18th October 2022

DRSSTC III (long pulse coil)


Recently it has become more and more common to use pulse-skipping drivers for a DRSSTC. The advantage for pulse-skipping is that when OCD threshold is hit, the driver does not shut down the coil until the next enable pulse, it only shuts the inverter down until the primary current falls below the OCD threshold and then starts again. This allows for higher ontimes and therefore higher power levels. This way we can produce huge and thick arcs even at low BPS. Another advantage is that we can lower the primary current and compensate for it with higher ontime. That allows us to run very high power coils while we can even keep the currents below the IGBT's datasheet ratings. 

Some pulse-skipping coils:


img.1. - Long pulse coil by Chris Ralph

vid.2. - Long pulse coil by Florian Delmas

Further reading on pulse skipping drivers can be found here:

My first long pulse coil

And with "first" I really mean second as my DRSSTC II already features a pulse-skipping driver. But in that case I'm using pulse-skipping to lower the primary current, so that my IGBTs are still running within their specification, as that coil is often used during my shows and needs to be reliable. My new coil will truly use pulse-skipping to push a coil into the limits, meaning above the specs primary current and very high ontime in the millisecond range.

I had some nice CM150DY-12H IGBTs laying around which I bought for very cheap (around 8$/piece). I have mounted those to a heatsink which I bought from TME together with a 10000uF 450V bus capacitor and 1uF 1200V snubber capacitors.


img.2. - CM150 module


img.3. - Bridge, heatsink, fans


img.4. - Bridge layout

After 3D printing some parts and getting myself another pulse-skipping driver from Philip Slawinski, I have put together a rough idea of how I want everything to be mounted. Bridge, fans, heatsink, driver. I'm powering the driver with a small ~50W iron core transformer which puts out around 18VAC. 18VAC after rectification becomes over 25V which could be a bit too much for the driver, that's why an LM317 linear regulator set up for around 22V output was used as well. 


img.5. - UD+ driver

img.6. - Painting it red

img.7. - 3D printed front cover

img.8. - Front view


img.9. - Whole front cover

img.10. - Top view

img.11. - Mounted driver

img.12. - Current transformers

I'm quite happy with this compact design. I still need to add MMC, soft-start circuit and a fuse. 

For MMC I choose to use trusty Czech Tesla TC343 150nF 1000V pulse capacitors. 9S3P combination for total of 50nF@9kV. These capacitors have been proven multiple times that they are perfect for DRSSTC use.

For a primary coil I bought 8mm copper tubing. My aim here is a lot of primary turns (at least 7) for high primary impedance. 


img.13. - MMC

img.14. - Primary coil tubing

img.15. - Winding of a primary coil

img.16. - Wounded primary

Again, I'm very satisfied with how it came out. The looks are nice and the coil was easy to wind. But... I have messed up the secondary coil. Somewhere I must have made an arithmetical mistake when calculating how long wire I will need and so I ran out of wire while winding. 


img.17. - Front view

img.18. - Primary inductance

img.19. - Finished secondary

vid.3. - First light

To fix the secondary I bought the rest of the wire and very carefully soldered it to the rest of the winding. I have used multiple layers of varnish and kapton tape to further insulate it. 

The primary inductance resulted in 37.7uH which together with 50nF MMC should give me a tuning range down to 117kHz. 

Unfortunately the resonant frequency of the secondary coil came out a little higher than I expected. With this setup I'm only using around 5.5 turns of primary coil which is not what I wanted. Nevertheless it works and I find redesign not worth it. 

The secondary coil is wounded with 0.3mm double coated wire on a 12.5cm diameter PVC pipe. The length of winding is ~45cm resulting in around 1300 windings. 

First high power sparks

With some help from my friend Tomáš we were able to tune the phase lead on board of the UD+ driver. We also set the primary current limit to around 450A. 


img.20 - Tomáš scoping the bridge

img.21. - Static load test

img.22 - Bridge waveforms

vid.4. - High power sparks

Yet another set of promising results. During static load test the coil was drawing well over 2kW of power. We let the coil running with an iron pan as a load running for over 5 minutes continuously and the temperature of the IGBTs did not climb even 5C above ambient. The efficiency of the switching must be enormous. I'm also very happy with the waveforms. On image#22 we can see voltage between C-E of one of the low-side IGBTs (Yellow - 100V/div) and primary current (Red - 100A/div). Near perfect zero current switching. The real limitation during the static load test was the iron pan getting extremely hot. Few minutes later and the wood underneath would catch fire. This was also quite drastically heating up the primary coil. 

After all these tests I was quite confident during the first high power run with secondary in place.

Results of first high power test

On the video#4 we can observe the coil performing at well over 3kW with on-times in the range up to 3ms! The thing I was most worried about was the solder connection of the different wires on the secondary coil but that seems just fine. On the other hand, other problems became present. 

First of all, better tuning is needed. With long-pulse coils we can expect the sparks to be up to 5x the size of the secondary coil. In my case I was lucky to get 3x. 

Furthermore, I need to upgrade the top-load. My current top-load is some aluminum tubbing wrapped with multiple layers of aluminum tape. This is a cheap and simple solution but it lacks the surface smoothness. This resulted in arcs continuously shooting from multiple breakout points at the same time. Also even more importantly, at low BPS the arcs just want to jump down to the ground strike rail and we even got some strikes to primary coil! This is very destructive and I was lucky the IGBTs survived. This calls for better top-load construction. Smoother and bigger.