Making an ultra high frequency Tesla coil... / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Making an ultra high frequency Tesla coil...

1 2 3 next

Move Thread

LAN_403

Inducktion

Fri Jan 15 2016, 07:50AM

Registered Member #3637 Joined: Fri Jan 21 2011, 11:07PM
Location: Buffalo, NY
Posts: 1068

Ok. So.

I know this is going to be riddled with issues just because of the frequency.
However.

I have wound a very, very small tesla coil secondary. It has a resonant frequency of 20 Mhz.

Yes, 20 mhz.

Now, the problem I'm facing; how to drive this....

I have vacuum tubes, and I can buy MOSFET's.

I know vacuum tubes would probably be a better solution in some ways, but they are inherently inefficient (dat filament) and are not small.

I have found several mosfets offered by Cree... They're SiC. 900 volt capability, and I'm unsure of which two models to purchase for this coil.

Model 1 has a gate capacitance of 150 pF, can handle 11.5 amps and has an on state resistance of 0.28 ohms. It costs $3.46 per. linky

Model 2 has a gate capacitance of 350 pF, can handle 23 amps, and has an on state resistance of 0.12 ohms. It costs $6.42 per. linky

Model one also has a slightly slower turn off delay compared to model 2 but it's small so it shouldn't matter too much.

Now my question is, should I even bother with trying to drive this with FET's, and, if so, which should I get that would work most efficiently? If not, how should I go about it with a vacuum tube?

As for the driver circuit I'm going with a PLL. I found a model of 4046 that can do 30 mhz. Specifically this one

Thank you!

DerAlbi

Fri Jan 15 2016, 08:08AM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

Hmmh. I would first tackle the problem at a higher level like a good topology choise.
If you go for something like a Class-H Edit: i mean Class-E drive the gate-drive requirements go down a little since its implementing ZVS and therefore the miller plateau is minimized. One could also think about a resonant gate drive in that respect.

What i also see as a problem is skin effect. I currently do not see how you get a coil with sufficient Q-factor done. The question is if you go for a LC-Tank (with the typical top load) or if you just use a Lambda/4 or whatever length optimized wire for the coil. My idea there is to pump up a standing wave inside the wire itself. I am not sure how this relates to the coils SFR.

I think you need extensive simulation on this.

Sulaiman

Fri Jan 15 2016, 12:28PM

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

You could start with something like this

and a sig-gen like this

WaveRider

Fri Jan 15 2016, 01:35PM

Registered Member #29 Joined: Fri Feb 03 2006, 09:00AM
Location: Hasselt, Belgium
Posts: 500

Keep in mind that the arc discharge will have a capacitance as well as a resistance. This capacitance can detune the coil significantly, so your PLL tracking system must have a tuning range that can accommodate the frequency shift range that takes place before and after arc ignition. (This depends on how big the arc actually is, as well.) Getting "perfect" ZVS is not always so easy either, because the changing arc size also changes the resistive loading as well. (No arc, small resistive/capacitive loading. Big arc, big resistive/capacitive loading.)

Have fun!

woodchuck

Fri Jan 15 2016, 04:33PM

Registered Member #39190 Joined: Sat Oct 26 2013, 09:15AM
Location: Boise National Forest
Posts: 65

I know of an Aussie amateur radio operator who made a 6M (52MHz) power amplifier by paralleling a bunch of IRF540's.

Craziest dang thing I ever saw.

Your choice of silicon (or silicon carbide) may not be the biggest obstacle you face.

Hazmatt_(The Underdog)

Fri Jan 15 2016, 06:43PM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

I was characterizing my coils for a while with my GR 1233 power amplifier, and all I was getting out of the coil was a short flame arc, not much unlike that of the plasma speakers.

So your arc is not going to be all that fantastic, it will be a dancing flame that propagates vertically, and length will depend on input power.

Getting down to the real nitty-gritty however, you most likely do not want to go with semiconductors

Why is that?

Simply put, the VSWR of the coil changes dramatically at the resonance point, this is because of the phase change right at resonance. Semiconductors have a difficult time with the large mismatches presented to them by the coil as the frequency changes slightly. I would expect VSWR changes over 10:1, and semiconductors typically have issues with 2:1, the reflected power causes dramatic heating and device failure.

Now you can better "match" your coil with a push-pull driver into a transformer, this has been done many times, and there is a "Popular Electronics" article which does this, but the output is a foot or less.

If I were you, I would change my design criteria to 13.56 MHz rather then 20 MHz. I have a 13.56 MHz amplifier in the garage right now that can deliver 2KW into a load, and these are fairly common for semiconductor processing, and the designs are out there, it's more realistic.
However, the mismatch will be difficult to manage, and you will have to be on top of your game to not blow-up the final stage if the coil has problems.
I would anticipate using a heavier gauge wire (doesn't matter anyway, the coil is "short") like 26 AWG, and a tungsten output terminal. The output terminal will be important so that the "electrical length" of the coil does not change much as it operates (maintain constant VSWR) and the heavier gauge wire will give more surface area for the heavier RF current you will be pushing.

You will also want to be able to control the input power to the final stage so that you don't just blow it up on turn-on, this can be done with a volume control (attenuator)

And you will also want a decent directional coupler that can handle the input power and will give you visual feedback to your degree of mismatch (VSWR). My PA has the coupler on-board, so look for a PA with the coupler built in, then all you have to do is attach a couple of meters to the output terminals (the meters will need a log scale as it is power, it's really voltage in terms of loaded into 50 ohms, but we calibrate the voltage scale to output power.)

There's more to it of course, but I think that's the major point.

I would probably go vacuum tube, they tend to be more forgiving, but for push pull output you will need a custom ferrite OPT.

Inducktion

Fri Jan 15 2016, 08:16PM

Registered Member #3637 Joined: Fri Jan 21 2011, 11:07PM
Location: Buffalo, NY
Posts: 1068

Hm.

I found videos of Zilipoper with his 17 mhz vacuum tube tesla coil, and I have to say his arcs are ridiculous.

However that being said I don't have even remotely the same power tubes he had. He was using some big radio frequency mommas. The biggest vacuum tube I have is a 6L6GC.

Would BJT's function any better than MOSFET's, or should I just go with vacuum tubes?

DerAlbi

Fri Jan 15 2016, 08:34PM

Registered Member #2906 Joined: Sun Jun 06 2010, 02:20AM
Location: Dresden, Germany
Posts: 727

BJT are great linear devices when it comes to low distortion power amplification. Thats what you need for good signal quality for High datarate RF but they allways yield extremely bad efficiency. There is actually a lot of development for mobilephone base stations going on because the power consumption is the highest cost for the communication companies.
I dont think you need linear behavior.. you need just fast switching. The benefit of the BJT is therefore worth nothing to you imho. For switching you might even run into saturation problems which cripples every BJT.
I think SiC Mosfets are the way to go. (In combination with an optimized topology) However i now understand as your frequency will change dramatically designs like Class-E will be complicated because they rely on a more or less constant frequency.
Please ignore "class-H" above. that was a brainfart, sry

I meant Class-E. Class E is easy. but it puts some stress on the switch. (it quadruples the needed voltage capability of the mosfet :-/ )

Hazmatt_(The Underdog)

Sat Jan 16 2016, 12:59AM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

If you are going to use semiconductors, I would recommend MOS, not BJT.

I was working on this problem with a class AB biased BJT power amp. It was push pull combined for roughly 150W output power into 50 ohms.
The bus voltage was a little over 100VDC. The amp was fairly stable into a fixed load, but as soon as the load changed dramatically one of the BJT's would explode, which took out the other pair because it was imbalanced. So a small imbalance in a high voltage amp causes severe imbalances.

FET on the other hand is pumping huge currents, like 30A from one device into a step-up balun. You have two of those so you're effectively pushing 60A at 28V or so into a step-up. What this gains you is if the fets see a change in say 10% of their operating current they can tolerate it without a meltdown. And if you combine two amps, it is more load tolerant to mismatches.

The consequence of this is that semis for RF are fairly pricey, and you would want a compensated design, which is why I'm recommending you just buy a pre-built brick amp.

Sigurthr

Sat Jan 16 2016, 02:01AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

If you're going to go semiconductor stick to Class-E or other "dual resonant" topologies. You can't hard switch 20MHz at any appreciable power level. Unmodulated, highly filtered DC fed Class-E can perform quite well even at 20MHz, it just becomes critical that you choose the right switch. If you try modulating it you're going to have reflected power issues.

Personally I'd go with a VTTC. You face literally no additional design problems for 20MHz than you do for 200KHz other than having to use ribbon instead of wire. I know 811A tubes can handle quite well and they're not really that pricy on eBay (~$30 ea). That's what I'd go with.

1 2 3 next

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