1. Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.
    Dismiss Notice

Building a DRSSTC Pt. 6 - Driver Design

Blog entry posted in 'Building a Dual-Resonant Solid State Tesla Coil', Jul 29, 2014.

5.00/5, 1 vote
Hi everyone, and welcome back to my blog!

Now that we have discussed the Tesla coil primary circuit, secondary circuit, and H-bridge, it's time to take a look at what we'll use to drive the IGBTs. If you'll recall from the last entry, we have to switch opposite transistors at just the right time so as to send a pulse of current into the primary tank circuit.

In order to switch the transistors, it's important to remember that the gate-emitter voltage on the IGBTs is what determines whether or not they are turned on. Since the collectors of the top two transistors are connected to 170VDC and the collectors of the lower IGBTs are connected to the emitters of the upper ones, we would have to have a fairly high voltage on the gates in order for them to switch on. However, here's a better idea: Use an isolated source connected directly across the gate and the emitter. We do this by using an isolation transformer commonly referred to as a "gate drive transformer", or "GDT". By sending a signal in on one winding (the primary), we can get the same (or very similar) signal on multiple secondaries which we can use to switch all of the transistors right when we need to.

Since we have 4 IGBTs, we'll need 4 secondaries--one winding to drive each gate--and one primary winding for the signal. In full-bridge designs, the preferred method is to use a pentifilar (5-winding) transformer, most commonly wound on a toroidal ferrite core. Now, before you just pick up a toroid and start winding, we need to explore why not all toroids will work.

Many toroidal cores are designed for chokes and are absolutely lousy at higher frequencies, as in GDT applications. This was my first mistake.

When digging through my junk bin I found the following core, which I had removed from an old ATX computer PSU:


I then removed the sheath from an Ethernet cable, which exposed four neatly-twisted pairs of solid copper wire. I figured this would be ideal, because I could wind multiple secondaries at once with little effort. Something else to note is that two of the secondaries should be in-phase with the primary, and two should be out-of-phase, which causes two transistors to be switched on while the other two are off, based on the input to the primary. I figured it would be easy to wind one pair one direction and another pair in the other direction to give me the proper phasing. So I did just that, and wound the primary and secondaries in sections, so it was easy to differentiate between the windings. When I was done, it looked like this:


However, when I tested it using my oscilloscope (I applied a 273kHz 12v square wave on the primary), my output, which should have resembled the input, was pretty much nil, aside from some significant ringing. I decided that I really needed to do some more research before winding my GDT.

When sourcing a toroidal core, there are many factors you need to take into consideration. In general, you want to have a very high permeability, otherwise the core will have significant losses at high (Tesla coil) frequencies. I eventually ordered four cores from Digikey (part number 495-3868-ND) which (if I remember correctly) have a permeability of 10,000. Generally you'll want an AL value as close to this value or higher for a Tesla coil GDT, otherwise it will have very poor performance.


I re-wound the GDT on one of the new cores, but this time I did not wind it in sections. In order to ensure best performance, I made sure the windings were fairly evenly-spaced. Once again, I used Ethernet cable wires, two of which were wound to give me phasing opposite that of the primary.


Looks much better, doesn't it? Time to run it at a high frequency to test it out:


The above waveforms are as follows:

Topmost square wave - Primary. 273kHz square wave supplied by an Agilent function generator
Middle square wave - Secondary 1. Out of phase with the primary, which is why the signal is inverted
Bottom square wave - Secondary 3. In phase with the primary

Secondary 2 is twisted and wound together with Secondary 1, and Secondary 4 with Secondary 3. All windings (the primary and the 4 secondaries) all have a 1:1:1:1:1 ratio, meaning the voltage and current on each should be the same. Generally the number of turns for each winding ranges from between 8 and 16, though you may need to experiment a bit in order to find the optimal number for your coil. You should use as few windings as possible without saturating your transformer.

So that's it for the GDT, and the first look at the actual build so far. We'll be getting to more posts regarding the actual build in a little while.

Before I end this entry I'd like to take a quick look at the circuitry used to send the signal to the primary of the GDT. This circuitry has to do two things:
1) Take the feedback signal from the Tesla coil (we'll get to the feedback circuit next time) and send it back to the GDT
2) Turn allow or disallow the feedback signal from passing through based on the interrupter state (HIGH or LOW).

In order to drive the GDT, you'll need two MOSFET drivers--one inverting and one non-inverting. These are effectively amplifiers that can supply sufficient current to the transformer to drive the IGBT gates. I will be using two Texas Instruments UCC27425 4A drivers in parallel. Ideally (if I had thought ahead) I would have purchased a single driver rated for a higher current, as these devices will likely heat up when running at over 200kHz! But two in parallel should work fine.

The UCC27425 is ideal for our application because it contains the inverting and non-inverting driver in a single package, and even has an "enable" pin, which means we can directly control it (turn it on or off) with the interrupter.

The following animation shows how high/low signals on each input (interrupter and feedback) affect the signal sent to the primary of the GDT.


I realize the animation is a bit "busy", so to speak, but hopefully this brief explanation will clear things up:

When the interrupter signal is high, the chip is enabled. Therefore, it passes the Feedback signal through, which drives the GDT. This ensures high efficiency, and that the primary and secondary can resonate properly. Notice that both inputs are connected together, so each driver output will always have an inverted logic signal compared to the other. This ensures that we can get an alternating current on the primary of the GDT.

When the Interrupter signal is low, it disables the chip, and no signal is allowed to pass through. This means the IGBTs don't switch when the interrupter is low, which ensures the Tesla coil remains off whenever you want it to be off.

This is ideal when you want to prevent the coil from overheating, or if you want to apply audio modulation to the Tesla coil. By switching the interrupter on and off at a frequency matching that of a note (for example, 440 Hz for an A4 note), then the Tesla coil will sound like it is playing an A4 note.

We will discuss the interrupter more in depth in one of the future posts, but for now hopefully you have a general idea of how the H-bridge driver operates, and how it responds to different input signals.

That concludes today's post. I hope you've enjoyed it, and as always, feel free to leave any comments or questions!
magvitron likes this.
  1. kimo3010
    can you please re upload the GIF .. thank you
    kimo3010, Feb 18, 2016
  2. DerStrom8
    Hello there kimo3010, I'll see what I can do. I was afraid this might happen....
    Thanks for pointing it out!
    DerStrom8, Feb 19, 2016
  3. DerStrom8
    I'm afraid I don't seem to have a copy of it on my computer or any of my external hard drives. I will continue to search but I will most likely need to create it all over again from scratch. This may take some time.
    DerStrom8, Feb 20, 2016