Terry filter, sparks firing in capacitor with video / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Terry filter, sparks firing in capacitor with video

previous 1 2 3 next

Move Thread

LAN_403

Steve Conner

Fri Nov 25 2011, 04:54PM

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

The resistors are the most important part of the filter! I believe that you can safely leave out all the other parts.

Indeed, as you found out, if you leave out the MOVs from a Terry Filter, it may be a good idea to leave out the capacitors too, and then all you have left is the resistors.

rulmismo

Thu Dec 08 2011, 06:14PM

Registered Member #4187 Joined: Fri Nov 04 2011, 08:08PM
Location: Spain
Posts: 43

Hi I did some measurements about the spark issue with my just made probe.

Current measurements are of mains, as they are done with a regular clamp that can not measure the low secondary mA currents.

I would appreciate some feedback, my intention is to remove cap at all, as last post suggested.

On the other hand I am quite convinced that all this stuff would fix adding the MOVs from oriignal Terry filter design...

Regards!
]terry_filter_measurements.pdf[/file]

rulmismo

Wed Dec 14 2011, 09:37PM

Registered Member #4187 Joined: Fri Nov 04 2011, 08:08PM
Location: Spain
Posts: 43

Not any comment?

Marko

Thu Dec 15 2011, 07:00AM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

rulmismo wrote ...

Hi I did some measurements about the spark issue with my just made probe.

Current measurements are of mains, as they are done with a regular clamp that can not measure the low secondary mA currents.

I would appreciate some feedback, my intention is to remove cap at all, as last post suggested.

On the other hand I am quite convinced that all this stuff would fix adding the MOVs from oriignal Terry filter design...

Regards!

Hi rulmismo - sorry, I missed your last post completely.

I'm pretty dazed about what's happening here as it appears that the transformer - capacitor connection is exhibiting resonant charging (pretty much like a DC resonant charging effect exploited to double the tank voltage in DC SGTC's)

For some reason I thought that resistors would stop it, though not too well apparently.

In my opinion you're using way too large capacitor anyway though, most people seem to use a few hundred pf of doorknob caps, I used 330pF along with 5k resistors for a 12kV, 40mA supply,

I'm starting to get worried if this makes any sense though as the cap only seems to slow down the voltage overshoot, and not reduce it's amplitude at all.

In order for the amplitude to be affected I think the resistor would need to be in range hundreds of kiloohms which is required to de-Q the parasitic resonant charging circuit, which has characteristic impedance in this range.

So in the end even if you remove the capacitor completely, I think the overshoots would continue due to transformer's parasitic capacitance alone, which might be a good few hundred pf.

It's starting to seem that MOV's are the true most important part of the filter, which of course are the hardest to find component but only their soft clamping action (instead of switching action of spark gap) is what really removes any voltage overshoot on the transformer.

Resistors are next I guess... but I'm not even sure anymore how much sense does it make to add any capacitors.

I hope some light will finally be shed onto this dark matter of tesla coiling soon...

Marko

Herr Zapp

Thu Dec 15 2011, 08:23AM

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

Pork sayeth: " ......as it appears that the transformer - capacitor connection is exhibiting resonant charging ...".

Can this be the same Pork who previously castigated poor Zapp for "panicking" by suggesting the the root cause of the observed phenomonon was a resonant condition within the NST secondary/Terry filter circuit??

One more question for OP rulmismo: is there any chance that your NST is a high power factor transformer, with an internal power factor correction capacitor?

I have previously experienced odd high-voltage resonances with a SGTC Terry filter setup that was in some ways similar to yours: two paralleled 15/60 General Electric NSTs, both with internal power factor correction capacitors; a Terry filter with physically large (100W) 1K wire-wound (inductive) resistors; 500pF ceramic "doorknob" bypass caps, and no MOVs. I had 3-ball spark gaps at both the input and output of the filter array (1" dia brass spheres), and at different input power levels and tuning conditions I would sometimes see arcing across the output spark gaps, and sometimes across the input spark gaps, but never both simultaneously.

If rulmismo's NST is a HPF unit, is there any chance that the C of the internal PFC capacitor and the L of the tertiary PFC winding inside the NST (for power factor correction) is somehow contributing to an unexpected resonance?

Herr Zapp

Marko

Thu Dec 15 2011, 09:40AM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Can this be the same Pork who previously castigated poor Zapp for "panicking" by suggesting the the root cause of the observed phenomonon was a resonant condition within the NST secondary/Terry filter circuit??

Hi Zapp

Well, from your post you seemed to be suggesting resonant ring-up with terry filter capacitors, which would require them to be near resonance with the leakage inductance of the transformer and which I proved that it can't be true. I'm sorry for looking all mean but I just didn't want any misconceptions to be spread!

Resonant charging is quite a different effect which occurs only after rapid turn-off of the spark gap, and results in at most 2x the peak voltage immedaitely after gap quench (which is still damaging enough to the transformer). I guess terry predicted this and that's why he added mov's in the first place!

This is consistent with both rulmismo's spice simulation and the fact that his caps would flash over while his safety gaps were firing.

This is important matter to be pointed out since I've heard many people just using a low pass filter and considering the mov's redudant (which is what I was swayed into in the end too).

I'm also completely unsure now how much would loading the filter output with a tesla coil tank help, I'd need to run some simulations to check that...
That way we would have capacitance of the MMC requiring to be charged too, so you might be right... But if we omit the filter completely and connect the larger-than-resonant MMC and the voltage will never overshoot more than 1x - oh wait, though why do we need the filter at all then, again? It's pretty proven that NST's don't like to be directly connected to TC tank circuits!

The thing is in the distributed internal inductance, capacitance and resistance of the transformer which is not too easy to ponder.

Marko

rulmismo

Thu Dec 15 2011, 10:12PM

Registered Member #4187 Joined: Fri Nov 04 2011, 08:08PM
Location: Spain
Posts: 43

Hi,
1-With just NST, output is 6.3kVp.
2-With NST and resistors+sg, output is about the same
3-With NST, cap+resistors+sg, output is 8,2kVp (+40%)
All the cases in permanent state, gap adjusted to not fire.
In case 3 is when sparks appear in C, only at switch on.

In case 3 is when it appears the "continuous sizzle of surface tracking" refered by Herr Zapp in the first posts. It is also when the input current (LV side) has a strong characteristic oscillation at about 150Hz plus the 50Hz.

I think that the NST is not HPF, it doesnÂ´t say anything about it in the plate.

Marko

Thu Dec 15 2011, 11:35PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Hm, that looks like a very high steady state rise compared to what is expected. Have you tried halving the capacitor value and then measuring it?

Also, what kind of "continuous sizzle" is that you have there? It could be messing with your measurement and even the max voltage because any tiny arcs that appear might induce additional charging effect.

On the other hand, the self capacitance of the transformer might be really high, as much as few nF (I have no idea anymore!) and with your high parallel added capacitance it's indeed starting to resonate like a tesla coil!

In any case I'm not sure why you insist on using such a large capacitance though. If Terry did so I suspect he did it deliberately while counting on mov's to remove the 2x voltage overshoot.
I have 330pF in my coil and am worried enough now

I'd definitely reduce yours but am not sure what kind of benefit would it be to remove them completely as Steve suggested. Guess we have to wait until an expert for this shows up.

I hope you got rid of that wooden board and metal can resistors though!

Marko

Herr Zapp

Fri Dec 16 2011, 01:17AM

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

rulmismo -

Regarding the "continuous sizzle" of corona, it should be easy enough to locate the source. At night, just turn the lights off, wait a moment or two to let your eyes acclimate, and examine all exposed metal that's at high potential, especially any points, corners or sharp edges. If you can hear the sizzle but can't visually locate the corona, use a 3' long piece of small diameter polyethylene tubing like a stethescope to "probe" the setup .

Later tonight I'll make some measurements of the inductance of the 1K, 100W wire-wound resistors in my filter array and post them as a value you can plug into your Spice model, if you want to.

EDIT: I went back and looked at the photo of your filter array, and the ends of the capacitor leads projecting through your terminal block are guaranteed corona generators and arc-initiators. Pull the caps back out of the terminal block so the ends of the leads are buried inside the metal sleeve of the terminal blocks; I almost guarantee this will eliminate (or at least reduce) your sizzling and may even cure your "arc-over" problem.

EDIT #2: Each of the wire-wound resistors in my Terry filter measure around 6 millihenries (turns out they are 160W parts, not 100W as I thought).

ANY sharp points, corners or edges on charged metallic parts are severe field-gradient "concentrators".

Herr Zapp

rulmismo

Fri Jan 06 2012, 07:28PM

Registered Member #4187 Joined: Fri Nov 04 2011, 08:08PM
Location: Spain
Posts: 43

Hi , I come back again after some additional findings

The "sizzle" just occurs with cap connected.
I tried to see any corona in dark room, but anything seen.
I tried to connect just the cap in parallel with the output, to discard any effect of my (to improve) chipwood base plate and same sizzle occurs.
I changed the tips of the caps to get them into the terminal blocks, but not noted any improvement.

I tried to change the cap value, with lower caps gets better (less overvoltage).
I changed input voltage to have a better measurement of the primary NST inductance. With this, and with winding resistance measurements I have done a SPICE model and surprise!, resonance at 156Hz with 2,25nF cap

So my reccomendation to everybody is take care about what value of C you use, it seems easier than you may think to get into resonance troubles just with the filter....

See pdf for graph and simulation.

]terry_filter_resonance.pdf[/file]

Update 21/08/12
I found that the variac I use introduce some kind of resonance when I move over 150V (you can hear th fizzle and the sg fires always at this point).
DonÂ´t know why, but in any case the additional leakage inductance of the variac probably will lower the resonance freq. stimated above. As far as I know it hasnÂ´t any caps in it.

i.e. switching on without the variac (220V direct) sometimes the sg is not fired, but wiith it you pass 150V and bang!

previous 1 2 3 next

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