Centre-tapped PWM matching Xformer, will it work? / General Science and Electronics / Forums

Forums

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

« Previous topic | Next topic »

Centre-tapped PWM matching Xformer, will it work?

previous 1 2 3 next

Move Thread

LAN_403

Ash Small

Wed Apr 02 2014, 09:48AM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

By 'bi-polar PWM', do you mean (and I've seen this basic arrangement in some motor drive scenarios), where the high side switches have, say, 48-50% duty cycle, and the low side switches have variable PWM, and where current circulates through one high side switch and the other high side freewheeling diode when the low side switches are off, and through diagonally opposing switches when the low side switches are on? (I believe this is something to do with back-EMF when used for motor drives.)

If so, the high side switches could be driven through a GDT and the low side switches could be driven via a mosfet driver? Presumably I'd need two uChips to accomplish this?

I'm obviously referring to a full H-bridge arrangement here. Should I post a diagram to illustrate this arrangement?

Steve Conner

Wed Apr 02 2014, 09:54AM

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

No, I don't. Go back and look at the diagram I posted that shows a bipolar PWM waveform. It is like regular PWM but with every other cycle turned "upside down", this cancels out the DC component so it can pass through a transformer.

There is no problem using bipolar PWM with GDTs in general. The Instructables induction heater circuit is just broken (what else would you expect from Instructables?

)

Dr. Slack

Wed Apr 02 2014, 10:49AM

Registered Member #72 Joined: Thu Feb 09 2006, 08:29AM
Location: UK St. Albans
Posts: 1659

Steve Conner wrote ...

No, I don't. Go back and look at the diagram I posted that shows a bipolar PWM waveform.

Er no, Steve, that diagram by no means illustrates what you are trying to put across. I would expect the high and low pulses to look equal, most don't, and to be labelled equal, none are.

Ash, every +ve cycle must be followed by a -ve cycle with exactly the same pulse timing, so the integrated area under the total waveform, the DC component, is identically zero for every cycle.

Steve Conner

Wed Apr 02 2014, 11:30AM

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

Sorry, it wasn't a very good diagram. Let me try again...
Scopetrace

Ash Small

Wed Apr 02 2014, 01:44PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

I've been puzzling over this all morning, and only just noticed the last two posts. I can follow that the +ve and -ve pulses need to be equal in order to eliminate any DC component.

Am I correct in assuming that the gate swiitches on at the beginning of the +ve pulse and switches off at the beginning of the -ve pulse, and that the period of the +ve pulse plus the period of the 0v part equals the time that the gate remains on, and that by varying the period of the 0v part you can vary the time the gate is on from a minimum period equivalent to the +ve pulse up to a period equal to 50% duty cycle? If this is correct I assume you need two GDTs, one to drive the gates of switches 1 and 4 and the other to drive the gates of switches 2 and 3.

While I appreciate that you are explaining how to use PWM with GDTs, I don't see how this could be accomplished using a TL494, however, as I mentioned in my previous post, if I drive the high side switches at 50% duty cycle through a GDT, and use the PWM function of the TL494 to drive the low side switches via mosfet drivers I can achieve effectively the same thing, and any 'reflected voltage' will just cause current to circulate through the high side switch that is 'on' and the freewheeling diode of the other high side switch, thus preventing high voltages that could destroy the switches, or have I missed something here?

I have been doing a lot of googling, but can only find examples of bipolar PWM which either relate to motor drive circuits (chopper circuits) or inverter circuits that produce a 50-60Hz sine wave output.

EDIT: I think I'm correct in assuming that any reflected voltage (back EMF?) would cause (or want to cause) current to flow as described above, rather than from ground through a low side and high side freewheeling diode. Am I correct about this point? I know if all switches were 'off', current would want to flow through one low side and the opposite high side freewheeling diodes.

Maybe I should post a schematic, or simulate it in Spice?

Steve Conner

Wed Apr 02 2014, 02:37PM

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

No. You can drive 2 or 4 devices off a single GDT.

A MOSFET or IGBT switches on when the gate is driven more than a few volts positive. For zero or negative voltages it turns off.

So, in a half bridge, you drive the two devices from two separate GDT windings 180 degrees out of phase. The high side device would turn on during the positive PWM pulse, the low side would turn on during the negative PWM pulse, and both devices would be off during the 0V part.

For a full bridge you add another two windings to the GDT and repeat as above.

When the bridge drives a reactive load, the output voltage won't necessarily fall to zero when all the devices turn off. This is the problem with the Instructables gate drive circuit. A hefty resistor across the primary of the second GDT would be a quick and dirty fix.

Ash Small

Wed Apr 02 2014, 03:05PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

"A MOSFET or IGBT switches on when the gate is driven more than a few volts positive. For zero or negative voltages it turns off."

Yes, I'm aware of this, but I was under the impression that the GDT would need to be driven -ve for this to happan, when using a GDT. I thought this was what Artlav was referring to earlier.

It looks like it's a lot simpler than I anticipated. Thanks Steve and Neil for your input. I'll get some TL494's and see what happens.

Artlav

Wed Apr 02 2014, 03:50PM

Registered Member #8120 Joined: Thu Nov 15 2012, 06:06PM
Location: Moscow, Russia
Posts: 94

Steve Conner wrote ...
A hefty resistor across the primary of the second GDT would be a quick and dirty fix.

No, it won't be.
As low as 10 Ohm gives only slight slopes on the output, and heats like crazy.

That does raise the question - how do you drive a full bridge of bricks from TL494, if possible at all?
I guess i should browse the DRSSTC section - someone there was brewing tea by induction from one of these bridges. :)

Ben Solon

Wed Apr 02 2014, 05:49PM

Registered Member #3900 Joined: Thu May 19 2011, 08:28PM
Location:
Posts: 600

Artlav wrote ...

Steve Conner wrote ...
A hefty resistor across the primary of the second GDT would be a quick and dirty fix.

It's quick and dirty- not a perfect or ideal fix. If the resistor gets too hot then you chose too small a power rating. A slight slope is all you really need to eliminate the components of the square wave that want to ring. Ringing is only bad when it causes large transients that can either punch through the gate isolation(extremely unlikely) or puts the igbt into the linear region or shuts it off all together before it's meant to be. If your turn off time is too slow then all you're doing is frying bacon on your heatsink.

Ash Small

Wed Apr 02 2014, 08:11PM

Registered Member #3414 Joined: Sun Nov 14 2010, 05:05PM
Location: UK
Posts: 4245

Steve Conner wrote ...

The majority of PC power supplies used a TL494 driving a half bridge through a GDT, with voltage regulation by PWM.

Well, I've just ripped an old ATX supply to bits, but it had a Fairchild KA3511, not a TL494, but it seems to do pretty much the same thing, but it has 24 pins instead of 16. Only problem is a few of the 'legs' are a bit short after I removed it from the PCB.

Datasheet here:

Maybe I should just order a handful of TL494's from RS, Only problem is minimum order is Â£20, and five TL494's are ~Â£1. I guess I'll have to wait until I have a list of other stuff I'll need first.

Unless I try soldering some wires onto the broken legs of the KA3511

previous 1 2 3 next

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