Serious Problem with IGBT Power Ratings / Performance

Haven't looked at any of your figures (a circuit diagram may help ) - but have you allowed for the cold resistance of the filament at turn-on?

You can see the circuit below.

D1 is not needed. Obviously your current-limit cirquit is not responding fast enough so you must delay ramp-up. Put a 10n-22n capacitor from G to E. Mosfet will suffer enough heat load at shutdown too. I also assume halogen is 300V rated. S.M.

And I hope the phrase "as soon as 500W lamp load is connected" means you connect at AC side. Anyway do use leak resistances parallel to capacitors, both the proposed gate capacitor and bulk capacitor. Don't leave gate floating. And do something about inrush current on bulk capacitor. S.M.

Gate is tied with Zener diode and transistor switch. Gate charging current is 200V/100K or 2mA. Gate also gets discharged when capacitor C1 voltage discharges on power OFF condition.

Yes, there is no inrush current limiting in this design for charging capacitor C1 and there is no discharge resistor for C1 and only 200V Zener diode leakage and load leakage discharges C1.

I have small filter capacitor at Load.

When output voltage increases, it also makes the gate zener to get almost same voltage when output nears 200V making the gate to go OFF. But somehow that is not happening. Zener diode of 16V is fine. I tested the circuit removing the DUT IGBT and I get 16V across the Gate Voltage Control Zener Diode,

I will reduce the load today and will see if DUT still works or not.

Lamp is connected to DC output and lamp is shown in the figure. Lamp comes on as soon as power is switched ON but never goes OFF as IGBT is getting damaged.

D1 was placed to discharge output load capacitance in case input power goes off. This diode limits excessive voltage on output.

Even if current limit switch is not functioning fast, the load is not high to beat the IGBT but circuit ior load s somehow beating it hollow. Current in the circuit should not exceed the limit unless Halogen lamp has different behaviour like initially very high current and then low current. I did not check the filament resistance of the Halogen lamp. 2N3904 is fast switching transistor with <39ns response so I do not doubt that parts except that current limit is near 5A level and load is only 2A to 3A when DUT becomes short circuited. Hellogen lamp glows but IGBT becomes dead and IGBT acts like a conduction in dead state. Perhaps these are not PTC IGBTs. My friend has picked these from market and they look non PTC type. I do not find heat sink hot so IGBT are not affected by heat but are becoming dead due to instanious some kind poor condition.

Check power dissipation in Tr. Possibly more than maximum rating (IFR840=125W). Exceeding the SOA limit.

Suggest reduce load power dissipation and check for proper voltages and dissipation. than increase load in steps.

I have changed the setting for current trip circuit to 500mA.

I have also added inrush current limiting 3.3mH inductor after bridge rectifier.

I also added 0.1uF capacitor in parallel to gate Zener diode.

I added two numbers 100K 2W discharge resistors, one at charge storage capacitor and another at load for minimum load condition.

I have tested devices at limited load current to 500mA at 200V by placing resistive load and changed load sense resistor to 1 Ohm.

I found that few MOSFETs like IXFH12N100F could deliver this current into load but was getting heated. Also few Infineon MOSFETs could deliver this current but some cracked out.

Many Vishay MOSFETs rated for high current high power failedto deleiver current more than 60mA.IGBT rated for 40A delivered only 60mA. Devices were not fully switching on and perhaps due to Gat current leakage that may be very high for these devices. Gate current 2mA at 16V was not enough and result was only 60mA into load.

Filament resistance of the Helogen bulb used earlier was 40 ohms. I tested 100W ordinary bulb filament resistnace and it was also about 40 ohms. It looks inrush current in bulbs is 10 times at starting condition.

So there were some good results that were successes due to extremly good feedback from all of you and I must thank you all for this. My engineer and Assistant both were busy entire day and made changes and tested many types of MOSFETs and IGBTs.

Now problem stil exists as I need 3A at 200V into load so what is the next plan that may work. MOSFETs in parallel are nasty. Only PTC IGBTS are good for this.

I am going to use the following IGBTs in future.

IXGK120N60B

IXGX72N60C3H1

IXGK120N60B is 560W rating. Better.

This article on IGBT loss calculations looks good one.

http://www.fairchildsemi.com/an/AN/AN-7520.pdf

The application note AN7520 applies to SMPS, working at 50kHz. The Tr is switched fully On/Off at zero current (the aim) to reduce dissipation in the transistor.

Your circuit is working at DC and tr is not being switched fully on/ fully off. Just functioning as a voltage regulator. Dissipation in the tr is very great.

Or am I missing something??

MY actual need is to drive push-pull stage power driver at 25kHz for which our power source is highly fluctuating. This circuit was considered to provide constant voltage to push-pull driver.

Now that is it wasting power, I tried to convert it into current controlled rather than voltage regulated version. I removed that 200V Zener diode. There was still drop across MOSFET and lots of drop across IGBTs at 600mA into load. I found that it was due to low current into gate and leakage current was beating the gate current. I removed that 100K series resistor also and let only 47k resistor be there. This made drastic improvement for MOSFETs but did only small change in IGBTs. I think IGBT need lots of gate current so I will change this circuit to isolated switching driver of 4A peak current. I have used 2.5A peak current isolated drivers but will push more current to IGBT gate now.

I want to limit the voltage to push-pull driver but do not want to use bulky transformer of 600W. I also need to regulate the voltage / current to my push-pull drivers as power becomes square of the voltage change. I am still thinking of method to regulate the voltage. I have seen other achieving it by using thyristor firing angle control in the rectifier network. What else is the way leftover? WhateverI do must regulate the voltage to about 200V or even lower to 150V for my push-pull circuit as transformer winding and output voltage are critical pulse shapes and they generate 25kHz 1kV sine wave. Small change in voltage is big change in power. My transformers are just meeting the margin of 500W at output and lots of wastage of power in filters and IGBTs.

It would be important to have the initial (cold) resistance value of halogen bulb. S.M.

From #10:

"Filament resistance of the Helogen bulb used earlier was 40 ohms. I tested 100W ordinary bulb filament resistnace and it was also about 40 ohms."

I assume this was cold resistance.

If this is so, here's the deal. The current is well within IGBT capabilities, but the dissipated power as heat on it, when you use current limiting, is not. Discard transistor in gate and all related cirquitry, and your cirquit will work just fine, if halogen is rated for 310V (220 rectified and filtered). If you're trying to drive 220V 500W halogen that way, IGBT will have to dissipate about 200W(!) continously as heat even after start up. S.M.

IGBT was destroyed almost instantaniously within few seconds. Perhaps heat on the chip is much higher than it could move out by conduction and then chip inside becomes short circuited and then no more heat is generated. Gate, Drain and Source all are shorted and only few milli ohms resistance is seen in damaged IGBT. This is also happening to MOSFETs.

We changed load to power resistors 150W 300 ohms and limiting current to 600mA. This worked well for MOSFETs but had problem with IGBT as output current was as low as 60mA. I then increased gate current by decreasing gate resistance and IGBT output current also increased. It is still lower than 600mA. Now I have series resistor of 47K alone to drive gate voltage Zener of 16V.

Yes, all these resistances were measured in cold condition to estimate initial surge current through filament injected by the DUT. On the Halogen bub 500W 230V is written and simple filament bulb 100W 230V is written. As filament mass is very little, its heating up will be only within few hundred AC cycles.

Specs of this IGBT IXGK120N60B are a bit confusing one 120A/200A. Hence, decided to order IXGX72N60C3H1 which also looks near to specs. Lead time 10-12 weeks. I am booking these now.

I have following suggestions

1. Connect resistor of suitable velue , Watts rating - across C1. This shall eliminate the un wanted voltage spikes ( may be (3-5) * Vm = 3 * 1.41*240 = 1015V )

Failure due to dv/dt shall be stopped.

2. Have some arrangement for cold start of halogen lamp. To eliminate the di/dt in rush.

Usually 470K 2W is used and I have inserted 100K 2W.

Please explain how a resistor can eliminate voltage spikes.

I'm also rather confused by: "( may be (3-5) * Vm = 3 * 1.41*240 = 1015V )", which doesn't seem to make sense.

He has stated peak voltage X current. Resistance helps in the input stage before rectifier where there is no load. These are added along with parallel capacitors and common mode choke. Sometime two resistors and capacitors are used and their center point terminated to Ground path and not neutral.

You can see such filters in all induction cooker circuits and RF circuits of high power or even low power like 100W AC-DVC supplies of TDK Lambda and many standard manufactured power supplies.

Perhaps idea is to have some minimum load when power supply is not in use and it can also discharge all charge that may exist somewhere. I think it is an anticipatory protection scheme.

I'm familiar with the circuit arrangement and it's purposes, - it's the comment about voltage spikes that mystified me, as did the expression "(3-5) * Vm".

You are right.

Why not use thyristors as in SCR or TRIAC instead as your main power switching device? They are meant for that function..

Years ago there was an article about incandescent light bulbs operating on DC.

Incandescent bulbs operating on DC at high voltages have a very short life. The electrons on the negative end of the filament are emitted and strike the positive end of the filament. This heats the positive end of the filament and the filaments gets too hot and opens.

Have you contacted the halogen/helogen lamp manufacture to see if the lamp can be operated on DC?

Have you done any searching on the WWW to see if halogen/helogen lamps can operate on DC?

In fact that scheme does not require filament for heating at all and high voltage near 600V is enough to start the plasma generation by accelerated electrons perhaps generated by Cosmic radiation.. I agree that such lamps had very low life. In fact this scheme works on already dead FL Bulbs having damaged filament.

This not the point of discussion here. Lamp was tried as Load for testing the power supply and it is not a power supply design for lamps. This is to generate 3A 200V DC power from highly fluctuation mains voltage to deliver constant power to next stage circuits.

Have you considered using a resistor as a load instead of a halogen bulb?
This would remove the problem of a very low load resistance when the circuit is first turned on.
The halogen bulb has a very low resistance when it's cold.(maybe about 3 ohm)
The halogen bulb resistance could be checked with an ohmmeter.

Yes. Load is now a 150W resistor 300 ohms.

Hello Guys, I have tried the tools at www.ezigbt.com for IGBT selection. I find them very useful. Have any of you tried?