Skip to content

SSTC 2 “Big Bad” the CW coil

Project description

After having learned a lot about solid state Tesla coils, and having played with DRSSTC’s producing big sparks, I started to grow nostalgic towards the unique beauty of CW plasma. Producing considerable CW sparks requires a lot of power, and hence from start I designed this project as a large and powerful coil, for power levels exceeding 10 kW.
But unlike most other types of Tesla coils, such a device also has a potential for being used as a powerful high voltage source for various industrial and scientific applications. While this time I had much more experience, the project was not without problems, and again I had an opportunity to learn a lot from my errors.

I decided to make the drive circuit as simple as possible for this coil, in attempt to minimize chances of unknown behavior that may lead to blown mosfets (Most likely due to bad experiences with my SSTC1). The topology involved simply taking the secondary base current feedback signal by a current transformer, clamping the signal with zener diodes, and amplifying it with a pair of UCC37322 gate driver IC’s which in turn drive the switching devices over the gate driver transformers. To start the oscillation, a weakly coupled output from TL494 PWM IC was used.

Driver circuit design

Driver circuit design

R1 is used to adjust the frequency of the TL494, and R2 adjusts it’s output signal strength. My idea was to set it so that it only barely gets the oscillation going, and then the CT signal easily overrides it. The current transformer is connected to pads named CT1 and CT2, which after clamping by zeners go directly into gate drive IC inputs. I hoped this simple approach would help reducing the chances of blown silicon. A relay K1 is used to “arm” the coil, and will assure it is always in off state when power is applied. A linear regulator for supply voltage based on LM741 and 2N3055 power transistor is also present on board.

Driver circuit PCB

Driver circuit PCB

Construction

This coil was intended to run from a rectified 3-phase supply, which was provided by a suitably large 3-phase variable autotransformer (variac). The powerhouse was originally a full bridge o FDH47N50 MOSFETS, but later switched to 30N60 IGBT’s as the project progressed. Each switching device is mounted onto a separate heatsink with no insulating pad, making the heatsink electrically live – this is essential when we want to push the devices close to their power dissipation limits. As example, the internal thermal resistance of the devices may be 0.2K/W. The interface to the heatsink may add 0.1K/W, and the heatsink to ambient another 0.2K/W, for 0.5K/W. A sil-pad insulator may add as much as 1K/W alone, tripling the total thermal resistance compared to no-insulation approach! The heatsinks used were PC CPU heatsinks, with each individual cooling fan also serving to insulate the heatsink from the wooden frame.

Mounting of the devices to heatsinks is also critical. Mounting using the single mounting hole available on the device is unsuitable for these high power levels because the pressure is applied to the package in a way that it actually lifts the die away from the heatsink. Instead, the devices are clamped to the heatsink with positive pressure over their dies, and rubber pads are used to even out the pressure. This assures an extremely good thermal interface. Coil frame and H-bridge construction is illustrated in the following pictures:

CNC machined plywod and parts gathered for construction

H bridge PCB's

H bridge PCB’s

Full bridge assembly underway. I found some nice CPU heatsinks that had straight air flow path. The heatsinks are live with each MOSFET device mounted straight onto it, and the fan acting as insulation

Full bridge assembly underway. I found some nice CPU heatsinks that had straight air flow path. The heatsinks are live with each MOSFET device mounted straight onto it, and the fan acting as insulation

Almost complete bridge

The H bridge was divided into two half-bridges, each of which had a separate PCB, decoupling capacitor, and an individual gate drive transformer for every device. This is a good approach to use when driving large devices with high gate capacitance. The coil was divided into two sections, a high voltage power section and a low voltage signal section, stacked atop of each other. In the end, though, the power section started to get crowded from upgrades while the control section stayed empty – it was hard to predict the exact requirements as the project evolved a lot over time. Some more construction pics in the continuation:

Circuit breaker and bridge rectifier in the power section

Control board and the front panel

For long time I’ve had problems with my secondaries overheating in CW coils. PVC tubes which I generally used have a very low melting point and can be easily deformed by heating. There are several better options however:

  • Cardboard tubes, such as used in various sheet material rolls, can make a cheap form that is also quite resistant to heat. Cardboard is quite strong and a good insulator, especially if impregnated with resin or lacquer prior to winding! Cardboard must be well dried prior to winding.
  • Heat resistant plastics, such as Teflon (expensive, hard to get)
  • Glass or ceramic – can withstand extreme heat, at cost of being fragile
  • Composites, such as fiberglass impregnated with resin

A big advantage of fiberglass for me was the fact that I could make it into any desired shape, instead of having to rely on tube sizes I can buy. I also already had all the required components lying around – so I proceeded to make my own secondary by lining the inside of a fairly big PVC tube. It was important for the outside of the form to be smooth, hence it took some more polishing and glazing afterwards before the wire could be wound. A form like this can easily stand temperatures of over 300 degrees celsius.

This time the secondary coil is going to be constructed from fiberglass in order to resist high temperatures. Inside of a PVC pipe was lined in multiple layers of fiber glass doused in polyester resin

This time the secondary coil is going to be constructed from fiberglass in order to resist high temperatures. Inside of a PVC pipe was lined in multiple layers of fiber glass doused in polyester resin

After setting and a pain of getting the PVC mold away, a crude secondary form emerges

After setting and a pain of getting the PVC mold away, a crude secondary form emerges

Obviously, it needs some more resin and polishing

Obviously, it needs some more resin and polishing

After some more work, the form is ready for winding

After some more work, the form is ready for winding

The winding setup. Secondary end caps are made of MDF (Medium density fiberboard). The secondary form is spun by a variac-controlled drill

The winding setup. Secondary end caps are made of MDF (Medium density fiberboard). The secondary form is spun by a variac-controlled drill

Winding completed

The whole assembly starting to take shape

The whole assembly starting to take shape. Holes around the secondary base are intended to direct some air from the heatsinks into cooling the primary and secondary coils

The primary coil this time needed to withstand significant RMS currents. I chose copper pipe for it’s large surface area that is open to air, making it both good conductor and easy to dissipate heat. On previous coils, I used to wind my primaries from insulated wire on PVC forms. The insulation would hamper the dissipation of heat from the wire, and the PVC would again tend to deform causing the coil to loosen and unravel. I hoped to have finally fixed that problem here – there was far less thermal resistance here, and the supports for the primary were made from 6 sturdy polypropylene rods. The tube primary also has an advantage of being easy to tap. I constructed my tap from 4 fuse clips on a small chunk of PCB, where two extra clips help keeping the tap firmly attached.

The primary was made of copper tube. secured to pieces of polypropylene rod by plastic zip-ties

The primary was made of copper tube. secured to pieces of polypropylene rod by plastic zip-ties

Getting ready for the test run

Getting ready for the test run

First light

With all major components completed, the coil was getting ready for it’s first light. I used my big 3-phase variac as a variable power supply, with all three phases rectified to produce quite smooth DC supply. I slowly ramped up the voltage, producing an extremely hot CW plasma, about 40cm long, with characteristic 300 Hz buzz. I wasn’t measuring the input power at that time, but it was clearly several kilowatts. At the time I was using stainless steel for my breakout point, which got extremely hot (due to it’s low thermal conductivity) and started glowing white, producing lens flare on the image.

Big Bad's first spark

Big Bad’s first spark. Note the corona around the topmost secondary turn

I only ran the coil for a few seconds at time, and after I looked closely at the images I noticed signs of corona around the top edge of the secondary. This is a bad sign because it indicates danger of possibly very destructive breakout from the edge of secondary coil, which could permanently ruin the coil. I had to stop the runs until this was fixed.

My solution was to add a smallish toroid/corona ring to the top of the secondary, so that it screens the top of the secondary electrostatically. Some pics of toroid construction:

Copper tube was bent by filling it with salt, and wrapping around a suitably sized round object

Copper tube was bent by filling it with salt, and wrapping around a suitably sized round object

Short sections of copper wire were cut to be used as supports. They will be soldered to the pipe and a round piece of PCB in the middle

Short sections of copper wire were cut to be used as supports. They will be soldered to the pipe and a round piece of PCB in the middle

After soldering the PCB was segmented in order to break paths to eddy currents, which can produce significant heating in CW coils!

After soldering the PCB was segmented in order to break paths to eddy currents, which can produce significant heating in CW coils!

Soon after I put this toroid on, the first in a cascade of problems occured…

Problems and failures

Soon after setting up the toroid, I attempted a longer run which resulted in a pair of blown MOSFET’s. I didn’t even get to take any pictures of plasma before I witnessed the breaker violently pop out. Subsequent inspection revealed a likely culprit: one of the MOSFET’s seemed to have really poor thermal contact with it’s heatsink! At the time I designed this coil, I believed the common advice to use the as little as possible thermal grease in order not to have overly thick grease layer increase the thermal resistance. Obviously, this time I used too little:

Another view of the heatsinks, MOSFET's removed

Another view of the heatsinks, MOSFET’s removed

Closeup of the suspected heatsink. Most of the area where the device was lacks grease mark, indicating poor thermal contact

Closeup of the suspected heatsink. Most of the area where the device was lacks grease mark, indicating poor thermal contact

In meanwhile, I started suspecting that MOSFET’s may still not be up to the task for the power levels I’m likely going to push this coil to. Calculation has shown that I could easily get as much as 50A RMS in the primary coil with current topology. I decided to do some DC load tests with the same devices and heatsinks, which showed rather defeating results. MOSFET was connected to high current 5V source, and drew about 25A from it in steady state (125W dissipation). I measured the tab temperature and used the datasheet thermal resistance figures to project the die temperature – it was shown to approach 90 degrees C! Obviously it was unreasonable to expect the MOSFETs to perform much better than they did for this coil. Hence I decided to move on to IGBT’s, which have far better high current characteristics. I re-used the HGTG30N60A4D’s that left over from my DRSSTC work.

I continued testing with IGBT’s on low power first, so I can take some measurements safely. This is where I started realizing that 12V fans used to force air through heatsinks slow down to a crawl while the coil is running! They were obviously being affected by the coil – at the time I suspected that it is due to changing electric field from the heatsink, which is live with RF voltage. But then my attention was drawn towards weird burning smells that started emanating from the coil. While trying to locate their sources during following runs, I witnessed the zip-ties, that used to hold the primary, start popping off the uppermost primary turn! At first I was convinced the copper pipe simply got too hot and melted them. But the other day I got the zip ties fall off again almost immediately after I started running the coil. Runs were low power and the primary stayed completely cold! I immediately started to suspect that the ties don’t like the electric field that forms between the wooden ring that I put over my primary, and the primary coil itself.

<to be continued>

 

Results

A video of a full power run, about 10kW. Overcurrent protection circuit was included, which can be seen acting once during the video.

Links and references

4hv forum thread http://4hv.org/e107_plugins/forum/forum_viewtopic.php?95545

 

Leave a Comment

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: