Stupid/lame IGBT question / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Stupid/lame IGBT question

Move Thread

LAN_403

LithiumLord

Thu Jul 02 2009, 02:36PM

Registered Member #1739 Joined: Fri Oct 03 2008, 10:05AM
Location: Moscow, Russia
Posts: 261

Well, I started a new project as many from the channel surely know :) I've ordered the electronics already, only LEDs and Vregs are not there (got some Vregs at home and LEDs are better to be obtained at flea market anyway where I'm going this weekend). However the question still worries me - I'm making a something close to Steve's DRSSTC .5, with the difference that I've designed an improved version of my older interruptor strobe reciever already, one featuring OCD with noise rejection trick still there, and as the drive part I'll use a stepdown GDT with 30vdc driver supply, but the power switch will be a halfbridge of IXGH40N60C2D1. They feel fast enough for a small coil I'm working on (aiming for a tiny 20cm secondary with all electronics in a used CDROM drive box beneath it), but the built-in diode makes me worry. Is there anything tricky about it, or it may be used the way it was designed for, without any freewheeling external circuits?

MinorityCarrier

Thu Jul 02 2009, 02:47PM

Registered Member #2123 Joined: Sat May 16 2009, 03:10AM
Location: Bend, Oregon
Posts: 312

If you think the the diode is not fast enough, why not add an anti-parrallel diode circuit using ultrafast DQ diodes like the APT2X101DQ120J?

LithiumLord

Sun Jul 05 2009, 11:17AM

Registered Member #1739 Joined: Fri Oct 03 2008, 10:05AM
Location: Moscow, Russia
Posts: 261

Well, just because the antiparallel circuit alone never works - and why should I use schottky diodes to lock off a built-in diode if I can just order a switch without one and only use a single UF per switch (btw that schottky needes some bigger heatsinks on my ISSTC, so using it would be extremely unwise in the new project which is expected to fit in a CDROM drive box - and the UF diodes, running cold on small ]-shaped sinks if used in full bypass circuit on FETs, in a single-diode bypass do not require any additional heatsinks at all as can even be placed on the same sinks as the switches).

Marko

Sun Jul 05 2009, 12:18PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

They feel fast enough for a small coil I'm working on (aiming for a tiny 20cm secondary with all electronics in a used CDROM drive box beneath it), but the built-in diode makes me worry. Is there anything tricky about it, or it may be used the way it was designed for, without any freewheeling external circuits?

I've never seen anyone using any extra diodes with IGBT's that already have them built in, and I have too hard time understanding why would 100ns Trr diodes be ''not fast enough''. Just use them as they are.

I would try to avoid any complex philosophy about the gate drive to increase the chance of everything actually working. Just use a pair of UCC's and a GDT, and prefferably 1:1 GDT if you are not going to push the IGBT peak current rating. It is the GDT that mainly limits the peak current, not the drivers. If I really wanted to mess with higher driver voltages I'd just use IXDD414's instead.

Also, don't put conductive blocks of metal such as a CD rom box under your coil if you don't want it to absorb large amounts of power.

Marko

LithiumLord

Sun Jul 05 2009, 01:25PM

Registered Member #1739 Joined: Fri Oct 03 2008, 10:05AM
Location: Moscow, Russia
Posts: 261

Thanks! As for the diode, actually I was afraid of the currents, not of the recovery times as it's clear that those being used in IGBTs are actual diodes, not a side-effect of the production process like they are in MOSFETs - therefore that would be just plain stupid to make a transistor more expensive while limiting it's parameters ;) However as I saw in some rare cases, the freewheeling current is sometimes more limited then the actual pulsed one - that's really rare and most switches have the built-in diode capable of handling the same currents the IGBT itself can drive, but as far as that's sometimes not working out that way a little paranoia kicked in :)

As for the drive, I ended up with 24/18v stepdown GDT because I don't want to use the optics here as the PSU would take too much space. Lol how did you guess I'm using 414's here - in fact I'm a big fan of Ixys stuff after those optocoupled drivers based off 408's ICs :)

>Also, don't put conductive blocks of metal such as a CD rom box under your coil if you don't want it to absorb large amounts of power.

Of course it will be away off the primary, just like in the present coil - but I don't feel like leaving the driver board uncovered as it's pretty scaled to feature not only the feedback, but also the OCD with LED bars (one for OCD readouts and another one for the threshold to visually represent the controller pot), the interruptor that is pretty massive with all those things, and also an antenna feedback that's used for the internal tuner (pretty much of something to be done just for an experiment - a XOR gate with additional 3-state output circuit to sinc it with the interruptor that charges an output integration cap while the signals of both the CT and antenna are phased and discharging it otherwise - the voltage off the cap can be checked with the same bar as used for OCD by flipping a switch, when it's maxed out it's good, the other way it's clear there's something wrong with the coil tuning, the less is the reading, the more tuning you need to do - hopefully it works out right and it would be a perfect scopeless tuner for field runs).

MinorityCarrier

Mon Jul 06 2009, 12:55AM

Registered Member #2123 Joined: Sat May 16 2009, 03:10AM
Location: Bend, Oregon
Posts: 312

The integrated fast-recovery diode fabricated in the die alongside the IGBT often isn't as fast as a stand-alone ultra-fast recovery diode that recieve extra process steps to improve speed.

Use of ultrafast recovery diodes in antiparallel configuration does increase complexity and cost, but there may be an improvement in efficiency by reducing turn-on losses. If you got the money, use of SiC Schottkey diodes in antiparallel configuration alongside a FRED in series conduction can provide a significant reduction in switching losses.

Just a thought. Maybe, as you say, it won't work for what you want.

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