Tesla Coil Build Thread

[Sigurthr]

Excellent work!

Shorten your antenna, you shouldn't have to move it past where it is now, but it is WAY too long. It should be no taller than the top of your primary (for a helical primary like we use).

What voltage are you running the DC bus off of?

EDIT: also replace those alligator clip connections with more of the 10ga(?) you're using. The entire HV side after the rectifier is high current high frequency and any impedance bottlenecks will severely hurt performance. I got a two inch increase in CW arc length by removing five inches of 12ga between the rectifier and bridge. If you can mount your rectifier and filtering cap(s) right on the bridge like I did with mine you'll squeeze even more performance out of it.

 

[Atomicrox]

Thanks!

Today it was around 120-127Vac, which gives 170-180Vdc. Mains voltage varies a lot around here, I've seen it as low as 110Vac.

Edit: the top of my primary is very low, the antenna'd have about 5cm.. I could put the coil on the top of a box, though.

 

[Sigurthr]

Yep. My antenna is a 15cm length of 22ga wire coiled up to a physical length of only 4cm. I used a pencil as a coilform. It needs to be very low so it does not get to the high voltage part of the e-field. Vertically from the base of the secondary to the topload the e-field increases in voltage gradient as you go upwards. You need to be under about 75V-pk in order to not blow the clamping diodes, and you really only need 4V-pk in order to function. So shorten up that antenna!

Looks good for 110-130Vac (remember that running the coil will load the mains and drop the line voltage some)! I typically run at 115Vac (122Vac loaded down) maximum. My bridge can handle it all day but my primary would overheat in minutes. As is I have forced air cooling....but 2.25kW CW is no easy task, even for 10ga wire.

 

[DashApple]

Nice work


Funny I never used any bus filter capacitors , not using gave me longer arcs than usning them

 

[Sigurthr]

Bus capacitance's effect on spark length depends on run mode (CW or interrupted) and pulse width when interrupted as well. For shorter pulse widths (50% or less) increasing capacitance increases energy per bang as long as the mains can fill the cap(s) fast enough. You still get some bangs when mains voltage is low and thus short sparks but it lessens the rate. For longer pulse widths you start getting an interrupted CW corona instead of the sharp arcs if you filter the dc bus. I once did some testing with single shot bursts, adding 1800uF nearly trippled arc length over no filtering. If you increased the burst rate or on time the sparks would shrink above a certain pulse rep rate. This is the point where the caps can't fill before firing again or in the case of PW where you've sucked down all the energy in the cap before the pulse ended.

The thing about not using any filtering or not enough is that you get feedback interruption because when the mains voltage is below the minimum needed to sustain oscillation there is no power going in to the resonator so no feedback signal coming out. This doesn't happen if you use an oscillator based topology like the PLL though. For standard auto-tuning drivers an optimum minimum filtration point can be found where while running CW the DC bus voltage ripple is nearly equal to the Vpk of the mains input. In other words if your coil takes 15V minimum to oscillate your DC bus will go between 15V and 170V (for 120Vac input) so that there is always oscillation but you get the visible effect of having no filter caps. I did this on my smaller SSTC for quite some time.

The main reason to keep oscillation going at all times is that UCC chips tend to self-oscillate at around 8MHz when no input signal is present which makes them overheat. Just like you minimize siwtching losses, you must also minimize the time the UCC chips are running before feedback kicks in. Each time the interrupter fires and there is less bus voltage than your coil takes to oscillate the UCC chips are running at 8MHz or so.

 

[Atomicrox]

Haven't been able to play with the coil yesterday..

TBH I didn't even check if the primary was getting hot. The bridge got warm but not very much. The 12V regulator got very hot, I'll probably need a better heatsink for it!

Thanks ionlaser555!

Sig, do you know what's the minimum filter capacitance to keep it oscillating?

As a side note, I tried to run it CW once (saturday) by disconnecting the interrupter and connecting the EN pins to Vcc. Nothing happened, like it wasn't oscillating at all.

 

[Sigurthr]

Is your 7812 heatsinked? What voltage are you feeding in to the 7812?

The only times I've seen a coil oscillate on interruption but fail on CW is when using an auto-tuning driver that isn't 100% optimized in layout and component selection. In fact, I had this issue recently with Po210's driver board. Often you can jump start the oscillation by touching the antenna with a metal object that you hold in your hand. A work around is a simple 555 timer tuned to the resonant frequency - just wrap the output wire around the antenna wire and it should work.

The minimum capacitance for oscillation depends on the impedance of the primary circuit, the voltage applied to the dc bus, and the minimum voltage needed to sustain oscillation (feedback) ((which is determined by lots of factors related to antenna shape/size/spacing, secondary geomettry, etc)).

Vripple = I / 2(f x Capacitance)

You will need to run CW to measure the current draw of the coil unless you have a fast storage scope. Just measure the mains draw RMS Amps and use mains frequency (100 or 120Hz if full wave rectified) for the above formula. You want Vripple to = input AC voltage - (minimum oscillation voltage x 0.707), so use that to find the minimum C needed.

 

[Polonium210]

My driver board was never intended to be used in CW mode.

 

[Sigurthr]

Good to know! Btw, in my head the previous statement about unoptimized layouts/components and the anecdote of your driver's operation were two completely seperate ideas, but upon re-reading it sounds like they are connected, this was not intended. Fact is I never did figure out why your driver wouldn't self-start for me.

 

[Atomicrox]

The 7812 is heatsinked but the heatsink sucks. Already ordered a new one but it'll take a while to get here. I'm feeding 15Vac rectified to about 21Vdc.

Yeah, I happen to be using some of Po210's modifications... The only thing I see that could even remotely cause that is the difference in antenna diodes. Steve's use 1n60 and Po's use 1n5819.

 

[Sigurthr]

I doubt if it is the diodes as I use 1N4148's.

It has more to do with layout, enable pin handling/biasing, and I suspect mostly the GDT and antenna DC blocking / coupling caps.

 

[Atomicrox]

Both of them use .1uF antenna/GDT caps.. unless the cap material also matters. I'm using 50V monolithic ceramic caps.

As far as enable pin/interrupter, my circuit is currently identical to Steve's. I'm probably going to try Steve's DRSSTC-style pulsed interrupter next.

As for the layout.. mine is kinda improvised

 

[robomont]

here are some more grounding suggestions.i learned these from grounding radio towers.
silicone jell all connections.this saves alot of problems later.
if you ground earth gets dry frequently then the next time you water it use a box of salt and pour down the hole.we use powdered copper and salt in the desert.they are called chemrods.

my personal favorite wire is the 1/2inch copper stranded that is woven like rope.
it has lots of surface are and when buried in the ground works better than a ground rod..ive seen a tower site owner spool a whole large spool of copper wire in a hole and pour a concrete base pad on top of it.he claims he never has severe lightning problems.

i really enjoy talking hv.i once worked for a year rebuilding hv equipment for the mining industry.

 

[Sigurthr]

Ya, radio grounding techniques work excellently for TCs.

Re: driver issues; I am leaning towards capacitor material/construction. The only difference between two versions of my driver where the first wouldn't self-start and the latter which would was I used HV film/foil caps for the dc blocking - coupling caps on the GDT and antenna instead of ceramics on the one which self-started. Every driver I have made since with them also self-started, and the one which I made recently using ceramics (Po210's) would not.

Steve's early drivers (like the ones you're using) all had moderate self-start issues which is why the first versions all have 555 starters or pushbutton starters. I don't know if it is connected or not to the cap issue though. I (like everyone else!) based my driver designs on Steve's drivers, so the overall topology is preserved - clearly no issue there.

 

[Polonium210]

My driver does self start.

I'm not sure what you did to it? :thinking:

I'm not liking getting slammed in this thread anymore

 

[Atomicrox]

@robomont:
The stranded wire sounds like a good idea, might try that if I have to build another ground

@Sig:
It's probably the caps, then. I'll try with an extra oscillator next time!

@Po210:
Relax, I said it self starts with an interrupter but not on CW.
No one is slamming you and even though you stopped answering my emails a few months back (for no reason at all) I still went with your schematics because AFAIK you have the best non-DR SSTC coil.