UCC21750-Q1: Driver Safety with respect to DESAT pin

Harsh,

Harsh Songara said:

1. The UCC21750-Q1 has DESAT pin as opposed to a OC pin on the UCC21732-Q1. Under the application information of the UCC21750-Q1 as attached here

We are in the process of uploading the full design files (waiting on the team to post them to the web). I will attach the full design files for the UCC21750 EVM here so you can use them before they get uploade. It will be same file. If you open schematic, you can change the variant in Altium and see the exact changes on the PCB in 3d view (which becomes no pop)

2626.UCC217XXQDWEVM-025.zip

The relevant protection diodes are shown below along with P/Ns, its from the same EVM guide, i recommend you to use both Zener and schottky. <1A for the schottyky and you are correct about the voltage rating, should be >VDD-VEE. The reason for schottky is not exactly the same as a zener clamp, basically its to prevent any noise from switching coupling to DESAT pin and bringing it below zero, you need that fast response there, a zener couldn't do that exactly as a schottky. Zener should be used as well.

Harsh Songara said:

2. On the above, just to be sure could you tell me if those three conditions are applicable only if the DESAT pin is not used. I believe not as if I were to use the DESAT pin I will need the clamp protection that these diodes would provide.

If desat, yes you need those protection diodes because of the nature of how DESAT detection works, wheras OC you do not. For the UCC217xx variants with OC pin, those Diodes are Not populated.

Harsh Songara said:

3. Do I need to clamp the Gate pin (after Rgon & Rg0ff) to VDD and VEE for this design? The datasheet suggests internal diodes upto VDD only. But since I'm planning to use a Bipolar supply do I need to do it externally for the Negative voltage?

That depends on the system need. Its not uncommon to see TVS across Gate/Emitter pins. In the EVM design, there is D9 which is the TVS diode.

Please let me know if you have further questions, if i've answered your question please let me know by pressing the green button.

best
dimitri

Harsh,

BTW I was discussing your question about using 2 Schottky diodes (one replaces the zener), and cathode of that schottky connect to VDD

This should be OK, since the ABS max of DESAT pin voltage could support a schottky there without risk of exceeding the Abs.Max condition of voltage on DESAT pin.

and i realize i didn't answer fully about purpose of TVS, it is to suppress overshoot/undershoot and ringing on the gate caused by switching and can be made worse by hard switching.

Best

Dimitri

Harsh,

Harsh Songara said:

Why must this current be less than 1A specifically? Is there any calculation for this that I can review/access?

We don't have a calculation, but a colleague had done a study using schottky diode to protect the gate driver itself from ringing, the spec for peak current capability should be on the order of 1A, and voltage rating should be set to at least exceed VDD-VEE.

Harsh Songara said:

I did my bit of research on this and I believe that I do understand the concept here. From my memory of reading the ISO5452-Q1 datasheet, I remember the HV DESAT diodes must have a low reverse capacitance and must be fast recovery. The capacitance is of great importance here because of the dV/dt getting coupled as a part of the Capacitive Voltage divider between the HV diodes and the blanking capacitor.

So, I think that the Schottky diode must be chosen such that this dV/dt at the DESAT pin with reference to COM must be low enough so that the DESAT pin does not get pulled lower than COM-0.3V.

This is one reason yes, which is why schottky is usally placed on in // with the zener and tied to COM/ IGBT emitter, to provide protection against DESAT going below com.

Additionally, the RED schottky can be used in lieu of 12V zener as there is not a chance of exceeding maximum condition on DESAT pin.

Harsh Songara said:

3. In order to alter my VDESAT threshold appropriate to my IGBT's VCE can I use a Zener Diode as used in other of TI's DESAT applications?

Yes, as shown in the above image, there is the 3V zener which is used to decrease the DESAT threshold. this would increase the apparent DESAT threshold by 3V.

You can see a previous thread discussing this here

Also, if you desire more precision, other parts in UCC217xx family that have OC pin, you are able to pick an arbitrary DESAT voltage and set it with a resistor divider.

Let me know if you have further questions

Best

Dimitri

Harsh,

Thank you getting back. This is actually what I wanted to ask, but yes this is clear now. However, a follow up question exists. Is using the Schottky+Zener in any manner more beneficial than the Schottky+Schottky as:

There shouldn't be a huge benefit to go either way. Zener could be used in cases where VDD is close to max supported, then better to use a zener.

Harsh Songara said:

How much should the HV diode's junction capacitance be smaller than the Capacitance of the Blanking Capacitor?

We don't have any guideline for this.

Harsh Songara said:

From my product research I've found that the Schottky diode has a capacitance which is not very much different from the minimum blanking capacitance I calculated. My minimum blanking capacitance is about 27pF. And some of the Schottky's I have seen have about 27pF too. So, when you say that the Schottky & both the Zener would be placed in parallel to the Blanking capacitor, I believe I would need to consider the junction capacitances of both these devices along with the blanking capacitor of my choice while calculating Blanking time? I haven't seen this particular in any of the resources I have been able to access to date. Was any such observation made by Texas Instruments while testing EVMs and Gate drivers in general. 

Of course the capacitance will be impacted by the diodes hanging off the desat pin as well as lots of other factors, but in our own documents we have never suggested to consider capacitive impact from the zener or schottky diodes, and I don't think we have any studies on this. Since they are on the same order as your blanking, you could consider the capacitance from your zener / schottky when reverse biased, it migh tbe more accurate. Regardless, the diodes cap is very nonlinear too, and theres even other variables to consider. it still will never be accurate as your calcualtion. Blanking time should be on the order that you calculated but will never be exactly what you calculated. Sorry, i dont have any more specific information than this.

Harsh Songara said:

I shall explore this option right away.

Great, if you need any more information about those parts let me know.

best

dimitri

Harsh,

The zener current spec as listed in the datasheet is different the "reverse" current listed is more like how much current it takes at the Zener voltage knee, it is rated for 500mW which is closer to a quarter Amp DC, peak current capability will be much higher. I believe this is is just specific to how zeners are rated as more of power/ protection device.

For HV diode:. The number you should be concerned about is the forward peak rating, which is it on the order of 10s of Amps (30A) so its perfectly capable.

The 200uA number you have suggested is the current flowing back throught the device when the device is reverse biased at its max working voltage, meaning Collector of IGBT is 1000V+ higher potential than VDESAT. This is not a number you should concern with because if your system is designed such that youre constantly pushing current back through the diode it will fail. You should spec that HV diode > than the max voltage you expect at collector. Or, you can place two in series.

Basically, those parts/ratings are OK!

Best

Dimitri