DRSS vs. SRSS / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

DRSS vs. SRSS

1 2 next

Move Thread

LAN_403

Ken Herrick

Mon Nov 30 2009, 09:07PM

Registered Member #2489 Joined: Mon Nov 30 2009, 06:23PM
Location: Oakland, CA USA
Posts: 17

Greetings all-

I'm new to 4HV but have been an off&on member of the Tesla List for some years. I'm now reconstituting my SRSS coil (for the old one see the attached image) and I have a basic question to throw out: Why resonate the primary?

I've done simulations of my tesla coil overall scheme that have brought this question closer to mind. In those simulations, where I take feedback from the secondary and have a non-resonant primary, the primary current waveforn is triangular at start, requiring the driving transistors to switch at current peaks. Within 10 cycles or so, however, the shape changes to sinusoidal as secondary resonance builds up and reflects back to the primary, and the switching shifts to near-zero-crossing. This is, I believe, as should be expected in hardware.

Thus my specific question: Given a) the same power draw from the mains, b) the same or similar power-capability of the power transistors, and c) appropriate feedback to establish & maintain oscillation, what additional advantages accrue with DR as compared to SR, beyond the avoidance of having initially to switch at primary current-peaks? If there is none, then is there not just a tradeoff between, in the DR case, having to design for a h.v.-tolerant primary and incorporate expensive resonating capacitor(s) vs., in the SR case, requiring power transistors that will withstand the initial 10-20 cycles of current-switching? Granted, the initial current-switching will result in more transistor power-dissipation, but would that not be preferable, especially in modest-capability coils?

Perhaps there's a DR advantage in that it might ring up quicker & thus cram a bit more charge into the top electrode before the spark extends very far and bleeds it off. But to do that, reaching the same breakout voltage as in a comparable SR case, I think you'd likely need quite a bit huskier transistors and/or a larger input power during the pulse burst. My comparison assumes essentially identical hardware in both cases.

Finally, as an aside, I note there's some comment on 4HV about secondary feedback schemes. What I am finding quite satisfactory is to incorporate a small 1:1 toroidal transformer, 60 uH or so for 100-120 KHz operation, in series with the secondary return to earth-ground. Clamp the primary to a low voltage with a MOV or zener diode or even just a pair of fast regular diodes in series. Bias the secondary's return-end to 2.5V with a pair of resistors--perhaps 22K each. Connect the secondary's hot lead directly to the input of a Schmitt gate e.g. an HC14. The 22K resistors plus the primary-clamping will limit the gate's input current and the initial pulse-burst turn-on will be sufficient to start the t.c. secondary into resonance and begin the feedback.

But just in case feedback would not commence from the first half-cycle of t.c. primary current, I incorporate a Schmitt-gate oscillator, running all the time and loosely resistor-coupled to the input of the above-mentioned gate, whose signal will get oscillation going and then be swamped out by the much lower-impedance feedback-signal from the 1:1 transformer.

Ken Herrick
Oakland, CA

Steve Conner

Mon Nov 30 2009, 10:08PM

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

Simply put, it's the difference between 14" of sparks and 14' of sparks. Dual resonance allows you to run primary currents of over 1000A and peak powers of hundreds of kilowatts. Of course not with that pair of IRFP460s borrowed from the mini coil

The DR concept evolved along with the use of powerful IGBTs.

By the way, there's a thread below asking about your spark augmenter topload.

Ken Herrick

Mon Nov 30 2009, 10:58PM

Registered Member #2489 Joined: Mon Nov 30 2009, 06:23PM
Location: Oakland, CA USA
Posts: 17

I appreciate that you can essentially null out the primary inductance & thereby facilitate much greater current flow--but I was postulating more-or-less equal hardware in both cases. To satisfy modest expectations like mine, why bother with primary resonance?

KCH

Steve Conner

Mon Nov 30 2009, 11:25PM

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

Well I suppose you're right, if you're satisfied with modestly sized sparks, then there's no need for dual resonance. But...

It provides an extra stage of impedance step-up. If I remember right, your entertainment machine used a Marx-like arrangement with a single turn primary. By taking advantage of dual resonance, could have used a multi-turn primary and a full bridge, making it cheaper and simpler.

The primary current in a SSTC is only sinusoidal under light streamer loading. As you push for bigger sparks, the secondary gets damped too much to reflect current back to the primary, and the waveform becomes squarish. Zero current switching is lost. This means that SSTCs hit a wall in terms of spark length, beyond which they self-destruct. Dual resonance breaks this barrier.

Primary current feedback only works with DR, and it gets the best performance out of IGBTs: they like to be turned off at current zeros.

Ken Herrick

Tue Dec 01 2009, 12:11AM

Registered Member #2489 Joined: Mon Nov 30 2009, 06:23PM
Location: Oakland, CA USA
Posts: 17

You keep late hours, Steve! My previous primary actually ended up about 3 equivalent turns, with about a total of 1800V peak of drive.

Correct me if I'm wrong, but won't the secondary become damped only while the spark occurs? Prior to that, I suppose there might be some damping due to residual air-ionization left over from the previous spark, but I shouldn't think very much. I see this in simulation (not any left-over damping, of course). For each "spark", there's the same build-up of secondary voltage each time, then a quick jump to near-zero when I switch a 100K "resistor" across it to simulate the spark. Exactly the behavior I saw in my real hardware via a nearby scope probe, as a matter of fact.

I'm not contemplating IGBTs since the 2 secondaries I have operate at ~100 & 120 KHz & I'd be worried about the uncontrollable conduction "tail" being too long. But have they become improved since I last paid attention to them?

KCH

GeordieBoy

Tue Dec 01 2009, 12:43AM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

As far as I'm concerned the justification for a Dual Resonant SSTC design is really a power factor correction one...

Sure you can achieve impedance transformation by getting any step-up ratio you want... Just choose hardly any primary turns, lots of secondary turns and really close coupling, to get good power transfer but there is a limit to how far this can be pushed.

When the resonant TC secondary is driven from a non-resonant link coupled primary the inverter's load current is always composed of two components. Those components are the reflected secondary current (which is mostly sinusoidal and in-phase with the inverter voltage when driven at resonance,) and the magnetising current of the primary (which is 90 degrees lagging and triangular since current through an inductor is proportional to the integral of the applied voltage.) This magnetising current serves no other purpose but to support the alternating magnetic field around the primary.

As the spark loading increases the Q-factor falls and the base-impedance of the secondary rises. Therefore the resonant sinewave component of the inverter current falls and the triangular magnetising current starts to dominate dramatically. Only increasing the coupling coefficient can act to regain the balance of wanted in-phase sinusoidal current against unwanted out-of-phase magnetising current. And there is a limit to how tight the coupling can be made before flashover becomes a problem in air.

This triangular magnetising current is not a particular problem for small SSTC's using MOSFETs as they prefer to switch lagging (net inductive) loads and can easily turn-off moderate currents at the drop of a hat without breaking sweat. However, trying to interrupt large currents quickly is just about the worst thing that you can ask of an IGBT!

The situation changes, however, if you apply power factor correction to the primary coil. This could be done with either a parallel PFC capacitor across the coil, or by using a resonant capacitor in series with it. Either way the out-of-phase magnetising current is cancelled out and the inverter then only has to supply the real power required to power the TC and not the VARs. The dual-resonant nature of the system also means that the overall network is very frequency selective. Despite the low loaded Q of the secondary when sparking the overall Q of the mode being driven can still be high and the load current will look reasonably sinusoidal. As Steve Conner said, IGBT's much prefer to be left to switch off naturally as the current goes through zero than be asked to turn off right at the peak of a large triangular current waveform.

As for why the DRSSTC uses a series resonant circuit instead of parallel, it comes down to an issue of impedance transformation and convenience of drive. The series resonant primary just happens to transform the reflected load of the secondary down to a very few ohms at the output of the inverter. (Without the secondary in plase XCp would perfectly cancel XLp at the drive frequency and the primary circuit would theoretically present a short-circuit to the inverter. It is only the eventual losses from secondary breakout that prevent the primary current from heading towards infinity!) This allows the multi-hundred kilowatts of peak power Steve mentioned to be sourced from a relatively low DC bus voltage of 400VDC or so. The series resonant circuit also presents a rising impedance to all of the high frequency harmonics of the inverter's output waveform so it can happily be driven with a voltage-source squarewave inverter. (A parallel circuit would require a current-source inverter instead.)

Dual resonant isn't without it's down-sides. In theory you end up expending energy to charge up both the primary resonant circuit components Lp Cp and the secondary resonant components Ls Cs during each burst. However only energy in Cs is immediately available to the spark upon breakout. In practice though, I have been told that most of the "trapped" primary-side energy makes it to the secondary side eventually anyway, and there is little power recycled to the DC bus at the conclusion of each RF burst.

-Richie,

Ken Herrick

Tue Dec 01 2009, 03:41AM

Registered Member #2489 Joined: Mon Nov 30 2009, 06:23PM
Location: Oakland, CA USA
Posts: 17

That's a good explanation, Richie, and thanks for it. I'll have some re-thinking to do...

KCH

teravolt

Tue Dec 01 2009, 05:34AM

Registered Member #195 Joined: Fri Feb 17 2006, 08:27PM
Location: Berkeley, ca.
Posts: 1111

I don't think I can add much to Richie's explination but when a series resonant circuit is driven by a AC source at the resonant frequency great voltages can be achieved at the L-C node. Like Richie said impedance transformation is the name of the game. In a regular tesla or any tesla for that mater the primary is tuned to the secondary in some manor. From what I have understood from this web sight is that spark length is a product of the amount of power you can deliver to the secondary. I am not shure if I am right about this but what makes a DRSSTC sapirior is ring up. When a DRSSTC gets to a certan power level the sparks have a tendancy to leap out ward to lengths that can be 2 or more times the secondary coil length. Steve McConner knows more than I and could offer a better description.

HV Enthusiast

Tue Dec 01 2009, 07:05PM

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

In simplest terms, when the primary is resonant it approaches zero impedance which means enormous peak currents can be generated in the primary.

Sure, as Richie said, you could use fewer primary turns to get lower impedance, but you will never get as low an impedance from say a single turn than you will when running a primary resonant.

Ken Herrick

Tue Dec 01 2009, 11:27PM

Registered Member #2489 Joined: Mon Nov 30 2009, 06:23PM
Location: Oakland, CA USA
Posts: 17

I'm uploading 4 dwgs to show what I've been simulating. Two are the double-resonant and single-resonant schematics, which are identical (to the left of C6) except for a) DR vs. SR, b) feedback from primary or secondary respectively and c) the t.c. ratio. The other two are waveforms for the two cases. In the SR schematic, the two t.c. primaries are there just to a) deliver mains dc to one of the C4/C6 cap's while the other one refers to mains common and b) diminish a bit the copper loss. In both, while the MOSFET ckt is a bit odd, it's essentially a half-bridge, with crossover-control, delivering ~650 V p-p to the 2 primaries or to the resonant-primary ckt. Also note that I use a pair of diodes at each MOSFET to decouple the intrinsic diodes.

For those familiar with SiMetrix's freebie simulation app, I'd be over the component-limit if I added so much as 1 more resistor! And although I'd installed a PSpice model for the transistors I'm actually using, that was too much for the freebie. Hence the STEs.

Now to the waveforms: At the top is the voltage drop in mains-supply capacitor C6; note that it's essentially the same in both cases, indicative of equal mains-drain during the pulse-bursts.

Next note the outputs at 200 us: some 90 KV greater in the SR case.

Next, the Q1 MOSFET voltage, essentially the same in both cases except for a bit of (simulation-caused?) sloppiness at 150-180 us.

Lastly, the Q1 currents: somewhat less overall in the DR case, but shifting phase in both cases with respect to the Q1 voltage, and--with maxima at which the MOSFET has to shut off of 170 A in the DR case and 226 A in the SR case.

There's not a whole lot of difference in my view. If I were to bother to tweak the SR simulation, for instance, so that its peak KV output became the same as the SR's, I suspect that its max. turn-off current would become very much the same. And I may not have the DR resonating capacitor exactly right, but it's close.

I'm tempted to parrot the fabled commercial and ask, "Where's the beef?" but I know better. So somebody save me before I commit to 10 KV capacitors and a 10 KV vs. 600 V primary.

KCH

1 2 next

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