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

1 2 3 next

Move Thread

LAN_403

rulmismo

Mon Nov 21 2011, 06:54PM

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

Hi everybody,

I have been testing my Terry filter, C=2.4n R=1K 100W x 2. I have no other load yet than the filter for a NST 8kV 60mA.

When I start adjusting the safety gap size to avoid gap-firing, it occurs a peculiar side effect.

If the spark gap fires for any reason, sometimes (usually when it quenches out) appear some loud sparks between filter capacitor. It is like for any reason an overvoltage appears between capacitor pads. It helps that capacitors are not very well separated in my design.

Capacitors are serial connected rated for 2x15kV and have no sings of damage after the sparks.

I tried to simulate it in Spice, but couldnÂ´t cause an overvoltage of this kind.

Please see it in
http://www.youtube.com/watch?v=-InGCuOysNc
and make any comment. should I provide an additional spark gap in parallel with the filter capacitor?.

Regards!

Herr Zapp

Tue Nov 22 2011, 04:43AM

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

rulmismo -

STOP!!!
You can't just apply high voltage to the input side of an unloaded Terry filter. Basically all you have done is connected capacitors directly across the output of your NST, and apparently this has created a resonant condition that is "ringing-up" and generating monstrous voltages across your NST. It's a miracle you have not already destroyed your NST!

Step through your video frame-by-frame, and you'll see that the arcs are jumping from the "hot" conductors to the metal case of the resistors, then either to your NST case ground or your bypass caps.

So, DO NOT apply power until you have the Terry filter connected to a Tesla coil tank circuit.

Also, you may have problems with your metal-cased resistors. These should be big tubular ceramic resistors, not low-voltage metal-cased ones.

Herr Zapp

Marko

Tue Nov 22 2011, 04:58AM

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

Hi

rulmismo,

Well cautiousness aside, Herr zapp is pretty much panicking on this one - that NST is nowhere near resonance with only 1.2nF of capacitance across it.

One important job of the safety gap is to protect your NST from possible resonances with the TC tank capacitor, and this is why it should be set without the filter connected to the tesla coil, so that the NST just barely doesn't fire it. (And by no means expand it later!) This can be done without the capacitors if you're afraid of "resonances" of any sorts as well!

The real problem with your setup is that you used wood as a HV insulator which is really a terrible idea for a TC. Make sure you get rid of wood and use proper clearances between components in future.

Herr Zapp

Tue Nov 22 2011, 05:45AM

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

Marko -

I think it's obvious that there is some sort of resonant condition here. This is not just the slow, continuous "sizzle" of surface tracking across the wood baseplate.

Notice that the NST is continuously powered throughout the clip. Initially, the NST output voltage is insufficient to even fire the safety gaps. Then, one gap is triggered by partially shorting it with the pick. When the first gap is triggered, instantly both gaps begin firing, but there is no other evidence of extreme voltages or surface tracking anywhere in the filter array. The safety gaps sizzle normally for 2-3 seconds, when suddenly there is (apparently) a large voltage spike that is able to jump at least 1" from the resistor case to the capacitor array/NST ground.

There is no way that the 4kV nominal output from one leg of the NST will arc 1 inch; something else is going on. Even if the bypass cap value is far from the precise "resonant" value, a completely unloaded circuit can still ring up a substantial overvoltage.

Step through the video a few frames at a time, and watch where the arc occurrs.

Rulmismo - can you post some close-up photos of your filter array, especially the area around the bypass capacitors? Can you place a ruler or coin in the photo to provide scale?

EDIT: After stepping through this clip multiple times, I now think that all the arcing is confined within the bypass cap array. What I originally thought was arcing from the resistor body to the cap array now just appears to be extremely bright illumination from the arcing within the cap array. Still think there is some resonant phenomonon going on, as the arcing appears to be much brighter than would be expected from a 4kV, 60ma arc.

Herr Zapp

Sulaiman

Tue Nov 22 2011, 08:20AM

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

Before Terry published his filter I had a filter for my NST which consisted of the 'earth' centre point
with two rows of ceramic capacitors in a 'vee' shape
and two large tubular wirewound resistors
all mounted on a clear acrylic plate
which was mounted with two 18 mH inductors
(c750 turns 0.4mm wire on 4" pvc)
on a wooden frame, then a safety gap.

I had unexpectedly long tracking between the 'hot' ends of the capacitor strings, which I 'fixed' by glueing pieces of acrylic onto the plate to increase the tracking distance.

So, I had exactly similar eht 'over-voltage' as described by rulmismo.

I never really investigated the problem
just assumed it had something to do with the inductance//capacitance
I vaguely remember discussions on Pupman regarding inductance between capacitors and (safety) spark gap.
and now I no longer have that setup as I gave away the filter to Navrit.

Interresting though.

rulmismo

Wed Nov 23 2011, 08:22PM

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

Hi guys, thank you indeed for the feedback.

spark gap from live end to NST center tap is about 5mm. I used roughly as clearance rule 10kV/cm, maybe it is too optimistic.

UPDATE: after reading a little bit, screwing terminals to chipboard table at 8kV/<10cm doesnÂ´t seem too good, and might develop internal tracking, doesnÂ´t it?. (I say "internal" as surface is some type of wood-like melamine sheet).

When I started testing, spark gap was only about 1-2mm as I was assuming that more that that was not enough to fire with 4kV. With that setup, spark gap fired inmediately, but no sparks in the capacitors appeared.

I increased gradually the spark gap, and about 4mm it started appearing sparks in the capacitor.
With about 5mm (uploaded video) it does not fire without help.

As it seems problem is appearing when arc is quenching, I am supposing that some inductive kick either from resistors or from dispersion inductance from NST may be causing it, I did a quick dirty simulation with some typical parameters of an NST (I donÂ´t know the real ones) and the specific values of the filter, and got a x3 magnification effect, so 30kV peak!
That could explain sparks jumping in caps, as space between legs is about 1cm.

To avoid that magnification my guess is:
- add parallel resistor to add damping: from simulations I would need about 1M 100W resistor, and that is not very common I think...
- add parallel spark gap to filter cap with series resistor: this seems better idea

Any better idea/critic thought?

BTW I tried to find a comparison of different materials just to check "how bad" is wood chipboard against regular HV materials, but couldnÂ´t find it, anybody has a reference?

Marko

Thu Nov 24 2011, 02:24AM

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

Hi guys

rulmisimo,

Wood is as wrong HV insulation material you can find, it's really just a matter of time until it catches fire with HV terminals stuck into it.

Even with good insulators like polypropylene, PVC or acrylic you still need to ensure at least 3x surface tracking distance than your safe clearance in air is for the particular voltage (which would be, say, 2x breakdown distance, so tracking distance would be 6x breakdown distance).

Another thing, I wouldn't use those metal canned resistors, it seems far too easy for them to arc over. Apart from that, I even think their inductance could be causing the apparent overvoltage problems (which only occurs after safety gap firing, much like problem with those "filter chokes").

Replace the resistors with a suitably rated number of series connected carbon composition resistors (I had the film resistors fail open in this application).

You don't really need 100W resistors here, you only really have 3,6W of dissipatiion on 1k resistors! (+some of the filter cap energy). For small transformers you can even use more resistance into couple of kiloohms which wouldn't hurt at all.

Now, to prevent any possibility of future messups, adjust your safety gaps with only the NST connected across them, so they just barely don't fire (+0.5mm at most!) and don't ever touch them after that again.

And as I said before, 50Hz resonance with this small cap is not a problem, and here is a proof if anyone is still worried...
Reactance of your transformer leakage L is 133 kiloohms, and of 2.4nF cap it's 1.33 megaohms -> 1.2 megaohms of impedance, 6,6mA of current -> voltage rise to about 8,7kV which is quite negligible (although you could probably use much less filter capacitance too without ill effects if this bothers you).

Various size and shape gaps will fire at different voltages so it's really best to find that out by experiment, fix the distance and not being tempted to expand it later!

PS. If you're still clueless what material to use instead of wood, just get a fairly sized piece of PCB and build the circuit on it - you can have all components soldered to one side without holes. I actually used small pieces of copper pipe soldered down to a PCB - thought you might like that idea as well as it can be set far more precisely and is less likely to move accidentally.

Marko

rulmismo

Thu Nov 24 2011, 06:33PM

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

Hi Marko,
I certainly will change the wood baseplate,
about resistors, I donÂ´t understand well your point, Terry Fritz design was for 250W resistors, and usual schematic shown everywhere is for 100W resistors. My very crude simulation including secondary coil gives me about 30W mean with two "bangs" per 50Hz cycle, so donÂ´t understand from where exactly cames the mentioned 3.6W.

Something that I just realised is that in Terry design are used some MOVs in parallel with filter capacitor, these probably can provide the clamping-damping needed to avoid any oveshoot. That makes me think than some claims in several posts that I found about the MOVs not being really necessary are not very true indeed, at least an additional safety spark-gap should be provided just after the NST.

I totally agree with you about how to adjust it. I think that will change to a wider polymer base plate and will add a safety gap before the filter.

Thank you
Raul

Marko

Thu Nov 24 2011, 11:32PM

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

Hm, I tend to forget you're using such large filter capacitors which add considerable amount of dissipation to the damping resistors... if your caps charge to 12kV every half cycle and assuming they dump all their energy into the resistors, resulting dissipation would be 0.5*C*U^2*100 = 17, 28W total.

So dissipation on your resistors would be somewhat more than 12W per leg with added I^2R losses (3,6W per leg) which is still much easier than 100W.

Note that you can increase value of your resistors without affecting capacitor discharge losses significantly, just I^2R losses... so a sweet spot for a filter would probably be when those two are about the same.

Have you tried running a simulation with non-inductive resistors? I'm no expert for this but I highly suspect the inductance of the resistors is causing you this overvoltage behavior... also I'm pretty sure the low pass filter alone without mov's has worked very well for many people and there's no way it could cause overshoots alone (being a first-order network).

If your filter is really punishing your NST as in simulation, you'd probably be better off not using it at all I guess! So do run some simulations for low inductance resistors.

Marko

rulmismo

Fri Nov 25 2011, 04:08PM

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

I did but resistors inductance does not affect.

Supplying just the fillter with the NST, after the arc quenches in sparkgap you have a second order LC circuit formed by dispersion inductances of NST and filter cap.

Attached LTspice file if anybody wants to try it.

http://www.2shared.com/file/8dEKscsK/Terry_filter.html

1 2 3 next

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