Dual Resonant Solid State Tesla Coil

Up until about 2005, the Dual Resonant (or Double Resonant) SSTC was a rarity. However, thanks to the hard work of Steve Ward, Steve Conner, Terry Fritz, Dan McCauley and others, many coilers are now building DRSSTCs. The DRSSTC works much like the original SGTC in terms of streamer appearance and in that it has a primary tank capacitor. DRSSTC's are more efficient than SGTCs, and completely controllable right down to streamer length, appearance, and sound. The DRSSTC can even be audio-modulated.

Overview

A DRSSTC is driven most often by a H-Bridge (or half-bridge) of IGBTs or MOSFETs, but some SCR models have been experimented with. SCRs are not often used because they are very hard to control and are generally too slow to handle TC frequencies.

It should be noted that the above losses are only the conduction losses (that is, losses generated from simply carrying current). Switching losses, which are caused by the instantaneous overlap of voltage and current, can also play a large factor. Typically MOSFETs have lower switching losses because they can switch faster, but DRSSTCs make use of what is called Zero Current Switching (ZCS). The primary current is a sinusoid, so it is ideal to have the devices switch when the primary current is crossing zero. Ideally the switching loss is zero (because the current is zero), but in reality, you never switch at exactly zero. In the case with DRSSTCs the switching losses are pretty low because of the low duty cycle, so IGBTs are still the switch of choice. MOSFETs could be used, but the power dissipation must be carefully considered.

How does it work?

A DRSSTC is different than the conventional SSTC due to the addition of a primary tank capacitor, hence, "Dual Resonant." When in a resonant state, the added capacitance in the primary circuit cancels the inductance leaving no reactive component. Primary current flow is now only limited by resistance in the capacitors (ESR) and resistance in the primary windings, which is usually on the order of a few hundred milliohms. In a SSTC, primary current is limited by the primary's inductance and streamer loading. This is the reason a SSTC can run CW or "continuous wave". So a key difference between SSTCs and DRSSTCs is that a DRSSTC usually operates in the transient state, while SSTCs can safely run in the steady state conditions. However, if you were to drive a DRSSTC at it's resonant frequency for too long, a steady state current of many times the safe limit might be possible. This could have the following adverse effects

• IGBTs blow from overcurrent - this would happen almost instantanly (few mS) depending on the size of the IGBT.
• Overvoltage of the primary capacitor - The extreme current flowing over the primary tank can create voltages tens of times more than the supply voltage. (use Ohm's law to figure out just how high)

Therefore we use an interrupter. An interrupter turns on the drive circuits for a determined amount of time (typically 50-300uS), and then shuts them off (typically 2-20mS). This low duty cycle keeps the IGBTs from overheating severely. Active overcurrent detection (OCD) circuits can arrest high currents that might develop before the interrupter shuts off. A common reason for sudden current draw would be a ground strike, which dramatically changes the impedance of the tank circuit (it increases the Q).

So what actually happens during a a single "burst" from a DRSSTC: It all starts off when the interrupter output says "go". When this happens, the bridge (either half-bridge or full H-bridge) turns on IGBTs in a manner that places the supply voltage across the tank circuit. This "step" input into the system results in a sinusoidal primary current, at a frequency determined by the LC combination. The first half-cycle of primary current peaks to a level determined by the surge impedance of the circuit (more on this later). Just as the primary current is about to reverse (that is, go "negative") we switch the output voltage from the bridge so as to add more energy into the tank. Determining when to switch can be done using feedback of the primary current. A current transformer can sense the polarity of the primary current, and use this information to determind what the output polarity of the bridge should be. It is important that you have the appropriate phasing of the feedback, or else the bridge will work to destroy resonance (not support it). So, as energy is continually added to the primary on each cycle, some energy is transfered to the secondary due to magnetic coupling. Eventually the secondary accumulates enough energy to charge its toroid capacitance to a suitable voltage to achieve a streamer. When this happens, the Q (the ability to store energy) of the tank circuit drops quickly as all of the energy is spent in the streamer (making it longer and hotter). Typically we only need to drive the streamer for about 3-4 RF cycles (the reason for this is beyond the scope of this article) to achieve its maximum length. So by that point, the interrupter says "stop!". The output voltage from the bridge goes to zero. If there is any energy left in the tank circuit at this point, the diodes that are anti-parallel to the IGBTs form a bridge rectifier, and the left over energy is recycled back to the main filter capacitors.

Controllers

Steve Ward built the first easy-to-build and quite reliable controller. Other controllers include Steve Conner's PLL driver.

Some basic design equations and guidelines

• Firstly, keep your resonant frequency low! IGBTs are getting faster, but most of them are still pretty slow (at least compared to MOSFETs). Keeping the resonant frequency low, allows for better zero current switching. For secondaries 3-4" diameter, 28-34awg magnet wire is acceptable. For larger 4-8" coils, 26-32awg wire is better. For 8-12" coils, 22-28awg is about right. Generally, around 2000-2500 turns is acceptable as an upper limit. The benefits of lower resonant frequency seems to outweigh the losses due to lower Q factor in the secondary. Also, large toploads seem to work well for DRSSTCs (typically the toroid diameter = the coil's height).
• After you determine how big you want your coil to be, and have found the resonant frequency of your secondary you can determine the primary circuit parameters. The surge impedance of the primary should generally be within 5-20 ohms.
$Z_{surge} = \sqrt\frac{L}{C}$

Generally, a lower Z will run higher peak currents and require fewer driving cycles, while a higher Z runs lower peak current, but for more cycles. From this equation you can determine the values of L and C for your primary circuit (once you choose Z). This equation also tells us the inductive reactance (Xl) and capacitive reactance (Xc) of the circuit (note, Xl = Xc). This is useful later.

• There is no exact way to know the primary current before the coil is built and tested, but generally the IGBTs are the limiting factor. Small TO-247 type IGBTs have been run up to around 400Apk reliably (but this will vary with IGBT type). At 400Vin, a half-bridge would be good for about 30" sparks. A full bridge would likely run about 48" sparks. The next common size up is the SOT-227 package, which have been run at about 1000Apk reliably. At 400Vin, half-bridge would probably generate 48" sparks, while a full bridge up to 80" sparks. Brick sized IGBTs seem to be fine running 2-3X their peak ratings. My largest DRSSTC required 1500A (at 650VDC input) to generate 12' sparks. Again, these are very rough estimates.
• With a rough idea of what currents may be present in the tank circuit, we can determine the tank capacitor ratings. First the voltage rating can be found with ohms law (V=IZ), where V is the tank cap voltage, I is the peak primary current (listed above), and Z is the surge impedance (also calculated above). So for a tank circuit of 10 ohms surge impedance, running 1000A, the tank cap should have at least a 10kV rating. The peak current rating for the cap is fairly straight forward, and needs to be at least that of the peak tank current (generally, the peak current requirement is easily achieved). The largest issue is the RMS currents. Below is a formula i derived that gives a rough estimate of the RMS current in the tank circuit:
$I_{RMS} = \sqrt{\frac{ONtime}{OFFtime}\times.7\times{I_{pk}}^2}$

The .7 factor is to roughly account for the envelope shape of the primary current. Anyway, this formula gets you pretty close to the expected RMS current in the primary. Your MMC capacitor should meet this current rating if you want to run the coil for any long period of time. My 6' spark coil runs about 65A RMS at 120BPS. My 12' spark coil runs about 160A RMS at 120BPS. It should be noted that doubling the break rate increases the RMS current by 1.414X, not by 2X, but the input power does increase by 2X. With such high RMS currents, it is clear why resistance must be kept as low as possible!

Waveforms

Primary Current Waveforms in relation to interrupter time

Interrupter set to 50uS
Interrupter set to 100uS
Interrupter set to 250uS
Interrupter set to 500uS
Interrupter set to 1mS
Waveform of a GDT

Note: These images were erroneously labeled in kV instead of kA. However, they are still accurate if one substitutes the labels.

• Do do everything possible to reduce primary resistance. Primary resistance can be so harmful that it can make your DRSSTC about as efficient as a typical spark gap coil. DRSSTCs have higher RMS currents than spark gap coils, so resistive losses are worse for DRSSTCs.
• Do minimize bus inductance (that is, the inductance between the main filter capacitors, and the IGBTs). Often, a laminated bus structure is the best way to go.
• Do use more capacitance in the primary tank if possible. (Performance often increases, but IGBT's suffer more current.) The idea here is to increase the rate of rise in primary current (by lowering the surge impedance), which in turn reduces the amount of time you need to drive the coil to achieve long sparks. Efficiency could be hurt if wiring resistance is not minimized.
• Do use high coupling, but nothing over .2. The higher the coupling the better the performance per IGBT current. However, higher coupling introduces a huge increase in primary-secondary arcover. Originally, the idea was to have coupling as high as possible, but now we realize performance increases with a lower coupling because the coil can generally be pushed harder. However, again current is higher in the IGBTs.
• Do not exceed 2/3 of the IGBT Vce breakdown rating for your Vin. Older IGBT bricks may need more derating (no more than 350/700V for 600V/1200V devices) due to their larger internal inductance which causes more severe voltage transients.
• Do not allow the output streamers to hit the primary or electronics. Use of strike rails, and metal enclosures is advised.

Commonly Used Components

IGBT's

Fairchild Distributors

• HGT1N40N60A4D - The favorite, good up to 1200A if you know what you're doing. In fact I've heard of them pushing 1700A. Unfortunately they are not made anymore.
• HGT1N40N60A4 - The TO-247 version of the above IGBT, still available, thermal dissipation is not as good.
• FGA40N60UFD - Cheapo version of the above IGBT, still available. I've personally put 200A (sparklength 27") through this IGBT, retail $3.90. • FGH30N6S2 - Same performance as FGA40N60UFD above, still available. Retail$3.30.

Other Distributors

• IRG4PC50UD - Best bang for buck. MSRP is $13, but Arrow electronics sells them for$3.90. Confirmed sparklength of at least 5'.

Capacitors

• CDE 942C20P15K - Everybody uses these. These are the .15uF, 2000V version, but any in this line should work. This capacitor is cheapest per value in this line. Approx $4 apiece if bought individually, can be bought from http://www.rell.com at$2.58 (min order of 52). The new "lead free" ones work just fine as well. The 940 series capacitors work as well, but are not as good.
• GE 42L3332 - Another popular capacitor. Some don't like them, but Daniel McCauley swears by them. Others have had no problems as well.

Note: Others have been known to work. When choosing a capacitor, the most important things to consider are:

• That it is a film/foil (not film or foil) capacitor
• Pulse current should be high
• ESR, ESL should be low. ESR moreso.
• MMC capacitors most often fail where the leads attach to the plates.

However, some still have these features and still fail. Mostly due to construction.

Statistics/Facts

• The DRSSTC's invention is credited to Jimmy Hynes.
• Steve Ward holds the world record for DRSSTC sparklength: twelve feet. This distance was achieved with his DRSSTC-2