MegaBrute CM600 gate driver / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

MegaBrute CM600 gate driver

previous 1 2

Move Thread

LAN_403

GeordieBoy

Thu Feb 05 2009, 03:00PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

As Dan said, a dedicated high-side driver is the best way to switch multiple paralleled power devices like IGBT modules. This is because it can have it's own floating decoupled supply capable of sourcing & sinking lots of amps to the gate capacitance very quickly. If the high-side driver is mounted right on top of the IGBT module, peak currents can be very high because it is not hampered by very much stray inductance at all. The design of the pulse transformer is now much more relaxed as it only needs to pass a signal indicating the switching instants, instead of carrying a wideband high current power waveform. Propagation delays, overshoot and ringing are all minimised.

Having said that, in a DRSSTC system the load current being switched by the inverter is sinusoidal and the IGBTs switch off softly anyway. Therefore the gate drive signal does not have to switch the IGBTs particularly quickly in this application. This is likely the reason that driving CM600s via a well designed GDT is workable.

If you insist on driving large MOS power modules with high gate capacitance through a GDT this is the way to do it...

Make the GDT with a turns ratio of something like 4:1 (This is done by twisting say 8 wires together and wrapping them through the core a few times.) Four of the wires are then connected in series to form the GDT's primary. The remaining four wires are connected in parallel to form a high-current secondary winding. Such a GDT transforms the very low impedance of the power device's capacitive gate up by a factor of 16 times. The effect of any leakage inductance on the primary side of this "step-down" GDT is reduced by 16 times as it actually appears smaller at the power device's gate.

In summary the driver must now operate from 4 times higher voltage than is required at the IGBT gate (say 80VDC), but only deliver 1/4 of the current. This greatly eases drive problems since it is rapidly changing currents (high di/dt) that usually cause the overshoot and ringing. The large gate drive currents only exist on the secondary side of the GDT, where the wiring can be kept short by locating the GDT right on top of the IGBT module. The elaborate multi-filar winding scheme for the GDT further acts to minimise leakage inductance by ensuring all of the series connected primary turns are very well coupled to all of the parallel connected secondary turns.

Such a step-down gate-drive approach is capable of swinging very large gate capacitances quickly. It is also the basis of those resonant gate-drive schemes used in Class E radio amplifiers.

-Richie,

Tom540

Thu Feb 05 2009, 06:18PM

Banned on 3/17/2009.
Registered Member #487 Joined: Sun Jul 09 2006, 01:22AM
Location:
Posts: 617

Actually the high side driver seems easier once the boards are taken care of. I wanted to make a dedicated drive like Dan's but instead of the IXYS driver chip I thought I would use a dual gate driver 27425 or 27325(inverting and non inverting outputs) driver chip pushing two n channel FETS much Like Steve's only eliminating the P channel FET. The upper FET would be capacitively coupled like in Steve's high side driver. I can draw up a quick schematic to explain further if needed.

Is there a problem with using two N channels? P channels are scarce and the ones I've found don't have the same current handling as the equivalent n channels. I thought the only advantage to the p channel was just for convenience in driving them? Or is that not the case? I always steer clear from using obsolete parts.

-Tom

Steve Ward

Sat Feb 07 2009, 05:28AM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

Just watch out for common mode noise getting into your high-side driver supplies. CM noise is a real issue when operating at 650VDC bus.

I dont know that i will ever go back to that high-side driver stuff. I design stuff like that at work now, and its annoying. The GDT approach seems easier to me, and probably more reliable.

My original high side driver suffered from under-voltage problems. As mentioned, when the drivers lost power, the output would turn ON for all 4 drivers! This was later fixed, easily, though i cant quite remember what i did to fix it. The schematics on my site are the updated ones that fail safely when losing power.

And, i would not accept an argument that the high side driver is faster. While this IS true, it doesnt help you. Because the nature of large IGBT bricks, you really do need to switch them slowly to avoid excitation of the parasitic inductance of the module. Remember that the voltage ringing you see at the outside of the module is EVEN WORSE inside the module where you cant get your scope probes. The only way to really get rid of the ringing is to slow down the switching transition (provided you have optimized the bus structure). The switching speed for CM300DY-24H modules using the "beefy GDT" approach seems completely acceptable at "sane" frequencies less than 70khz.

I have many coils that run from this design and have thus far proven to be very reliable. At the same time, Jeff Larson's twin of my DRSSTC-2 still uses my high side gate drivers and has been very reliable for years now as well. Really, it comes down to execution and implementation, not just the schematic :P.

Tom540

Sat Feb 07 2009, 07:07AM

Banned on 3/17/2009.
Registered Member #487 Joined: Sun Jul 09 2006, 01:22AM
Location:
Posts: 617

Well there's the other side of the discussion I was looking for. I already have all the stuff for both except I'd have to make boards for the high side driver. I guess I'll try GDT's first and if that doesn't work the way I want then ill do the high side stuff. I need to stop changing my mind.

previous 1 2

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