Something new I'd like to see / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Something new I'd like to see

previous 1 2

Move Thread

LAN_403

Goodchild

Sat Nov 03 2012, 09:47PM

Registered Member #2292 Joined: Fri Aug 14 2009, 05:33PM
Location: The Wild West AKA Arizona
Posts: 795

Steve Caton and my self have been working on a coil now for about 6 months that uses a modulation scheme I would call innovative.

It's similar in setup to Steve Ward's latest QCW but we are planning on doing true ZVS phase shift bridge design. ei full phase shifting of one side of the full bridge, not just shutting it off early. It will also use the switches parasitic junction C to make the system ZVS by commutating the current from the primary into the parasitic C on turn off or turn on. This basically gives perfect linear rise and fall time controlled by dead time and parasitic C. Extra external C can be added to further slow down the rise and fall time to any value we want.

We also plan to use 8 water cooled full bridges all driving an impedance/current sharing ferrite transformer network that couples all the bridges into the Tesla coil load and galvanically isolates them from each other.

Any RF envelope shape should be achievable up to 800A.

Dr. Drone

Sun Nov 04 2012, 04:48AM

Registered Member #290 Joined: Mon Mar 06 2006, 08:24PM
Location:
Posts: 1673

Dr Hankenstein

Sun Nov 04 2012, 02:49PM

Registered Member #1642 Joined: Sun Aug 17 2008, 11:36PM
Location: Black Canyon City
Posts: 96

Ahhhh! Now that's innovation, Eric! That's so far over my head that I have no idea what you're talking about!

Back to my arm-chair for a cold one.

Keep us posted on the progress.

Hank

Goodchild

Sun Nov 04 2012, 09:04PM

Registered Member #2292 Joined: Fri Aug 14 2009, 05:33PM
Location: The Wild West AKA Arizona
Posts: 795

Long story short we put big power in and we get big sparks out

After all isn't this how the pro's do it?

ARM-CHAIR COILING LIKE A PRO

8155308547 Bed097a97d Z

Marko

Mon Nov 05 2012, 04:53AM

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

Hi guys

Wouldn't there also be significant saturation trouble with the modulation transformer having to pass many amps of DC through a winding that produces hundreds of volts? I assume large airgap would be required for this to work, like used on some tube amplifiers. So a custom transformer would be required...

Goodchild wrote ...

Steve Caton and my self have been working on a coil now for about 6 months that uses a modulation scheme I would call innovative.

It's similar in setup to Steve Ward's latest QCW but we are planning on doing true ZVS phase shift bridge design. ei full phase shifting of one side of the full bridge, not just shutting it off early. It will also use the switches parasitic junction C to make the system ZVS by commutating the current from the primary into the parasitic C on turn off or turn on. This basically gives perfect linear rise and fall time controlled by dead time and parasitic C. Extra external C can be added to further slow down the rise and fall time to any value we want.

We also plan to use 8 water cooled full bridges all driving an impedance/current sharing ferrite transformer network that couples all the bridges into the Tesla coil load and galvanically isolates them from each other.

Any RF envelope shape should be achievable up to 800A.

Hey - I'm now curious, why is Steve Ward's phase shift modulation "nontrue"? Is he using some quasi-PWM approach with duty cycle reduction on one half-bridge?

Anyway, the utilization of ZVS like you're describing is long known and promoted by richie burnett as "class DE" topology. I'm fairly convinced that the effect occurs naturally in pretty much all simple non-DR SSTC's, as well as DRSSTC's adjusted for slight phase lag...
The behavior seems to appear naturally by negative feedback from miller effect, which halts the turn-on for a bit until the output capacitance has drained! Of course, there's nothing wrong in trying to bring some control into this by adding adjustable deadtime.

Of course, the effect is negated if the deadtime is made too large; benefit of phase shift control is that power can be varied while keeping zvs all the time, unlike ordinary pwm. And Richie Burnett likely experimented with this before any DRSSTC's even existed!

I'd say there are at least two pretty good modulation techniques that may apply well for tesla coiling, depending on some factors:

1.) Phase shift control and ZVS: best used with MOSFET's and high frequencies, since they are fast at turning off current but suffer from output capacitance and diode recovery losses;

2.) "bang bang" control and ZCS, for lower frequency systems that use IGBT's; IGBT's enjoy good ZCS and hence this technique may be ideal for regulating power in DRSSTC's, induction heaters or audiomodulating CW coils.

Lately I'm getting more and more convinced that for QCW coils, stacks of parallel mosfets may actually be a better choice than igbt's due to better high frequency performance and ease of paralleling.

As a part of my armchair coiling career, I've lately been convincing Steve Ward to switch to parallel mosfet stacks instead of transformer-seriesed igbt bridges. It's simpler and cheaper, and currents in QCW coils are low enough that they could be handled by a sensible number of mosfets.

And now for some truly adventurous souls, to get even more power you can dunk your multi-mosfet inverter into liquid nitrogen, and enjoy 10x reduction in ON resistance! I performed this experiment with DC load tests and results are very promising. Gate drive voltage required is somewhat higher though, 18V was needed to turn all devices fully on!

I'd really like to try that with a tesla coil, but I don't have an easy access to LN2. Hope someone here will learn to build a cryocooler at home and make some liquid air for this purpose!

Hope that's enough innovative thoughts for now -

Marko

Goodchild

Mon Nov 05 2012, 08:25AM

Registered Member #2292 Joined: Fri Aug 14 2009, 05:33PM
Location: The Wild West AKA Arizona
Posts: 795

You bring some good point's Marko,

I'm not playing the who did what first game, the ZVS topology has been around for a long long time. I came across the idea from a mentor of mine that introduced it to me. If I'm not mistaken he was one of the original advocates of ZVS technology when it first emerged many years ago for SMPS design.

The marvel that is ZVS at least the way we plan to do it does happen at a small level in a normal DRSSTC or SSTC assuming dead time is present. However with the addition of external C in parallel with the parasitic C of the switch the rise and fall time can be made to suit any switching speed (to within reason of course). Furthermore the tail current and miller effect of the switch are now none existent due to the fact that there is no charge stored in the device at turn off (unlike ZCS). ZVS is not just for MOSFETs, matter in fact I would even go as far to say that IGBTs stand to gain more from it than MOSFETs.

I also have done a fair amount of research into whether MOSFETs would be a cheaper alternative to IGBTs. From what I gathered a MOSETs solution that would be rated to a high enough spec would be ether more expensive or about the same as the IGBT solution we plan to use. We did however strike it lucky with some IGBT's that have a positive tempco making them perfect for paralleled operation. Lately I have seen a lot of newer IGBT with this positive tempo and it seems to be becoming less of an issue.

We still plan to use a matching network though not only to 100% ensure proper sharing but also to galvanicly isolate the bridges so in the event of a single switch failure we don't lose 16+ switches. It also has another added advantage of impedance matching the inverter to any Tesla coil load of our choosing simply by changing a turns ratio. In a QCW this is a huge bonus as ringup time plays a large part in the proper operation of the system. Normally this would require changing out the tank capacitor for a different value.

So in summery I think the IGBT/transformer method is a better approach, on price it comes in very close to a MOSFET solution but has added advantages such as impedance matching and short circuit isolation.

Steve Conner

Mon Nov 05 2012, 11:22AM

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

Woooo, go Eric!

Just don't induction melt your steel frame.

Steve Ward and I had lots of discussions about paralleling H-bridges.

One way to do it is with a tree of two-winding common-mode chokes. Exactly the same method used for paralleling variacs, you can see it in a Powerstat catalog or the like. The chokes don't have to pass the full power of the inverters, so there are no saturation worries. They see the full current, but only a tiny amount of volt-seconds determined by the differences in switching speed between the multiple bridges. So they can be made with a few turns of thick wire on a moderately-sized core.

At least in normal operation. If one bridge failed, the chokes would see a huge excess of volt-seconds and saturate hard: I proposed adding a sense winding to detect this and shut the gate drive down on the next cycle.

Dalus used this approach (though without the sense windings) in his "Resonant Rise" coil with good results.

The other way is with a small air-cored ballast inductor on each bridge output. Steve Ward's Gigantor uses multiple primaries, which is just a special case of this, the leakage inductance between primaries acting as the ballast.

You seem to be doing it the third way, as Steve did in his Tesla gun, and similar to the output combiner on the Harris 3DX-50. The drawback is that each transformer must pass the full output power of its inverter, so they end up bigger and more costly for a given output power. The advantage is the ability to combine the outputs in series instead of parallel. But is that really necessary? The 3DX-50 had to do it to match a rack of 100 H-bridges into a 50 ohm load, but in a Tesla coil you can have any load you want by designing the primary circuit appropriately.

Goodchild

Mon Nov 05 2012, 05:01PM

Registered Member #2292 Joined: Fri Aug 14 2009, 05:33PM
Location: The Wild West AKA Arizona
Posts: 795

Oh you saw the steel frame, that was an experience building that thing.... Just imagine two EEs that have never welded before teaching them selves how to do it on a final product. Oh well those are the moments I live for.

You are 100% correct Steve we did pick the more complex route, but not without just-cause. Matter in fact we originally planned to use the common mode choke idea using binocular type cores. With our coil being so big and not really knowing what kind of tank impedance we are going to need in the end we thought it best to be able to change out inverter impedance easily for two reasons. One the sang on the bus caps may become so large that we end up hitting a voltage limitation. The second is that our tank caps are not easily changed or configured as they are large expensive MICA transmitting caps. Flexibility is our driving motivator.

The air core basest (multi primary) approach we also considered, but came to the conclusion that the primary would just be way to cluttered with 8 primaries with 8 to 10 turns each.... You do get the isolation with this method though!

Although these transformers have to pass the full power I think we are going to get away with some $15 toroid cores I picked out at mag-inc. No air gap needed! The cores have relatively low losses up at 500KHz and the maths say they should work for our application.

The last reason is you just can't beat the feeling of absolute isolation between bridges. I sleep better at night knowing it's bulletproof and we don't have to depend on a circuit to shut off the switches in the event of a switch failure (although we will have one of those as well

previous 1 2

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