13.56MHz CW coil, 250W target, class-E drive topology / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

13.56MHz CW coil, 250W target, class-E drive topology

previous 1 2 3 4 next last

Move Thread

LAN_403

...

Fri Jan 26 2007, 06:00AM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

Looks good BP!

It amazes me how you can use a few coils of wire and carefully chosen caps to make a coil run at these pure evil frequiencies, I would be lost just trying to follow in your footsteps.

Keep it up!

Steve Conner

Fri Jan 26 2007, 10:22AM

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

Nice work, BP! I think you have the highest frequency Class-E coil so far.

BTW, 7 primary turns seems too much for such a high operating frequency. Didn't WaveRider only use 2 for his?

My gut feeling is that you may be mismatched because your scope doesn't display the waveform accurately. If you tweaked it to make it "look" right, you are probably underloaded, and the drain voltage is spiking high, then shooting back below zero and making the reverse diode conduct. Using fewer primary turns should load it more heavily.

BTW, I've never heard a MOSFET bubble, that is gross!

You must be getting the die hot enough to melt and burn the plastic case.

Marko

Fri Jan 26 2007, 11:35AM

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

My groundplane is a scrap sheet of copper clad that failed to etch properly (at all), so it's just a sheet of copper that's thin in some spots. =P

You aren't having any losses on it? Tried elevating the coil, or maybe cutting a trace in board to break the ''eddy ring''?
That copper looks like pretty well coupled to your primary, to me.

My gut feeling is that you may be mismatched because your scope doesn't display the waveform accurately. If you tweaked it to make it "look" right, you are probably underloaded, and the drain voltage is spiking high, then shooting back below zero and making the reverse diode conduct. Using fewer primary turns should load it more heavily.

Yeah.. lots of trials and errors, although waveform looks really good to me.

It simply must came to gross dissipations, since big mosfet is way too slow to properly switch on and off at taht frequency, it would just drift in linear region and die.

Smaller mosfets like irf730 are faster, but with 1ohm ON resistance, just putting lil little current through it at those frequencies gets him hot, then he gets uber hot, and ON resistance may increase dramatically, by factor of 5 or so before critical temperature.

But I guess that's all the fun with classE sport.

BTW, I've never heard a MOSFET bubble, that is gross! You must be getting the die hot enough to melt and burn the plastic case.

Where bad mosfet dies go when they die. ^^ .

WaveRider

Fri Jan 26 2007, 01:26PM

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

Hi BP.. Great going!!

I am sorry I didn't get back to you straightaway, but I'm hoping to do some simulations this weekend.

I have a couple of thoughts to offer, tho'..

1. Your RFC is way to large. 470uH is likely to suffer self-resonances which can suck power away from your spark. Use your scope to see how much RF voltage there is on your power supply leads.
2. The ground plane under your coil should be connected to the ground of the driving circuit.
3. Copper losses at 13.5MHz are fairly high, so use the thickest winding wire possible to cut down on losses.
4. In my latest coil version, I used 10 primary turns. If number of turns roughly goes as the inverse of frequency (keeping impedance constant), roughly 4 turns should suffice...altho' this should be taken with a grain of salt, because your coil geometry is different. Only simulations and measurements will verify this.
5. _Do not etch_ the groundplane board. Eddy currents are what give you the shielding.
6. Using a Faraday cage can improve performance by reducing coupling into all the "bench spaghetti."

Later!

Marko

Fri Jan 26 2007, 01:34PM

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

5. _Do not etch_ the groundplane board. Eddy currents are what give you the shielding.

I don't understand this one at all... The board is practically a shorted turn coupled with priamry probably better than secondary, and there must be excessive ohmis loss and items on fire.
'shielding', whatever it means, must be from some unknown mechanisms?

WaveRider

Fri Jan 26 2007, 01:44PM

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

Yes Firkragg, you are quite correct that it is a bit like a shorted turn. Where your misunderstanding lies is in the fact that the current is distributed over a large area of copper, so ohmic losses are actually very small when compared to a shorted wire turn.

Steve Conner

Fri Jan 26 2007, 02:37PM

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

Yes, good conductors like copper will mostly reflect the EM field rather than absorbing it. The induced eddy currents are what cause the field to be reflected, so as WaveRider says, anything that eliminates them will allow the field to pass through and make the shielding useless.

What this means in the land of circuits (as opposed to fields) is that putting a copper plate near a coil will mostly just reduce the inductance, instead of introducing large ohmic losses. Hence items on fire = 0

Think how much harder it is to induction heat copper or aluminium than iron.

An interesting consequence of shielding is that a quarter-wavelength whip antenna inside a Faraday cage effectively becomes a Tesla coil. The radiated energy can't escape the cage, so it gets reflected onto the antenna again, and an extremely high voltage builds up at the antenna tip.

I've heard stories of people getting air breakout at 144MHz by this method, at considerable risk to the high-powered valve RF amplifiers they used to drive it, since the impedance seen at the base of the antenna is changed drastically by the shielding too, and becomes a serious mismatch to 50 ohms.

I should also mention that when I said about accuracy of scopes, I meant that BP's scope won't show him what the waveform really looks like, because it only has a bandwidth of 20MHz, so it attenuates the harmonics that give the waveform its shape. If he tunes to get a nice picture on screen, the actual waveform is probably spiky and horrid.

It's still an impressive achievement to get breakout at all at these frequencies, once again, congratulations BP

Marko

Fri Jan 26 2007, 04:42PM

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

What this means in the land of circuits (as opposed to fields) is that putting a copper plate near a coil will mostly just reduce the inductance, instead of introducing large ohmic losses.

OK, but even considering inductive resistance of 'shorted turn is large compared to it's ohmic resistance and losses are lil low, how is it benefical then?

I don't see why would it make any difference if it was a mesh or a board etched with strips wich minimizes eddy currents.

All the time I tought only purpose of conductive plane under the coil is to act as a low-loss 'plate' with capacitance towards secondary, playng earth.

It keeps electric field between itself and coil (cage does it even better) and so it keeps trouble associated with it.

I don't see why does mgnetic field bother anyone nor why should it be shielded from by coupling large piece of copper to primary, I tought it will just result in losses.

*meaningless rant removed for benefit of humanity

ragnar

Mon Jan 29 2007, 11:59AM

Registered Member #63 Joined: Thu Feb 09 2006, 06:18AM
Location:
Posts: 1425

Today I made some circuit boards to hold these variable caps (0-500pF, and I'm hoping they won't break down at 200V for now, but that's probably stretching my luck). They look like they were used in recieving equipment *shrug*

Because everything is so shakey with the 13.56MHz setup, and I had so many issues with efficiency and tweaking the operation of the flywheel, this one is designed to run at 6.78MHz, and I'm using an IRFP450 (which I've run successfully at this frequency before; well, at least, using the primary-in-drain-circuit topology.

I've taken the same core for the RF choke and wound seven turns of thick enamelled wire around it... I'll see how it performs -- I've no way of measuring it, but I'll count the number of turns on the original primary, since it's specced as a 470uH core when I purchased it.

The tuning inductor is 'squeezable'.

If you look at the final pic (network3.jpg), +ve/-ve power goes in at top-right, the primary is connected to the two terminals at bottom-right (the unconnected trace will be drilled and screwed through to a grounded heatsink). The gate driver (with self-DC-biasing transformer) will connect to the middle (right).

This thing is HEAVY.... and I haven't even put a heatsink on it yet... it just looks evil don't you think? :D

Marko

Tue Jan 30 2007, 06:10PM

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

Blackplasma, I don't know how do you have patience just to design all those pretty boards.. really great work.

I'l jealous of your variable caps. It's somehow, almost a shame to use such beauties in a SSTC.

And.. ..what maerial is your drain-choke-core? Low-density iron powder?

previous 1 2 3 4 next last

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