Designing acpl-337j desaturation circuitry

The ACPL-337j has it built in so you dont have to worry about it.

• Integrated Fail-Safe IGBT Protection
  • DESAT Detection, “Soft” IGBT Turn-off and Fault Feedback
  • Under Voltage Lock Out (UVLO) Protection with Feedback

https://www.broadcom.com/products/optocouplers/industrial-plastic/isolated-gate-drive-optocouplers/highly-integrated-sma…

Thank you for fast reply. I know thak acpl-337j has it implemented but I want to know what diodes should I use as DESAT diodes and false fault prevention diodes (last page of datasheet). Correct me please if I'm wrong but I assumed that theese diodes are additional external components between pins 14 and 13 (false fault protection diodes (schotky and zener)) and betwen pin 14 and IGBT colector (regular diode and zener diode).

Thank you.

DESAT protection is built in to trigger a fault condition at 7V - Vf. Additional diodes can be added if you wish to change this fault condition.

If you wish to change the fault condition you would have to specify what your new fault condition is before selecting diodes and then work them out with the value sums on that page of the datasheet.

-----------

datasheet quote

The value of VCE  that triggers DESAT to signal a FAULT condition is nominally 7  V –  VF. If desired, this DESAT threshold voltage can be decreased by using multiple DESAT diodes or low-voltage Zener diode in series. If n is the number of DESAT diodes, the nominal threshold value  becomes VCE,FAULT(TH)  = 7 V – n  × VF. If a Zener diode is used, the nominal threshold value becomes VCE,FAULT(TH)  = 7 V  – VF  – VZ.  When using two diodes instead of one, then diodes with half of the total required maximum reverse-voltage rating may be chosen.

-----------

I think the built in protection is pretty general purpose, you would need to know what your expected fault condition was before you start adding diodes.

Oh, great! Looks like my reading skills are my largest promlem. So conclusion is that for driving some general purpose IGBTs and MOSFETs no diodes are required at all. Additional diodes are required only for cases where specific DESAT threshold is required. So is it ok if I wire everything up without diodes and look at short circuit current waveform (over IGBT) with scope and then if too large current is present for too long, I then put some diodes to reduce treshold.

Thanks.

Yes thats the way I'd go about it some people prefer doing the math for everything in the system but the hands on approach of wiring it up and testing is more fun!

I will try then. Thanks a lot!

"So is it ok if I wire everything up without diodes ..."

I'm not clear what you are saying here. It's a good idea to use the diode marked 'D DESAT' if you use the protection. Unless the load supply voltage is very low, it saves you from having to have a high power resistor for the 1k. It also saves you from having to limit the voltage at the output pin so it doesn't go above its absolute maximum rating. [There are probably other things it does, too, that I don't understand.]

When the IGBT is off, that diode will be reverse biased. It needs to have a reverse breakdown voltage that's larger than the highest voltage that could be seen at the collector - that might be the supply voltage or it might be the voltage that the collector rises to if an inductive component were switched and you rely on something to catch it above the supply voltage. The datasheet points out that it needs to be fast recovery too.

The datasheet is saying you can add more diodes if you want to adjust the voltage at which the protection triggers. You'll probably want to do that (I just had a quick look at a datasheet for an IGBT that could manage tens of amps and hundreds of volts, and the kind of power that it would be dissipating if the on collector voltage got to 5V was a bit alarming.)

I don't understand the equation - I'd have thought it should be 6V - Vf (on the basis that the resistor will be dropping an extra volt if it's carrying 1mA).

According to the datasheet, the zener and Schottky diode can be added to try and stop spikes from whatever it is you are switching from skipping past the RC filter and setting off the protection. They are worth putting in.

If you don't want to use the protection, you'll need to fool the chip into thinking that there's no fault. If you were to just leave off the diode, the capacitor would charge when you turn on the transistor and, after a fixed delay, would set off the protection (once it reached 7V). An appropriate resistor in place of the capacitor could hold it below the threshold level.

I disagree with Lucie. I think you should do the maths and use the protection that they've been kind enough to build into the chip for you.

"So is it ok if I wire everything up without diodes ..."

I'm not clear what you are saying here. It's a good idea to use the diode marked 'D DESAT' if you use the protection. Unless the load supply voltage is very low, it saves you from having to have a high power resistor for the 1k. It also saves you from having to limit the voltage at the output pin so it doesn't go above its absolute maximum rating. [There are probably other things it does, too, that I don't understand.]

When the IGBT is off, that diode will be reverse biased. It needs to have a reverse breakdown voltage that's larger than the highest voltage that could be seen at the collector - that might be the supply voltage or it might be the voltage that the collector rises to if an inductive component were switched and you rely on something to catch it above the supply voltage. The datasheet points out that it needs to be fast recovery too.

The datasheet is saying you can add more diodes if you want to adjust the voltage at which the protection triggers. You'll probably want to do that (I just had a quick look at a datasheet for an IGBT that could manage tens of amps and hundreds of volts, and the kind of power that it would be dissipating if the on collector voltage got to 5V was a bit alarming.)

I don't understand the equation - I'd have thought it should be 6V - Vf (on the basis that the resistor will be dropping an extra volt if it's carrying 1mA).

According to the datasheet, the zener and Schottky diode can be added to try and stop spikes from whatever it is you are switching from skipping past the RC filter and setting off the protection. They are worth putting in.

If you don't want to use the protection, you'll need to fool the chip into thinking that there's no fault. If you were to just leave off the diode, the capacitor would charge when you turn on the transistor and, after a fixed delay, would set off the protection (once it reached 7V). An appropriate resistor in place of the capacitor could hold it below the threshold level.

I disagree with Lucie. I think you should do the maths and use the protection that they've been kind enough to build into the chip for you.

I would like to use protection for sure. It is my first time that I'm doing something like this with ACPL-337j. So can you tell me the procedure on how to "do the math". I was thinking something like this:

1. search the datasheet for DRAIN to SOURCE voltages (in case of MOSFETS) or emmiter to collector (in case of IGBTs)

2. read voltage that equates to some current that is ok (for example 15A for 20A mosfet (leave some head room))

3. use diodes (D DESAT) in configuration that read voltage from step 2 is the treshold

Please tell me if I'm wrong, I would like to learn the procedure. For testing I'm using IRFP460N mosfets because I have a lot of them at home and IGBT bricks are expensive burning fuel for learning.

Thanks.

Yes, use the graphs in the datasheet. [When I said maths, I just meant simple arithmetic like Ohm's Law and power calculations.]

Complications are the fact that the on resistance changes with temperature (increases as the device warms up - which it will do if it's carrying a reasonable current - that one helps you). A further complication is if you are pulse driving the MOSFET at a current above the continuous rating.

So you need to be considering the power dissipation and your thermal management and not just the currents involved. For the MOSFET, when it's hard on it looks much like a low value resistor, so the dissipation quadruples for a doubling of the voltage.

If you play around with the figures you should be able to see if it's clear-cut and you can easily position the threshold to protect the device with a reasonable amount of headroom or whether it's really tight and going to be prone to false triggering when you might want to review your heat management or choose another device.