Gate Driver Modification not working

[AQUAMAN]

I have modified a concept2 gate driver

I have a low pass filter and then I want to clamp the voltage of this from 0.7V to -0.7V which I feed into an opamp

The diode used to clamp to -0.7V however stops the gate driver working, does anyone have an ideas why?
You can see my schematic

FWIW the gate driver works perfectly without this diode in.

Gate Driver swings from -10 to +15V

[miguelvp]

I so shouldn't do it, but can't really help it...

It will never work submerged in salt water.

[max666]

I don't know what is going on, but what happens if you use a 10k resistor instead of the 1k?
With the 1k resistor there are about -9mA flowing back into the driver, maybe he doesn't like that.

[AQUAMAN]

With 10k I think the low pass will attenuate what I want to view too much

The driver works without this 1 diode in, how did u come to 9mA conclusion?

[pa2ees]

I agree with Max666.  I would check with the datasheet of the device to see if there is any problem sinking that much current.  I would also try the 10k resistor, just to see.  Update: try a lower value capacitor with the 10k to not attenuate what you want to view.  If that's not good enough filtering, put two filters in series.

Also, what do you mean by "The diode used to clamp to -0.7V however stops the gate driver working" ?  Describe some of the symptoms that you see, for example a waveform that is different, or the voltage output goes to 0V, or 2.379V, or something to help us understand what you mean by 'not working'.  If the diode is there, does the output go positive but not negative?

I understand a gate driver should be able to source and sink a significant amount of current, so provide us with more details, and someone will provide you with good direction on solving your issue. 

As an electronics engineer/hobbyist, the type of problems you are facing provide you with the greatest learning and growing opportunities.  Embrace it, and share what you learn!

[AQUAMAN]

Well basically input to that opamp is clamped to -0.684V like it should be, but the voltage at the gate terminal seems to be stuck at around -2.8V or similar and it never changes

[AQUAMAN]

Just checked my simulation, probably u are right

Tomorrow I will try it with 2k

[miguelvp]

More info for whoever wants to dig deeper:

http://igbt-driver.com/products/scale-2-plug-and-play-drivers/2sp0320

The particular driver is a 2SP0320V_S

App notes are in that page including a Do's and Dont's
Do's and Dont's with SCALE™-2 Gate Drivers

2SP0320V and 2SP0320S Manual

More Info:
http://igbt-driver.com/design-support/articles

Edit: love the automotive application picture so I had to attach it.

Edit2: from the do's and don'ts:

Use of Diodes
The applications circuits, especially for gate driver cores, contain several diodes. The required types of diodes are clearly specified in the Application Manuals or Application Notes. The types of diodes (e.g. using a Schottky diode instead of the low-leakage diode BAS416) must not be changed, as this may lead to incorrect driver operation.

But by looking at the application notes your approach is way different (therefore that's why you mentioned it's a modified gate driver)

[T3sl4co1l]

You could go with 10k and 100pF and not change the rise/fall time very much.

Tim

[miguelvp]

Advantages of Advanced Active Clamping

That's for an older igbt

but a related article:
http://www.powerguru.org/driving-igbts-in-parallel-is-plug-and-play/

And this is the paper they refer in the manual /4/
Driver Solutions for High-voltage IGBTs

[AQUAMAN]

More info for whoever wants to dig deeper:

http://igbt-driver.com/products/scale-2-plug-and-play-drivers/2sp0320

The particular driver is a 2SP0320V_S

App notes are in that page including a Do's and Dont's
Do's and Dont's with SCALE™-2 Gate Drivers

2SP0320V and 2SP0320S Manual

More Info:
http://igbt-driver.com/design-support/articles

Edit: love the automotive application picture so I had to attach it.

Edit2: from the do's and don'ts:

Use of Diodes
The applications circuits, especially for gate driver cores, contain several diodes. The required types of diodes are clearly specified in the Application Manuals or Application Notes. The types of diodes (e.g. using a Schottky diode instead of the low-leakage diode BAS416) must not be changed, as this may lead to incorrect driver operation.

But by looking at the application notes your approach is way different (therefore that's why you mentioned it's a modified gate driver)

You are right I haven't modified any of the standalone gate driver. I merely have a PCB that is attached on top of the gate resistor which my RC, diodes and opamp are on.
THe driver is still driving a big 1700V 1400A IGBT though and so the actual Gate-emitter voltage is supposed to remain unchanged (other than perhaps the few mA directed into my RC/opamp, which is negligible when the peak gate current is probably getting close to 10A).

[AQUAMAN]

I tried a 2k resistor today with 6pF

The gate driver "works" but instead of -10 to 15, it goes from -6.5 to 16.5 (so for some reason the inbuilt protection incase of over +15V doesnt work anyway). And of course due to this shift my area of interest is now outside the voltages I have clamped between    It also seems like there could be a little bit of distortion in the general waveform pattern but it could just be my eyes

So I think probably as it is those who suggested it are right - the gate driver doesnt like the constant 9mA current that goes through this diode when its off - it probably thinks the gate is leaking massively and has some kind of protection, or it just simply doesnt like it

Ill see what happens when I use 10k but I dont have an 1-2pF capacitors, but in simulation this still tends to attenuate the signal I put into the opamp if I use large value resistors (I am actually looking at the variation in this signal with temperature, so my sensitivity decreases from say 4mV/C to 2mV/C)

[AQUAMAN]

It works with 5K

However!

It seems that there is significantly more random noise in the circuit. I am sure the way I have clamped is not ideal, and my 'PCB' is actually just a bit of strip board soldered on top of the gate resistor with some wire which is about 5cm long from this to the emitter to make a ground connection. But at the moment I take a measurement and I average over 20 samples on the oscilloscope (I have not yet considered adding the ADC, although I will have to) and it seems that although the majority of samples are good, there are significant amount that have huge oscillations in them, maybe 4 or 5 out of every 20 samples. Im not sure why this is. I bet if the IGBT was properly switching an alternating load current these would be even worse.