Good enough for brick IGBT gate drive ~160khz? / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Good enough for brick IGBT gate drive ~160khz?

1 2 next

Move Thread

LAN_403

Arcstarter

Fri Aug 07 2009, 09:08PM

Registered Member #1225 Joined: Sat Jan 12 2008, 01:24AM
Location: Beaumont, Texas, USA
Posts: 2253

Hey guys, i need a nice powerful gate drive for my IGBT brick, and i am going to run the gate at something like 24-30v (unless told otherwise). I will be pushing it to 160khz, and it is for a DRSSTC halfbridge, so a big gate drive is needed.

I have few components, and have a budget of about nothing, so i decided to use a couple mosfets, with an isolated drive signal (GDT) for safety. This schematic was drawn in a couple minutes, so if needed i can redraw it if it confuses anyone (i understand, but that is because i am the one who drew it). It is pretty much what Steve Ward used (it was taken from his brick driver), so i am fairly confident it will work.

For the power source i would probably make a seperate smps with around 30 volts output driven straight from the mains and a uc3842 to drive it. Or i could just use iron cored transformers, but that would be a bit bigger, and less of a challenge.

Thanks alot guys!

hvguy

Sat Aug 08 2009, 04:57AM

Registered Member #289 Joined: Mon Mar 06 2006, 10:45AM
Location: Conroe, TX
Posts: 154

Looks good, but why not place the GDT after the circuit? That way you avoid the isolated PS issues, among others... Also, 24V is fine for the gates since your driver is so low impedance.

Arcstarter

Tue Aug 11 2009, 04:34AM

Registered Member #1225 Joined: Sat Jan 12 2008, 01:24AM
Location: Beaumont, Texas, USA
Posts: 2253

Well, i do not have a GDT that big for one thing. I mean, bricks have such big gate capacitance. I am not even sure if this brick will like 160khz, but as i have been told Semikron bricks tend to be faster.

Also, the 24-30 volt PSU would likely be switch mode, maybe some sort of inverter (royer is another thought because of simplicity, or maybe flyback, but i would rather not) so a high speed diode bridge and cap is all that is needed for a fully regulated though high power supply (if i do in fact use the switch mode). That would be hard to kill. So isolation is not too big of a worry.

hvguy

Tue Aug 11 2009, 05:32AM

Registered Member #289 Joined: Mon Mar 06 2006, 10:45AM
Location: Conroe, TX
Posts: 154

You don't need a huge core. Something around .75-1in would be fine if you wind it right and the material is good. You will need two circuits, two power supplies(or a dual secondary), and still need an intermediate gate driver to drive them. A single GDT with two secondaries is much simpler to implement, and, as I'm sure you know, simpler=more reliable. It also means less delay and higher isolation... That being said the high side gate driver can be a good learning experience if you have not built one before and the royer circuit works well as a simple ps for it if you chose to go that route.

HV Enthusiast

Tue Aug 11 2009, 11:59AM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

hvguy wrote ...

Looks good, but why not place the GDT after the circuit? That way you avoid the isolated PS issues, among others... Also, 24V is fine for the gates since your driver is so low impedance.

Because switching power and high current through transformers is very difficult and leakage inductance becomes a major issue.

The best way for larger bricks is simply to switch the "signal" through the transformer for isolation, and then have your power stage on the high side with the IGBT brick.

Here is a very simple gate driver i designed for CM600 bricks. (I also sell the boards for these on my website) Doesn't even require an isolated switching power supply - just a linear supply off a 60Hz transformer. Very simple.

Dan

hvguy

Tue Aug 11 2009, 06:56PM

Registered Member #289 Joined: Mon Mar 06 2006, 10:45AM
Location: Conroe, TX
Posts: 154

Yes, the transformer design does require more attention, but it saves you an isolated power supply (SM or Linear) for each IGBT (three for a full bridge), and at least two dozen components. It also has the added benefit of galvanic isolation between any sensitive silicon (30V GDICs) and the HV rails, which means your IGBT can fail without damaging any of the driver. I'm not saying it doesn't work, I've used both, but the transformer greatly simplifies the design. Steve W. and Finn are good examples of this. Both tried the discrete drivers for there large coils but ended up using GDTs instead. For someone on a budget and little experience with DR coils I feel like the discrete drive is an unnecessary complication, though it could be a good learning process.

Of course, the process could be made easier if you bought a few of those nice PCBs Dan is offering for gate drive

Arcstarter

Tue Aug 11 2009, 08:10PM

Registered Member #1225 Joined: Sat Jan 12 2008, 01:24AM
Location: Beaumont, Texas, USA
Posts: 2253

Because switching power and high current through transformers is very difficult and leakage inductance becomes a major issue.

The best way for larger bricks is simply to switch the "signal" through the transformer for isolation, and then have your power stage on the high side with the IGBT brick.

That is exactly what i was thinking. The GDT driving the mosfets will provide isolation *and* the inverted drive signal. Adding the mosfets would not bring the inverted drive signals back in phase, so nothing to worry about there. (maybe not, a professional opinion is always good)

The power supply is nothing difficult. A large ATX transformer would be nice (large for a bigger winding window...) and easy. Also, i could make yet another winding for the power supply for the logic and gate drive board. One transformer will do everything. And it can be run straight from the wall with no transformer to run the smps' logic.

hvguy

Tue Aug 11 2009, 08:55PM

Registered Member #289 Joined: Mon Mar 06 2006, 10:45AM
Location: Conroe, TX
Posts: 154

The schematic you have shown provides the correct drive signals for the N/P half bridge without the bootstrap arrangement since the output of your GDT will be inherently bipolar, as you've noted. This will save you a bit on the parts count. You need to consider using a negative supply rail as well with your half bridge since your current schematic will not provided any negative bias at turn off. I'm still not sure what your afraid of with the GDT, it really comes down to this: GDT=less parts=less failure modes=more reliable. Before you commit to the discrete driver I highly recommend you try a GDT, just throw one together on the bench.

teslacoolguy

Wed Aug 12 2009, 06:05AM

Registered Member #1107 Joined: Thu Nov 08 2007, 10:09PM
Location:
Posts: 792

I personally don't see any reason not to go with a simple GDT. Like Hvguy said they are very simple, robust and will provide plenty of power for bricks (i use them on cm300's no problem). Go to TSC ferrite and buy some part number: 5000 38-19-25-0002 toroids on this page

Use about 6 windings of cat5 and parallel the 4 *striped* conductors for the primary. You will have to scope the outputs to find the desired value of gate resistor but that's all pretty straight forward.

Kizmo

Fri Mar 26 2010, 02:09PM

Registered Member #599 Joined: Thu Mar 22 2007, 07:40PM
Location: Northern Finland, Rovaniemi
Posts: 624

Sorry for bumping oldish thread.

I have some CM300Dy-12NF bricks here (600V 300A 5th gen from mitsubishi)

Could these bricks do 100kHz in drsstc? Obviously i need very beefy low impedance gate driver (similar to steve wards driver)

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob