Should I worry about shoothrough? / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Should I worry about shoothrough?

previous 1 2 3 4 next

Move Thread

LAN_403

Steve Conner

Mon Apr 25 2011, 07:15AM

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

Those gate drive ICs are more trouble than they're worth. They look great on paper but tend to self-destruct if you get the PCB layout wrong.

MOSFETs and fast IGBTs will work fine all driven by a single GDT with no deadtime. With older, slow IGBTs you may need to introduce some. But you can easily do that with a single GDT. There are three possible drive voltages to the GDT primary: +, - and 0, which gives all devices off. That's what the deadtime control feature on chips like the TL494 is for.

James

Mon Apr 25 2011, 07:45AM

Registered Member #3610 Joined: Thu Jan 13 2011, 03:29AM
Location: Seattle, WA
Posts: 506

bwang wrote ...

It's a sad fact of life that MOSFET drivers hate shorts

Also, if you wind a GDT, be careful with your choice of core; you should scope its output before using it.
The vast majority of the cores found in ATX power supplies do not work.

Yeah most of the cores you find in those are powdered iron, and the ferrite toroids for interference suppression are intentionally lossy at high frequencies. You want a good low loss, high permeability ferrite. I bought a bag of suitable cores on ebay that work great. Even if you have to buy them retail they're cheaper than those ICs.

Couple of good examples of properly driving a GDT. You need to make sure you can push plenty of current to handle the gate load of whatever mosfets or IGBTs you use.

IC:

Transistors:

In the latter circuit you can omit the diodes across the transistors, they're not needed.

Ash Small

Mon Apr 25 2011, 10:05AM

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

Thanks for all your replies

Steve McConner wrote ...

Those gate drive ICs are more trouble than they're worth. They look great on paper but tend to self-destruct if you get the PCB layout wrong.

This is primarily why I'm trying to aviod IC's. This is my first attempt at a full bridge circuit.

Steve McConner wrote ...

MOSFETs and fast IGBTs will work fine all driven by a single GDT with no deadtime. With older, slow IGBTs you may need to introduce some. But you can easily do that with a single GDT. There are three possible drive voltages to the GDT primary: +, - and 0, which gives all devices off. That's what the deadtime control feature on chips like the TL494 is for.

Bwang and others have confirmed that it is possible to use just one GDT.

Patrick, frequency will be ~25kHz, I beleive I mentioned this in the first post above.

Now for the clever (or maybe not so clever bit, I'll await any comments!)

The circuit below shows an astable multivibrator with squarewave outputs 1 and 2.

I could use these outputs to switch a couple of transistors on a conventional centre-tapped push-pull GDT, with four secondaries.

However, If I replace R1 and R4 with the two halves of the centre-tapped primary, this simplifies things to the point where the GDT driver circuit consists of just two transistors, two capacitors and two resistors.

Now for the REALLY clever bit (I think). As TR1 won't begin to switch off until TR2 is on, this automatically introduces 'dead time' where both transistors are on, so both halves of the centre-tapped primary are conducting. This is effectively the same as the '0 volt' state that Steve mentions above. This is determined by the time it takes the transistors to switch off. (Td(off))

Any comments will be appreciated. (Am I on the wrong lines completely or MIGHT this actually work?)

EDIT: I suspect there must be a reason why this won't work, otherwise everyone would be doing it, but I don't know what that reason might be. From what I've read on astables, there 'seems' to be no reason why it shouldn't work.

(BTW, I may use an ungapped TV or monitor flyback core (or two) as the GDT core if I can't find a suitable toroidal one)

Steve Conner

Mon Apr 25 2011, 02:55PM

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

That is exactly how the output stage in the TL494 works, it has two open-collector transistors that can be used with a centre tapped GDT primary, and has been used that way since the 70s.

The drive capability isn't that great, though, and you need to be sure whether it turns both off or both on for deadtime. I can't remember which way it is, but neither is ideal: both on needs a current limit resistor, both off and the GDT can ring.

A chip like the SG3525, with two totem pole outputs capable of forcing the 0v deadtime state, is better.

We mostly use one or other of these PWM chips and connect the two outputs to two driver ICs like the UCC3732x. Then each chip drives one end of the GDT primary. This can drive big bridges at hundreds of kHz.

Ash Small

Mon Apr 25 2011, 03:12PM

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

Steve McConner wrote ...

That is exactly how the output stage in the TL494 works, it has two open-collector transistors that can be used with a centre tapped GDT primary, and has been used that way since the 70s.

So are you saying that the circuit I suggested isn't used because a TL494 does exactly the same thing, but uses up less space? I assume using power transistors my proposed circuit could drive more power than a TL494? (It looks like the TL494 is limited to 250 mA and 40 V)

Steve McConner wrote ...

The drive capability isn't that great, though, and you need to be sure whether it turns both off or both on for deadtime. I can't remember which way it is, but neither is ideal: both on needs a current limit resistor, ........

Would the current limiting resistor be needed because the impedance of each half of the primary would change if both were 'on' at the same time (increasing current flow by a factor of several times), or simply because twice as much current would be drawn if both halves were 'on'?.....or some other reason?

I'm only planning to run this at 25kHz, for maximum core efficiency (Ftyp=25kHz, Fmax=100kHz, N27 cores)

ScotchTapeLord

Mon Apr 25 2011, 05:04PM

Registered Member #1875 Joined: Sun Dec 21 2008, 06:36PM
Location:
Posts: 635

The "third state" we're throwing around here would work if you were not using a center tap. It will work with the center tap, but when both go low you will be wasting a significant amount of power, as you'd be pulling current with both transistors. Also, if using a center tapped GDT, R1 and R4 would be useless as you cannot source to a positive center tap.
It would be better if you introduced the dead time with both transistors going off (pulling no current through each "leg" of the primary instead of current equally through both), which is how the TL494 works.

Not to mention, if you don't have a good 50% duty cycle you could see some shifting of your output, and *I think* that could cause your third state to shift and because an active state for certain switches. With your astable circuit, your duty cycle is at the mercy of your component tolerances. With the TL494 you have accurate timing.

Ash Small

Mon Apr 25 2011, 06:05PM

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

ScotchTapeLord wrote ...

The "third state" we're throwing around here would work if you were not using a center tap. It will work with the center tap, but when both go low you will be wasting a significant amount of power, as you'd be pulling current with both transistors.

It will only be 'wasting power' for something like 250 nS or thereabouts, so I consider this to be negligible. (OK so it's 250 nS 50 thousand times a second, but still......1.25% of the time, X 2 = 97.5% as efficient as both being off)

ScotchTapeLord wrote ...

Also, if using a center tapped GDT, R1 and R4 would be useless as you cannot source to a positive center tap.

I don't follow you here. R1 and R4 are connected to a common live feed (see above illustration). Simply replacing them with the two 'halves' of the primary effectively creates a centre-tapped primary.

ScotchTapeLord wrote ...

It would be better if you introduced the dead time with both transistors going off (pulling no current through each "leg" of the primary instead of current equally through both), which is how the TL494 works.

I agree, but that would complicate things.

ScotchTapeLord wrote ...

Not to mention, if you don't have a good 50% duty cycle you could see some shifting of your output, and *I think* that could cause your third state to shift and because an active state for certain switches. With your astable circuit, your duty cycle is at the mercy of your component tolerances. With the TL494 you have accurate timing.

I am trying to 'visualise' the effect of both opposing primaries being 'on' at the same time, though.

Will the currents 'cancel out' as in a filter choke resulting in no current flow in either primary winding? Will the magnetic inductances in the core 'cancel out', thus greatly reducing the impedance in both primary windings and leading to huge current flow? Or will the current flow just be twice that usually flowing in one primary winding?

Maybe I'll have to set up a test and measure it?

As this thread progresses I'm increasingly tempted to try it, just waiting to finish making the heatsinks, which is taking ages with just hand tools.

Steve Conner

Mon Apr 25 2011, 06:22PM

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

Ash Small wrote ...

So are you saying that the circuit I suggested isn't used because a TL494 does exactly the same thing, but uses up less space? I assume using power transistors my proposed circuit could drive more power than a TL494? (It looks like the TL494 is limited to 250 mA and 40 V)

Does exactly the same thing, takes up less space, costs less when assembly labour is factored in, has soft-start, can be pulse width modulated to regulate the output voltage, etc.

In "Electric Motors And Control Techniques" there is a circuit for a DC motor drive chopper that uses the two transistor astable as the PWM, and a tunnel diode as the current limiter. The TL494 hadn't been invented, I think.

wrote ...

Would the current limiting resistor be needed because the impedance of each half of the primary would change if both were 'on' at the same time (increasing current flow by a factor of several times)

When both transistors are on the induced EMFs cancel and only the DC resistance of the windings limits the current. It's exactly the same problem as shoot-through in a bridge.

Ash Small

Mon Apr 25 2011, 06:32PM

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

Steve McConner wrote ...

.
In "Electric Motors And Control Techniques" there is a circuit for a DC motor drive chopper that uses the two transistor astable as the PWM, and a tunnel diode as the current limiter. The TL494 hadn't been invented, I think.

.
When both transistors are on the induced EMFs cancel and only the DC resistance of the windings limits the current. It's exactly the same problem as shoot-through in a bridge.

I'd better start googling current limiting diodes then, Thanks for the tips Steve.

Steve Conner

Mon Apr 25 2011, 07:10PM

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

Ah, no, I meant that the tunnel diode was used as part of the circuit that sensed motor current and limited it. I just pointed it out to give a feel for the general archaic weirdness of the circuit.

I wasn't recommending it as some sort of band-aid for your circuit, if you can even find a tunnel diode nowadays.

previous 1 2 3 4 next

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