Skip to content

DRSSTC1

Introduction

When I first learned about solid state Tesla coils, they seemed like expensive, power hungry machines that only produce puny little sparks. Then one day, I ran onto Steve Ward’s website and his amazing DRSSTC, one of the first in the world! Seeing those sparks that looked so huge in proportion to the coil, produced by IGBT’s – a new and enigmatic component for me at that time. I immediately knew I want to built a Tesla coil like this myself, regardless of how hard it might be.

A DRSSTC stands for Dual Resonant Solid State Tesla Coil, which differs from an ordinary SSTC in having an additional resonant capacitor in series with the primary coil, which cancels out it’s inductive reactance and allows the inverter to feed huge peak powers into the system. In order to prevent the inverter from blowing up, this condition is only allowed to last for about 100-200  microseconds, during which the resonant primary current may peak in kiloamperes. The drive is shut off after that, with bursts like this repeating at a rate of few hundred hertz – this disruptive mode of operation with high peak powers allows for growth of sparks to great lengths, with appearance and sound very similar to SGTC sparks. The firing rate and pulse length is controlled by a special modulator circuit called an interrupter. A lot more information about DRSSTC’s is available on Steve Ward’s site and 4hv forum.

Project description

I started this project about 7 years ago, when I just began high school. Due to my lack of experience I decided my best shot for success would be to rely on Steve Ward’s design. The controller schematic that was widely used during that time is shown below:

DRSSTC controller schematic, courtesy of Steve Ward

DRSSTC controller schematic, courtesy of Steve Ward

The basic idea behind this controller is primary current feedback – the primary signal is measured by a current transformer, and clipped by zeners to approximate square wave, and this signal is fed with positive feedback into gate drive chips and ultimately used to control the IGBT bridge. The positive feedback causes the system to become unstable and oscillate, with current growing in amplitude rapidly – until either turned off by the falling edge of interrupter pulse, or by the overcurrent detection circuit constructed around U6 and U7. This circuit, which senses the primary current amplitude with another current transformer, is there to prevent destruction of the inverter from the over-current in case the interrupter signal becomes corrupted, or during events like heavy ground arcs. It turns off and then re-enables the drive after a set period of time (a few milliseconds) set by the NE555 monostable circuit. The trigger current treshold is controlled by a trimmer resistor on LM311 input. The flip-flop U2 is there to synchronize the turn-off of the igbt’s with the falling edge of the input signal: since the interrupter or OCD may turn off at any time, they could make the IGBT’s turn off heavy current during the peak of the cycle, resulting in high turn off losses and possible destruction of the IGBT’s.

Driver board construction

My own implementation of this circuit on a prototype board is shown on the picture below:

My first ever DRSSTC driver board - photographed recently because I couldn't find any pictures from the time I constructed it. The empty sockets used to house UCC3732x gate driver chips, which were re-used in other projects

My first ever DRSSTC driver board – photographed recently. The empty sockets used to house UCC3732x gate driver chips, which were re-used in other projects

Note how I used a yellow powdered iron core in a common mode line filter – in front of a proper ferrite choke – clearly I didn’t have too much idea of what I was doing at the time!

The circuit lacked OCD at first, though it was added through subsequent upgrades. The interrupter was designed into a separate handheld enclosure:

Interrupter based on a single NE555, constructed on a small piece of prototype PCB

Interrupter based on a single NE555, constructed on a small piece of prototype PCB

Interrupter enclosure closed

Interrupter enclosure closed

Unlike the Steve Ward’s schematic, my interrupter was powered by an additional 15V supply line coming from the Tesla coil. The trimmer resistor used to set the ON time could only be accessed by a screwdriver, because I was afraid that a potentiometer could be accidentally turned to a too long setting and blow the coil up – especially true for initial tests with no OCD!

Today, I wouldn’t recommend an interrupter design like this to anyone. An example of an improved circuit is shown in my mini DRSSTC project, though I still wouldn’t recommend the practice of interrupter not being electrically isolated from the coil due to several reasons:

  • When humans are standing near large Tesla coils in operation, high voltages get induced on them by large changing electric fields
  • Holding a plastic interrupter box in hand may cause significant displacement currents flowing into the circuitry and disrupting it’s operation, as I had it happen in this project.
  • A box can be made of metal, but in that case touching any part of the box may produce an unpleasant electric shock. Some people still actually prefer this, because holding the box entire time during coil operation allows wide surface area for current passage and generally no shocks are felt; high frequency current only actually produces “shock” due to localized pinpoint heating on the site of contact.
  • Another possibility is to use a plastic box that is shielded from inside with adhesive metal foil tape. This would both protect the circuitry and prevent shocks, provided there are no metallic protrusions on the box that could create shocks if touched, like switches. Those may be left floating, but then they would again act as a path for the noise to get into the circuitry.
  • Some people prefer to use an opto-coupler chip to insulate their interrupter electrically. However, I’m not sure whether I would support this practice for larger coils, since induced voltages can quite easily surpass the opto-coupler rating and destroy it.

The only truly bullet-proof solution is to use fiber optics, which is today practiced by most expert coilers for running large DRSSTC’s. However, those tend to get expensive, and popular HFBRx IC’s that could transmit DC signal are also becoming obsolete, leaving the interrupter link issue still somewhat open-ended. I will discuss this more along with my future digital TC driver implementations!

First power section and coil component attempts

While the control section was replicated quite successfully, my first attempts at power electronics section didn’t share the same luck. I at first wound a secondary with a rather high resonant frequency of 400kHz, and was subsequently talked into making another one for <150kHz in order to minimize switching losses in the IGBT’s. The high frequency secondary was ultimately re-used for SSTC1.

Some basic parts that are about to become a Tesla coil

Some basic parts that are about to become a Tesla coil

Winding jig used to wind a secondary, which later proved to have way too high resonant frequency

Winding jig used to wind a secondary, which later proved to have way too high resonant frequency

Secondary, toroid and a box assembled. Do not a box made of ferromagnetic material this close to your secondary base!

Secondary, toroid and a box assembled. Do not a box made of ferromagnetic material this close to your secondary base!

Toroid closeup

Toroid closeup

Due to cost and difficulty of acquiring IGBT’s in Croatia, I opted for a half-bridge for now instead of a full bridge. Zeners are used to protect the gates and networks of parallel schottky diodes and 3.9 ohm resistors are used on each gate to accelerate turn-off, and provide some dead time for the devices. Note that in these primitive drive circuits, there is no other hardware mechanism for generating dead-time. I was actually long puzzled how cross-conduction isn’t a bigger problem in SSTC’s – only in recent times I have started to realize that the Miller effect may be playing a major role.

Transient voltage suppressors (TVS) were used across every device in order to block any voltage spikes that may occur during switching. Note that this is a dangerous way to use suppressors, because under extreme currents the IGBT reverse diodes may develop a voltage drop that is higher than TVS forward voltage, causing them to conduct and blow up! For this purpose, always use either bidirectional TVS or at least add an anti-series schottky diode to prevent forward conduction!

Half bridge schematic

Half bridge schematic

8x HGTG30N60A4D fast IGBT with internal diode

9x HGTG30N60A4D fast IGBT with internal diode

I also had trouble acquiring proper energy storage capacitors for this coil. Generally people use large screw terminal style capacitors with high current connections, but I had to be satisfied with a pair of snap-in capacitors shown in the picture. They still hold over 150J togehter which should be enough for the coil of this size, which may have bang energy of 10J or so.

Energy storage capacitors, 2x 1000uF 400V

Energy storage capacitors, 2x 1000uF 400V

The DRSSTC primary circuit endures high peak and RMS primary currents which are hard on the tank capacitor. These high currents generally also imply high resonant voltage rise, up into kilovolt range. The common solution is to use many small capacitors in a series-parallel configuration, so called multi-mini capacitor or MMC. The capacitors need to have both low losses at high frequencies, and endure high peak currents which is best satisfied by polypropylene film + foil capacitors, such as are WIMA FKP1 or Cornell Dubilier 942C series. This kind of capacitor has a 2-stage construction, with a pair of metal foil electrodes and a floating metalized film electrode in between. The floating electrode is present over almost entire length of the capacitor, while the foil ones are in less than 1/2 each. This effectively produces two capacitors in series. The foil provides good high current contact to capacitor terminals, and the metalized film electrode gives the capacitor much valued self healing property, because the thin metal film easily vaporizes around the site of failure. Some people have reported good results with fully metal-film caps, though they are known to fail at the electrode-terminal connections and I would recommend against them. But, I knew nothing about this at the time and bought polyester film capacitors that have lossy dielectric and are poorly suited for high currents. I’m quite certain this was a main cause of the initial poor performance of the coil.

A multi-mini capacitor of 8 series 400V 1uF polyester film capacitors. Do not use polyester, nor any other type of capacitor other than film+foil for your DRSSTC's!

A multi-mini capacitor of 8 series 400V 1uF polyester film capacitors. Do not use polyester, nor any other type of capacitor other than polypropylene film+foil for your DRSSTC’s!

The primary coil construction was also quite dodgy, but it was OK compared to the blunder I made with the resonant capacitor. In any case I’d still recommend to use copper tube instead of relatively thin solid wire like I did in this occasion! I used four PVC pipes as supports, which were filled with polyester resin and fastened with one screw each to the wooden base.

Primary coil construction

Primary coil construction

If I recall correctly, I achieved my first light with 400kHz secondary, but blew the igbt’s aftera while. The new secondary with a lower resonant frequency was then put it:

The whole test setup. A single IGBT was mounted beneath each PC CPU heatsink visible under the coil base

The whole test setup. A single IGBT was mounted beneath each PC CPU heatsink visible under the coil base

The driver PCB wired up for a test

The driver PCB wired up for a test

Everything crammed together into a PC power supply box. The ill-attempted MMC is in the upper right corner

Everything crammed together into a PC power supply box. The ill-attempted MMC is in the upper right corner

If I recall correctly, I couldn’t make the system work with primary current feedback  – I suspect the culprit might again have been the bad capacitors. I used secondary base current feedback which is not a very good approach for a DRSSTC, since it doesn’t guarantee zero current switching for the inverter. Nevertheless, I succeeded producing some sparks, about 35cm in length, which I found quite exciting at the time!

Perhaps the longest spark generated by DRSSTC, before the IGBT's died

Perhaps the longest spark generated by DRSSTC, before the IGBT’s died

Soon after that, the IGBT’s gave up their life for the second time, and I suspended the project after realizing how many pitfalls have I gotten in. I have used very long ON times on the interrupter in order to reach even this spark length, and as it can be seen from the oscilloscope shot, the primary current would settle virtually to a stationary value by the end of the bang! Now this looks like an indicator of poor tuning: the current couldn’t grow because the primary circuit was way off resonance.

Oscilloscope capture of primary current and drive signal waveform. It can be seen that ON time used was exceptionally long

Oscilloscope capture of primary current and drive signal waveform. It can be seen that ON time used was exceptionally long

Since these experiments are about 7 years old, I can no longer be sure about the exact cause of poor performance and IGBT failures – but I suspect the culprit may lie in combination of several causes:

  • I have used a lossy primary tank capacitor, which might have robbed a portion of power from the output
  • The use of secondary current feedback has likely prevented ZCS operation of the inverter, resulting in increased losses
  • Improper tuning has resulted in reduced peak power and made me require excessive ON times to produce significant sparks
  • The lack of OCD may have resulted in excessive current ringup and IGBT death, once I happened to tune up the primary into resonance.

For the end, I’m going to take an opportunity to summarize all mistakes I made in this stage of the project:

  • Bad choice of secondary resonant frequency. Try to have the resonant frequency under 100kHz for discrete IGBT’s, and perhaps even under 50kHz for coils based on slower IGBT modules (bricks).
  • Bad choice of primary resonant capacitor
  • Mounting the coil atop a ferromagnetic box
  • Not being careful about the possible forward conduction of transient voltage suppressors
  • Failing to shield the interrupter, causing it to be susceptible to noise
  • Too thin primary conductor
  • Too small toroid for a DRSSTC
  • Poorly insulated gate drive transformer, susceptible to breakdown.

The coil was ultimately dismantled and parts shelved until I felt ready to take on another attempt.

Revival – DRSSTC v1.1

About a year later started a new, fresh DRSSTC construction. I re-used many old parts, though the mechanical construction was changed and I managed to fix majority of my previous mistakes.

A new, more suitably sized toroid

A new, more suitably sized toroid

Making primary supports in acrylic. After cutting they are glued in a manner that allows simultaneous drilling of holes

Making primary supports in acrylic. After cutting they are glued in a manner that allows simultaneous drilling of holes

After drilling, the supports are hot-wire bent and copper tube is slowly and carefully wound through. Finally the supports are bolted to the base.

After drilling, the supports are hot-wire bent and copper tube is slowly and carefully wound through. Finally the supports are bolted to the base.

The new MMC is made using trusty 942C20P15K capacitors

The new MMC is made using trusty 942C20P15K capacitors

The full bridge PCB was etched by covering it fully with lacquer, and then scraping away the sites where copper is unwanted - a rather slow and painful process!

The full bridge PCB was etched by covering it fully with lacquer, and then scraping away the sites where copper is unwanted – a rather slow and painful process!

Four IGBT's soldered into their positions. Insulating pads were again required to insulate them against the heatsinks, and were again problematic to install!

Four IGBT’s soldered into their positions. Insulating pads were again required to insulate them against the heatsinks, and were again problematic to install! The double sided PCB attempts to minimize parasitic inductance by using one side for positive, and another for negative supply rail

The finished H-bridge in all of it's glory. Note the large amounts of decoupling capacitors, which were re-used from my previous MMC attempts!

The finished H-bridge in all of it’s glory. Note the large amounts of decoupling capacitors, which were re-used from my previous MMC attempts! Two sets of current transformers are visible, that provide signals for current feedback and limiting.

Backside of the bridge, bleeder resistors visible

Backside of the bridge, bleeder resistors visible

plywood support for the control board

plywood support for the control board

Mountining  into case

Mountining into case

control board fitted

control board fitted

Despite all the improvements, my bridge design still wasn’t the very best. I’ve left very little spacing between the traces between which considerable potentials could appear, and the insulation between IGBT’s and the heatsinks was still hard to place properly. I had to glue the sil-pads to the heatsinks in order to stop them from moving during mounting!

I hoped to find a larger filter capacitor this time (which I left the space for) but in the end I had to put back the two 1000uF 400V caps I used before. In short time everything was ready, and the new version of the DRSSTC was ready for it’s first light !

First light setup

First light setup

First light v1.1

I brought the setup to the garage and proceeded to tune it while gradually increasing the power level. I found out that I need a few more primary turns to reach a good tune, which I hacked up with some heavy duty cable. When I was satisfied, I cranked the power up to full and witnessed some wild, roaring sparks that stretched nearly a meter in length!

I only managed to take one picture though, that didn’t turn out too well due to long exposure and inadequate lighting. But it clearly shows appearance of flashovers between primary and secondary, which are the reason I had to stop my runs this time!

O_O

First credible DRSSTC run

The other day, I raised my secondary slightly and gave the coil another go. This time I brought an oscilloscope and a current transformer to monitor the primary current. I could observe the primary current rapidly peaking the 600A OCD limit, which was not a good news. I didn’t expect the IGBT’s to handle much more, since this was already over 2x their pulse rating! And then, upon closer observation, another unpleasant surprise on the bridge which I later managed to photograph:

Arcing between H bridge traces

Arcing between H bridge traces

I never fully determined what caused this problem. The arcing seemed to occur between one DC bus rail and CT output, which left me puzzled because I would expect this to result in a large explosion – but somehow, the arcs remained small and the coil continued to work well despite their occurence! For a while, at least. While I was trying to find the source of these arcs, the coil at once went completely silent. Thinking I blew an IGBT, for which I had no replacements, the project was mothballed for several months. Later I found out that all IGBT’s are actually OK, and that culprit was some unknown component on the control board. This was, sadly, pretty much the end of my experimentation with this coil. I again summed up the mistakes made in the project, and focused onto correcting them in another project – a small, portable DRSSTC that I could easily transport around for demos. The short resume of problems –

  • I guess the inter-trace arcs on the bridge speak for themselves.
  • I also started realizing that smallish 30N60 IGBT’s aren’t really up for the task of producing meter range sparks without over-stressing them extremely. Running IGBT’s at multiples of their transient SOA was a trend when DRSSTC’s were first invented, but rapidly fell out of fashion due to piles of blown devices that apparently started to accumulate in people’s workshops. Hence I decided I want to upgrade this coil with a completely different engine, based on CM300 IGBT modules.
  • The characteristic impedance of my tank circuit was extremely low, resulting in very rapid rise of primary current, that led to flashovers and also unnecessarily over-stressed the IGBT’s. This means a smaller capacitance, higher voltage MMC, and a higher inductance primary to begin with.

That would be it for now: I have bean working on the CM300 DRSSTC driver, which may as well get a whole new page (DRSSTC 2.0), and will post some proggress as soon as time lets me. Thanks for reading!

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: