Wednesday, August 1, 2012

DRSSTC Δt1: Windwindwindwindwindwind.

I think I spent way too much time designing my (very first!) Tesla coil, so now I have to catch up by actually building it before the end of the summer. Over the weekend, I got started by winding the secondary. It's ~1500 turns of 28AWG magnet wire around a 4" PVC pipe. Having wound motors by hand before, I'm in no way a hand-winding purist and whatever gets the job done fastest is what I would go with:

I made a makeshift live center out of a 4" RC car wheel (riding on a shaft in the drill chuck) and got to work winding using the MITERS lathe as a hand-winding jig. (Not powered, although I'm sure that would work as well.)

In total it took about 2-3 hours to do. Definitely tedious work, but not as bad as winding a motor I think. I calculated that if I had turned the lathe on it would have been done in 7.5 minutes. But, it also might have snapped halfway through and ruined the whole thing. It would also have been difficult to check for crossed turns at that rate. I'm sure that with practice it could be done.

Before I did the winding, I cut four slots into the top of the PVC pipe in order to fit in two interleaved toroid holders. These are just polycarbonate cutouts that hold the 5" aluminum ducting that makes up the top load:

CineStar 6 and 2x4 Scooter, for scale.
I will probably coat the windings in polyurethane finish to protect them. But otherwise, the secondary is done!

In the design, I talked a bit about the driver's power electronics but I hadn't quite finished the signal board's layout at that point. The signal board is actually pretty simple: it will use one of the wootstick 2.0 STM32F103C4 dev boards that I used for several motor controllers. The wireless programming may come in handy for this project.

As far as the STM32F103C4 is concerned, its only job is to drive four LEDs in short bursts of square waves, which is the beauty of optocoupled gate drive. It should, ideally, have no idea that it's connected to Tesla coil power electronics. I will attempt to get it running at 72MHz so I have plenty of time resolution for fine-tuning the drive frequency.

The left-hand side of the board has the wootstick 2.0 board and LED gate drive outputs. The right-hand side has some switches for driving a precharge relay, main contactor, and gate drivers from a separate 14.8V battery. The two sides are connected by an isolated 12V-to-5V logic supply (DCR021205).

The tricky bit will be connecting this board, which will be some distance from the coil, with the power board, which will be heatsinked to the coil base. The plan for now is to use a series of shielded 3-wire cables: two for the gate drive signals and one for 15V power to the gate drivers. The good news is that the gate drive signals are fairly low impedance, since they have to drive LEDs. That should make them less susceptible to EMI. The bad news is that the entire thing could still get stuck by an arc. So, around the set of three cables for some distance out from the coil, I will add an outer shield to secondary (coil) ground.

Next up: building the primary and the base.


  1. Shane
    I'm working on a little project and I was hoping perhaps you could join our little group and give some input
    I have a group doing rocket stuff there "Team Prometheus" and a lot of what you do is of interest to me :) Perhaps we can do some work together on one of the many projects? Or please contact me

  2. Hey Shane,

    Glad to see you're a fan of Digi-Key! :) I made an SSTC driver too and a video of it:

    I found your blog when developing a BLDC driver and have been checking in periodically. Keep up the good work!


    1. Hey Chris,

      Thanks! That's really neat. I can't wait for the SiC FETs to come down in price a bit - they're getting to be competitive with IGBTs. What gate drivers did you use on your SSTC?