MMC disaster or "Curiosity Killed the Cap"
Kristian, Mon Jul 02 2007, 10:47PMI was just getting my spark gaps set and thought I would hook up my MMC to my primary gap, just to see. I'm wondering if I wired things correctly. I had the gap in parallel with the NST (15KV, 60mA) and the cap bank(string of 10 942C20P15K-F in series) in series off one side of the NST. The gap was set to half the distance the nst's where reliably firing across. I brought it up slowly with the variac, and it started to bang, but not in the gap. It was arcing on the underside of the MMC board. I shut it down immediately. Like a dumbass, I flipped the board over so I could see and fired it up again. The bottom side of the board was lighting up like crazy and the gap never fired. On top of all this, the caps were now resting directly on the floor and started arcing to the concrete through the casing of one of the caps. There was some smoke and a black spot on one of the caps. No worries though, I bought extra caps for just such an occasion.
The MMC bank is rated for 20,000V. I had my gap set for about half that. It seems that it would take the ~1/8" space in the primary gap before it would take the >1" space between the leads of the caps. Does Anyone know how I could test the caps to see how many I killed?
Re: MMC disaster or "Curiosity Killed the Cap"
J. Aaron Holmes, Mon Jul 02 2007, 11:11PM
Probably the NST was just arcing to ground through your cap. Don't forget your NST is midpoint grounded! If the NST is sitting on the cement floor, then even if you only connect one leg of the NST HV to your MMC which is also sitting on the floor, you may be providing a path to ground via your MMC (via arcing to the cement). Of course, when this happens, your NST is basically shorted out, so the voltage is now far below what is required to jump even your small 1/8" gap. Lift your MMC off the ground and your gap will probably start firing.
EDIT: Not sure on the testing. Perhaps ramp up the voltage on each cap slowly with a variac to make sure it isn't shorted internally. Shorted caps are the worst, since they'll cause all others in the string to be overvolted and possibly destroyed.
Cheers,
Aaron, N7OE
J. Aaron Holmes, Mon Jul 02 2007, 11:11PM
Probably the NST was just arcing to ground through your cap. Don't forget your NST is midpoint grounded! If the NST is sitting on the cement floor, then even if you only connect one leg of the NST HV to your MMC which is also sitting on the floor, you may be providing a path to ground via your MMC (via arcing to the cement). Of course, when this happens, your NST is basically shorted out, so the voltage is now far below what is required to jump even your small 1/8" gap. Lift your MMC off the ground and your gap will probably start firing.
EDIT: Not sure on the testing. Perhaps ramp up the voltage on each cap slowly with a variac to make sure it isn't shorted internally. Shorted caps are the worst, since they'll cause all others in the string to be overvolted and possibly destroyed.
Cheers,
Aaron, N7OE
Re: MMC disaster or "Curiosity Killed the Cap"
..., Mon Jul 02 2007, 11:43PM
I am not sure that you have you caps wired in correctly to start with. The caps should be in series with the primary (you should never run without a primary/secondary, as all of the energy from the NST will be going into either the spark gap or the cap, neither of which will really like it), and the cap/primary in parallel with the gap.
Also, I doubt you managed to kill the 942's, it wouldn't even suprise me if the one that arked to ground is still more or less alive. I am not sure exactly what it needed to get them to self-heal, but if you find one that is shorted you could probably fix it by charging it to a few hundred volts and shorting it a few times.
..., Mon Jul 02 2007, 11:43PM
I am not sure that you have you caps wired in correctly to start with. The caps should be in series with the primary (you should never run without a primary/secondary, as all of the energy from the NST will be going into either the spark gap or the cap, neither of which will really like it), and the cap/primary in parallel with the gap.
Also, I doubt you managed to kill the 942's, it wouldn't even suprise me if the one that arked to ground is still more or less alive. I am not sure exactly what it needed to get them to self-heal, but if you find one that is shorted you could probably fix it by charging it to a few hundred volts and shorting it a few times.
Re: MMC disaster or "Curiosity Killed the Cap"
Kristian, Tue Jul 03 2007, 02:12AM
Aaron,
The NST is mounted on 3/4" ply and is grounded to the mains, BUT the bolts holding the casing to the ply were in contact with the cement. On the first test, the cap bank was stood up on high voltage stand offs with plenty of clearance. I only flipped it when I noticed it was arcing on the underside of the board (caps arcing to each other, not to earth), and not arcing the gap. I flipped it to get a better look at where it was arcing from \ to. The dumb thing is I had a 3'x5'x1/2" sheet of HDPE leaning against the wall right next to where I was working.
...,
I did not have the primary connected at all, just the gap and the caps. I know this is bad for the components, but the gap is very robust
and I've been told that these caps will take a tremendous amount of abuse. I figured they would be able to withstand a second or two of misuse. The run time for both tries with arcs was about a second total. I shut down immediately each time. Really, I just wanted to see how loud it would be, hence the title of the post.
Also, I have physically shown the MMC to some very experienced people and there was debate about whether or not the caps were to close together. I was curious to see if the spacing was adequate. The caps have slightly more than 1" between leads with a 10x voltage differential. I have just now cut and epoxied 1/2" thick lexan strips between all the leads with differentials. The arc would have to jump about 3" now. If that was the problem, correct me if I'm wrong, but it should now be taken care of.
Anyway, what about testing the caps? If I ramp up the voltage on each cap slowly with the variac to make sure it isn't shorted internally, what am I looking for, as in how will I determine a cap is shorted internally by this method?
Kristian, Tue Jul 03 2007, 02:12AM
Aaron,
The NST is mounted on 3/4" ply and is grounded to the mains, BUT the bolts holding the casing to the ply were in contact with the cement. On the first test, the cap bank was stood up on high voltage stand offs with plenty of clearance. I only flipped it when I noticed it was arcing on the underside of the board (caps arcing to each other, not to earth), and not arcing the gap. I flipped it to get a better look at where it was arcing from \ to. The dumb thing is I had a 3'x5'x1/2" sheet of HDPE leaning against the wall right next to where I was working.
...,
I did not have the primary connected at all, just the gap and the caps. I know this is bad for the components, but the gap is very robust
and I've been told that these caps will take a tremendous amount of abuse. I figured they would be able to withstand a second or two of misuse. The run time for both tries with arcs was about a second total. I shut down immediately each time. Really, I just wanted to see how loud it would be, hence the title of the post.Also, I have physically shown the MMC to some very experienced people and there was debate about whether or not the caps were to close together. I was curious to see if the spacing was adequate. The caps have slightly more than 1" between leads with a 10x voltage differential. I have just now cut and epoxied 1/2" thick lexan strips between all the leads with differentials. The arc would have to jump about 3" now. If that was the problem, correct me if I'm wrong, but it should now be taken care of.
Anyway, what about testing the caps? If I ramp up the voltage on each cap slowly with the variac to make sure it isn't shorted internally, what am I looking for, as in how will I determine a cap is shorted internally by this method?
Re: MMC disaster or "Curiosity Killed the Cap"
Steve Ward, Tue Jul 03 2007, 02:12AM
The self healing happens instantly. If a cap reads short circuit, then its likely toast. I agree that the cap that arced through the exterior plastic is likely OK.
Steve Ward, Tue Jul 03 2007, 02:12AM
The self healing happens instantly. If a cap reads short circuit, then its likely toast. I agree that the cap that arced through the exterior plastic is likely OK.
Re: MMC disaster or "Curiosity Killed the Cap"
Kristian, Tue Jul 03 2007, 03:14AM
Steve,
You are one of those people that have seen this thing first hand. I showed it to you last month in Kalamazoo. What do you think happened? Was it the spacing of the caps on the board that caused the initial falure? You think it's alright? How can I make sure all the caps are still functioning?
Kristian, Tue Jul 03 2007, 03:14AM
Steve,
You are one of those people that have seen this thing first hand. I showed it to you last month in Kalamazoo. What do you think happened? Was it the spacing of the caps on the board that caused the initial falure? You think it's alright? How can I make sure all the caps are still functioning?
Re: MMC disaster or "Curiosity Killed the Cap"
ArcLight, Tue Jul 03 2007, 06:32AM
Like you, my little coil is powered by a 15Kv, 60ma NST. I originally used one string of ten CD-942 .15uF-F caps (in the interest of full disclosure, it was actually two strings of 5 mounted in series on either side of the primary and gap with the gap mounted in parallel with the NST.) However, I kept destroying caps (one at a time, not the whole bank at once) after only a few minutes (broken up into 15-20 second segments) of run time. Sometimes in a fairly spectacular manner ie. the middle of the cap lit up like the sparkgap and would have a 3/8†to .5†hole burned into the side of it after only a second or so. The burn was always next to the bleeder resistor which was mounted 5/16†from the cap. Immediately after the cap failed the bleeder resistor failed and would burn a black crater in the poly cutting board that it was mounted on.
After burning-up 5 or 6 caps I started mounting the bleeders on the opposite side of the poly and then when a cap failed you would see the output of the coil quit and the inside of the cap would light up. Then the bleeder resistor would arc over and burn the familiar black crater.
After having two more caps fail with the resistors mounted on the opposite side of the poly (not as spectacular, but still fatal for the cap,) I started using two strings of 19 caps. What a difference. The coil runs better (more, thicker, longer arcs) and no more blown caps even after many nights of hard running.
I'm not sure what to tell you about testing your caps. Even if a meter said that a cap is good, I don't know if I would trust it at voltage. That's why I make my MMCs so I can change individual caps and resistors with little effort and no unsoldering. I mount the caps on 5/16†poly cutting board with 1.25†#6 steel screws, washers, and nuts (because of cost, $7-$9 per hundred) and brass washers (because of low resistance and priced at $6-$8 per hundred.) If a failure were to occur I can just take off the nuts and washers and remove the cap. Then loosen the screws, replace the resistor on the back side of the board, tighten the screws, replace the cap and then replace and tighten the washers and nuts. It's more expensive, but I believe that the flexibility is worth it.
Regards, ArcLight
ArcLight, Tue Jul 03 2007, 06:32AM
Like you, my little coil is powered by a 15Kv, 60ma NST. I originally used one string of ten CD-942 .15uF-F caps (in the interest of full disclosure, it was actually two strings of 5 mounted in series on either side of the primary and gap with the gap mounted in parallel with the NST.) However, I kept destroying caps (one at a time, not the whole bank at once) after only a few minutes (broken up into 15-20 second segments) of run time. Sometimes in a fairly spectacular manner ie. the middle of the cap lit up like the sparkgap and would have a 3/8†to .5†hole burned into the side of it after only a second or so. The burn was always next to the bleeder resistor which was mounted 5/16†from the cap. Immediately after the cap failed the bleeder resistor failed and would burn a black crater in the poly cutting board that it was mounted on.
After burning-up 5 or 6 caps I started mounting the bleeders on the opposite side of the poly and then when a cap failed you would see the output of the coil quit and the inside of the cap would light up. Then the bleeder resistor would arc over and burn the familiar black crater.
After having two more caps fail with the resistors mounted on the opposite side of the poly (not as spectacular, but still fatal for the cap,) I started using two strings of 19 caps. What a difference. The coil runs better (more, thicker, longer arcs) and no more blown caps even after many nights of hard running.
I'm not sure what to tell you about testing your caps. Even if a meter said that a cap is good, I don't know if I would trust it at voltage. That's why I make my MMCs so I can change individual caps and resistors with little effort and no unsoldering. I mount the caps on 5/16†poly cutting board with 1.25†#6 steel screws, washers, and nuts (because of cost, $7-$9 per hundred) and brass washers (because of low resistance and priced at $6-$8 per hundred.) If a failure were to occur I can just take off the nuts and washers and remove the cap. Then loosen the screws, replace the resistor on the back side of the board, tighten the screws, replace the cap and then replace and tighten the washers and nuts. It's more expensive, but I believe that the flexibility is worth it.
Regards, ArcLight
Re: MMC disaster or "Curiosity Killed the Cap"
Coronafix, Tue Jul 03 2007, 07:16AM
Kristian,
I would suggest that next time to connect everthing up before trying it out.
It sounds like you've set the gap, finished all the pieces, set the safety...I think
you just got to switch it on!!
Here is how everthing should be connected.
Don't fire the thing up without a load, ie the secondary attached.
You've got a variac, so you can ramp it up slowly, unlike me, I just switch
it on at the full 15000V and watch sparks fly!!
Good luck,
Michael
Coronafix, Tue Jul 03 2007, 07:16AM
Kristian,
I would suggest that next time to connect everthing up before trying it out.
It sounds like you've set the gap, finished all the pieces, set the safety...I think
you just got to switch it on!!
Here is how everthing should be connected.
Don't fire the thing up without a load, ie the secondary attached.
You've got a variac, so you can ramp it up slowly, unlike me, I just switch
it on at the full 15000V and watch sparks fly!!
Good luck,
Michael
Print this page