then on again ..... still nothing. The realization slowly dawns that there IS a limit, and that you've foolishly exceeded it, and now it's all over.

Also, there are odd, unexpected resonances and voltage nodes that can occur in a spark-gap coil. For instance, I know of a 15/120 NST powered TC that utilized an R-C filter array ("Terry filter") between the NSTs and the tank circuit. This filter array had 3-ball protective spark gaps at BOTH the input and output terminals, and the main spark gap was a multi-segment parallel copper pipe arrangement in a pressurized box so high velocity air was blown between each pipe segment. The ball gaps were adjusted so they fired at a substantially higher voltage than the main spark gap. However, at some combinations of input power level, toroid size, breakout conditions, etc. either the "input" or the "output" sets of ball gaps would fire like crazy, but never both sets simultaneously. Bear in mind that these ball gaps were in parallel with the main gap that had a smaller total gap spacing. All interconnecting wiring between the NSTs, R-C filter, and tank circuit was short and direct.

Just an example of unexpectedly high voltages that can develop in odd locations in TCs from less than obvious causes.

Herr Zapp

Steve Conner

Mon Oct 04 2010, 09:17AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Herr Zapp wrote ...

The ball gaps were adjusted so they fired at a substantially higher voltage than the main spark gap. However, at some combinations of input power level, toroid size, breakout conditions, etc. either the "input" or the "output" sets of ball gaps would fire like crazy, but never both sets simultaneously.

I've observed this too. I think the reason is that the ball gaps are 3-electrode while the main gap is only two electrodes.

The higher firing voltage of the ball gaps is predicted on the assumption that the voltage is distributed evenly across the two gaps, but there's no mechanism to make sure that actually holds in practice. The NST's output is symmetrical with respect to ground, but its inductance is so high that the whole circuit essentially floats at radio frequencies.

If your NST's winding centre tap is connected to a different ground than the centre electrode of the safety gap, that is even worse! Almost any amount of RF voltage could appear between the two grounds in the kind of scenarios we encounter with Tesla coils. Not only does that encourage the gap to fire, but it doesn't protect the transformer any more.

Once one of the ball gaps fires, all of the tank capacitor voltage appears across the other one and it'll go too. So, under unbalanced conditions their voltage rating is effectively halved.

HV Enthusiast

Mon Oct 04 2010, 11:56PM

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

Herr Zapp wrote ...

The GE 42L caps will work fine IF you run them within their voltage and current ratings. However, the Cornell-Dubilier 942C series film-foil caps have higher peak current ratings than almost any other rolled-film caps, and have a "self-healing" design feature that help them to survive overvoltage conditions that will kill other caps. Obviously, a properly-designed MMC will not subject the caps to either current or voltage peaks greater than the cap's ratings.

Herr Zapp

I agree. Although many people will knock the 42L series, they do work great, but as Herr said, are much less tolerant of an improper tune which can lead to more voltage spikes / peak currents.

However, the usefulness of the 42L series is about gone. In the olden days (about 10 years ago), they were very inexpensive, so a 0.33uF, 2kV MMC 42L cap was very attractive at about $3.50. Now, the same cap goes for over $16.00 each! Not sure why of the increase in recent years, but thats a lot of money and most folks just opt for the better 942C MMC capacitors.

Herr Zapp

Tue Oct 05 2010, 12:24AM

Registered Member #480 Joined: Thu Jul 06 2006, 07:08PM
Location: North America
Posts: 644

EVF -

Several years ago, a supplier of audio system components in Wisconsin had a huge inventory of GE 42L capacitors, in the 2kV and 3kV series. In quantity, these could be purchased for as little as $0.50 each. Alas, all are gone now, but I ended up with several hundred each of several different values. As far as I could tell, these were first-quality parts, and I never experienced any failures when they were used in spark-gap TC MMCs.

At a price of $0.50 each, it was painless to overdesign an MMC without any concerns about capacitor cost.

Herr Zapp

1 2 next

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