This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

UCC23513: uvlo inquiry

Part Number: UCC23513
Other Parts Discussed in Thread: ISO5852S

Hi,

my customer is building I type 3 phase 3 level inverter.

They wonder if we have any device without UVLO protection.

As you know, the I type 3 phase 3 level inverter's shutdown sequence is: the internal 2 IGBT(s2 and s3 in the picture) should be turned off after a period after the outer 2 IGBT(s1 and s4) is turned off, otherwise the IGBT will be damaged.

So if the internal IGBT gate driver like UCC23513's supply voltage is not stable and fall under UVLO threshold, the internal IGBT will be turned off when the outer IGBT is still on, and it will damage the IGBT.

But if the gate driver doesn't have UVLO protection, they can shutdown the power stage according to the correct procedure(first outer 2 IGBT, and then inner 2 IGBT).

So they wonder if we have any device who don't have UVLO.

I'm not sure if their comment is correct, please give me some insight.

  • Hi Howard,

    Thank you for your question. I work on the applications team in the high power drivers group.

    In short, no, all of our drivers have UVLO thresholds. This is done to prevent potential damage caused by under driving switches. A driver with no UVLO protection could pose a risk to the system. The 12-V UVLO on the UCC23513 is tailored to IGBT operation.  We also encourage customers to connect a capacitor near the Vcc and VEE pins to bypass noise.


    If this answered your question, could you please press the green button? If not, feel free to ask more questions.

    Thank you and best regards,
    Zachary

  • Zachary,

    thank you.

    Then how can we relieve the customer's concern given that all our products have UVLO protection feature.

  • Howard,

    All UCC23xxx parts currently have 12V UVLO. IGBTs generally requiring >10V UVLO threshold.

    UVLO is a pretty important protection that protect IGBT from damage by not allowing to be driven with insufficent drive voltage. I dont think it would be a very good idea to remove it. It is better for them to ensure their supply is stable rather than not having UVLO at all. I am not even sure of ISO gate drivers without that feature.


    From my understanding, youre saying their concern is that UVLO-induced fault in one IGBT will lead to large VCE transient on other that will damage it, correct?

    There could be other ways to help with such an issue, for example drivers featuring desat protection.

    Could this reference design for 3phase 3level inverter be useful for customer? outside drivers use ISO5852s which features desat protection, inner use UCC53x0 without that protection feature

    Unfortunately as you know we don't currently have opto-compatible driver with advanced protection features.

    Best

    Dimitri

  • James,

    youre saying their concern is that UVLO-induced fault in one IGBT will lead to large VCE transient on other that will damage it, correct?

    No. For example, if S2 turn off when S1 is on, and the inductor's current is from left to right (which means S3 and S4 current is from E to C from either the IGBT or the body diode), S2 C will be +VDC, S2 E will be -VDC. S2 will endure 2*VDC which is the total DC bus voltage, which is not allowed and will damage S2 itself.

    Gate driver featuring desat protection won't help in this case.

    The T type inverter in the reference design you show doesn't have such problem. 

  • Hi, Howard,

    I agree with the comments Zachary and Dimitri gave you yesterday about eliminating UVLO.

    If this is a big concern for your customer, then I would suggest they implement a supply voltage supervisor (SVS) IC to monitor the bias supplies, and if there is an issue, have an orderly shutdown of their system. TI just happens to have an entire line of SVS ICs to choose from.

    Let me follow-up via email. Can you mark this thread resolved?