Full bridge driver design / General Science and Electronics / Forums

Forums

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

« Previous topic | Next topic »

Full bridge driver design

1 2 next

Move Thread

LAN_403

Mads Barnkob

Sun Jan 25 2009, 09:20PM

Registered Member #1403 Joined: Tue Mar 18 2008, 06:05PM
Location: Denmark, Odense C
Posts: 1968

Hey All

I want to build a TL494 driver at 16khz driving a full bridge of IRFP260s for a rather beefy flyback. I got some 6A maxim single mosfet drivers normal and inverted.

I'm have been looking at different SSTC and Inverter schematics for driving full bridges, but there is a couple of things really nagging me. They seem overcomplicated and redundant, and this is because I do not understand the purpose of the "middle" full bridge that drives the GDTs.

I have searched google quite some and only found patents and bridge suicide net solutions :/ blerg.

First of all, this is not a bashing of the people that made these, I respect your work but I would have made it different (if things work as I think they do :) )

Lets look at Uzzors Grand inverter

PWM -> transistors (instead of mosfet drivers) -> "middle" full bridge (because the smaller transistors are too weak?) -> GDT -> full bridge

I feel the "middle" bridge is somehow taking up space and messing with my head, but I hope its there for a reason beyond that the other components are too weak.

The same is going down in Wards SSTC2

and

PWM -> mosfet driver -> "middle" bridge -> GDT -> full bridge

Again I can only see the middle bridge as a amplifier to take load of the mosfet driver, if its 2 outputs where connected directly to the GDT it would be the same.

So am I right in my explanation of these two circuits? Could I possible use my 6A mosfet drivers to drive the GDT directly for a full bridge?

uzzors2k

Sun Jan 25 2009, 09:55PM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

It all depends on the driving frequency and power of the first gate driver. In the case of my full-bridge driver I found the first stage of BD transistors were much too weak to drive four IRFP450 mosfets at 140kHz, with minimal dead-time. I was never able to test with "real" gate driver chips, but I assume some regular UCC's could have done the job without needing the mosfet stage afterwards. So really, they are just there to amplify the signal in-order to drive the gates fast enough. In your application at only 16kHz, a tenth of the upper drive frequency I designed the full-bridge driver for, your 6A gate driver would be enough for acceptable gate drive, certainly with no duty cycle control.

Dr. Dark Current

Sun Jan 25 2009, 10:01PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

6A drivers should be fine for driving an IRFP460 fullbridge. Of course you should use two gate drivers, driving the GDT primary in a push-pull fashion. Or if necessary, you can even double up the drivers for better current handling (putting 2 in parallel on each side).

However 16kHz is very low. Most "flybacks" like to run at +-40kHz.

Another thing is, you should watch for resonance or exciting the sub-harmonics, because if you hit any resonant frequency (which is quite possible), you WILL destroy your transformer or bridge.

Mads Barnkob

Sun Jan 25 2009, 10:02PM

Registered Member #1403 Joined: Tue Mar 18 2008, 06:05PM
Location: Denmark, Odense C
Posts: 1968

Its a big semi disc style tv flyback from dr. spark, so I assume its resonant frequency is for NTSC 15734 Hz (picture of flyback:

)

I understand that dead time is a fine walk, its okay that we know one mosfet is off before the next conducts so we avoid shorting our bridge, but too long dead times would kill the free wheeling diodes.

I would like duty cycle control for sure, thats the whole idea of using a TL494 for it, but you do not see that happening?

I might be boned, the MAX4429 mosfet drivers can drive 2500pF to 18V at 40ns delay, 25ns fall and rise time, so 90ns (1,1MHz?), but the IRFP260s got 4000pF input capacitance, so at 16kHz they should still be able to handle this as they got a fair bit more time to do it?

uzzors2k

Mon Jan 26 2009, 06:58PM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

Haha, he really went through with it! Chris sure is a man of his word.

You'll be fine with mosfet drivers and deadtime. I've made it work discrete transistors before, without problems. When hard switching an inductive load which you'll be doing, the freewheel diodes will only see the peak current the switches see. Also with minimal deadtime the current never gets a chance to rise very high, so longer deadtimes don't do any more damage than short ones. The diodes will just conduct longer.

Steve Conner

Tue Jan 27 2009, 09:39AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Yes, Uzzors' gate drive setup is rather complicated. I can only assume it's a hack because he didn't have any gate drive chips handy.

At a pinch, you can drive the main GDT direct from the outputs of the TL494 itself, except you need a centre-tapped primary on it, because the TL494 outputs can only pass current in one direction. You could probably get away with this driving two TO247 MOSFETs at only 16kHz, but it would be unlikely to work at SSTC frequencies.

For more gate driving power, simply connect two gate drive chips like the popular TC442x, MAX442x, UCC3732x, whatever, one to each of the TL494's outputs. Any of these chips should drive any number of MOSFETs satisfactorily at 16kHz. You need pull-up resistors on the TL494's outputs when doing this, and the driver chips should be non-inverting.

(If you have inverting driver chips already and want to use them, you can wire the TL494's outputs as emitter followers, and use pull-down resistors.)

uzzors2k

Tue Jan 27 2009, 11:48AM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

Steve McConner wrote ...

Yes, Uzzors' gate drive setup is rather complicated. I can only assume it's a hack because he didn't have any gate drive chips handy.

Aye.

Have you thought of using the SG3525? It has internal push-pull outputs which could drive a one winding primary GDT directly, with no need for external components.

Mads Barnkob

Tue Jan 27 2009, 12:29PM

Registered Member #1403 Joined: Tue Mar 18 2008, 06:05PM
Location: Denmark, Odense C
Posts: 1968

Im going to use TL494 as I got them at hand, with MAX442X drivers, got both normal and inverted ones, actually I thought I had to use a normal and a inverted, but I am still studying the TL494 datasheet, so not that far in the design yet :)

I figured a standard 15:15:15:15:15 GDT would be fine. Not looked into, just a copied thought.

So I could just use a normal and a inverted driver chip and use pull up on the normal and pull down in the inverted, or should I just stay with 2 of the same, got 4 of each.

Steve Conner

Tue Jan 27 2009, 12:36PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Use 2 of the same. The TL494 has already taken care of the deadtime and bipolar drive for you, all you need to do is boost the current.

Mads Barnkob

Fri Mar 06 2009, 10:47PM

Registered Member #1403 Joined: Tue Mar 18 2008, 06:05PM
Location: Denmark, Odense C
Posts: 1968

So here is my first revision of a full bridge flyback driver, its mostly based on uzzors multipurpose inverters with a twist and less silicon :)

its running at 15644 Hz that was about the closest I could get to NTSC frequency (15734 Hz), I am using a TI TL494, their datasheet uses f=1(2*Rt*Ct) for push-pull, while some others like infineon uses 1.1/(2*Rt*Ct), hopefully I chose the right for my ic.

the PWM is directly copied from uzzors, as I have yet to understand how that control, ref, input loop is determined, I know it uses 0.5V to 3.5V to control 97% to 0% pulsewidth.

Pull-up resistors according to TL494 datasheet, I used MAX4420 6A mosfet drivers, they are supposed to have decoupling on both sides, but i figured the distance to that cap on the board would be about the same as a piece of wire to the other side side (1-4 and 5-8 needs decoupling according to datasheet)

Powersection is also almost identical to uzzors, its build to be driven with up to 200VDC. I left out overcurrent protection in the first place, I do not expect to push over 1kW through a flyback without it failing before the bridge.

Maybe some 5 ohm gate resistors...

Suggestions and constructive critisism is more than welcome :)

1 2 next

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