Overvoltage protection

You need to consider the "safe operating area" graph for that device. See the data sheet.

You are outside of the limits.

Regards tm

Of course that is the reason: I'm out of the limits. But how to prevent it? The transistor is for sure able to work for 50 ms at 300 A (being fully "on"). So, the problem may occur at the end of the pulse. I'm not able to measure it, because (1) it is too short, (2) it may damage my oscilloscope, (3) each try damages a transistor (10$). And I think, that designers of welders know how to solve the problem...

Regards P.

On a sunny day (Mon, 11 Feb 2013 20:44:01 +0100) it happened Piotrne wrote in :

Assuming for a moment that is really *is* that back EMF, you could use some diode to feed it back to the supply (capacitor). ------------------------------------power | | load | | | |------------------a diode k - transistor switch | | === S--- --- | | /// ///

But there is an other issue, HOW do you know you have 50 A or whatever max current you want in the transistor? The correct procdedure is to measure that current, and switch the transistor off before its limits has been reached. Either by a sense resistor, or a current transformer in point S, the pulse from that should be compared to some reference and remove the transistor drive. In fact if the circuit is purely resistive, you would need to add some inductance. But it likely is very inductive... You would need to scope the current (via a curent transformer with termination resistor) that would be safe for your scope, and start with a very short pulse to see what happens. But always provide that drive voltage reset circuit.

This is a great BIG transistor. I don't know the gate capacitance and the Miller effect charge, but the 47 Ohm gate resistor is going to cause very slow turn-on and turn-off of the transistor. Leaving the transistor in the linear region for milliseconds while it is turning on and off is going to cause excessive heating.

I think you have to turn it on and off much more abruptly, and use inductors to control the rate of current rise. It may make sense to run the switching frequency higher, recirculate current with an inductor and free-wheel diode, and modulate the PWM to control weld current. You might want to look at what commercial welders do.

Jon

Methinks the operation of the FET is beyond the I*I*T rating as well as the voltage rating at transformer turnoff.

Not clear what you're trying to accomplish. The devil will be in the details.

I spent several years tinkering on and off with a welder for battery tabs. Didn't get very far. Then I acquired a digital storage scope and a current probe. Had it working a lot better in a day. There's no substitute for making measurements to see what is actually happening.

Again, depends on what you're welding, but for me, constant current was not what I wanted. The resistance of the weldment and the contacts and...and...and make a significant difference is how much energy ends up where you want it.

Switching to constant energy by discharging high voltage from a cap thru an SCR into a transformer dramatically increased the reliability of the welds. That's how the CD welders do it. Of course, the transformer is an art.

Save some money while you are learning about power electronics :)

Less than $1.50 each. Maybe they are real.

I don't think you are turning the device fully on fast enough. It has an input C of about 0.02 uF. Can you reduce the 47 ohms?

Good luck.

Not sure what you mean by the "Burner"? I can only assume you are referring to the electrodes?

In any case, I am willing to bet that you are suffering from induction kick back.

Basically you are using 50VDC, but the release of this current with your pulse generating many more times that on the drain of your MFET..

Don't get fooled with that ZENER symbol for the foot print.. It does not do what you may think and the current in this return pulse of HV does not need to be high to damage the fet. Looking at the Cap rating on the gate, I would say it would be easy for it to short through the gate.

You may want to report in which manner the fet shorted? Gate to drain?

Source to Drain only ? etc.. This would better tell the story.

Maybe you should be considering a snubber around the FET.

P.S.

THe pulse current ratings of over 1k amps are only good for less than 400us.

Jamie

Electrodes, of course (they are not called "burner" by welders?)

I have simulated it in PSpice (SwitcherCAD) and it shows me pulses of about 45kV... Very bad. It also destroyed 4 drivers (IRS2001) in

Well, the transistor from the second experiment has shortened D to S only, the first (which I did not expect to destroy and turned off the circuit a few seconds later) is totally "thermally" destroyed.

I have replaced the driver with an IR2151 (which required to modify the generated PWM), reduced the resistor to 10 Ohm and asked the owner of the current source to complete it with an overvoltage protection (varistor + capacitor). I'll report results...

Thank you P.

Is this for a TIG machine? If so, is it using HF?

Can you put up a schematic of your design somewhere?

tm

How is the transformer configured? You mention just one diode, so is it jus t halfwave rectified making for only DC loading of the secondary? Once you get past the transient overvoltage issue, you may be confronted with a sat urated core short circuit issue.

halfwave > rectified making for only DC loading of the secondary?

It is a 3-phase transformer with a fullwave rectifier consisting of 90 diodes mounted on 6 aluminum plates (radiators). It provides a nice, slighlty wavy voltage which does not go to 0.

The whole device should be welding small (thin) elements by melting only their surface. Therefore, such short pulses of current should be used.

Today I have been trying to get rid of the interference caused by the ionizer. It simply resets the ATMega32 each time when the ionizer is turned on. It even influences separately the LCD display (2*20 characters) and the microcontroller: sometimes only the LCD shows strange thing, while the microcontroller keeps working and sometimes the microcontroller resets. Originally, the ionizer was turned on by the microcontroller (through a small, optically isolated circuit) and I suspected, that the distance between wires going to the ionizer and the microcontroller are too close (of course, the ionizer is turned on and off at the primary side, 230V). But even after completely disconnecting these two devices (microcontroller and ionizer) and keeping only common power supply, the problem persisted. Finally, a filter inserted between the mains and the ionizer removed the problem. Unfortunately, coming closer with other wires (the high current circuit) brought the problem back, even without turning on the current. So, it has to be solved too...

I'll report the progress :)

Regards P.

is this a spot or arc welder?

Jamie

just halfwave

odes

There are industrial line transient suppression devices based on the ckt sh own below. The advantage is it's a non-magnetics solution, simply parallel your rectifier with another that charges a large filter cap to peak line, t hen that becomes your high voltage spike clamp level:

Please view in a fixed-width font such as Courier.

. . . . . . --------- . | | . >----------------+--|PH1 +|------------> . | | | TO HIGH . 3PHAC >-------------+--|--|PH2 | CURRENT . | | | | LOAD . >----------+--|--|--|PH3 | . | | | | -|------+-----> . | | | | | | . | | | --------- --- . | | | /// . | | | . | | | . | | | --------- . | | | | | . | | '--|PH1 +|-------. . | | | | + | . | '-----|PH2 | --- . | | | --- . '--------|PH3 | | . | -|-------+ . | | | . --------- --- . /// .

...may be patent protected, but that's your headache...

Offhand, i would say your final gate drive is grossly incorrect. Switching FETs like being turned on hard and fast, and turned off hard and fast. Take a good look at your gate drive without the final transistor, better place a capacitor of 10 to 100 times C(gs) instead of the final transistor to account for Miller capacitance.

?-)

Ummm- load is inductive, Crss is miniscule, Qgs is only 180nC, gFS is something huge like 300A/V in active region, it's a usecs switch. The 47R is not a problem, the MOSFET is an EMP bomb.

it just halfwave

diodes

It is generally a good idea to wire any high current stuff with twisted pairs as much as possible. Same with any sensitive stuff. I can hardly believe you left off the line filters though.

?-)

halfwave