• State TI Thinks Resolved
  • Locked Locked
  • Replies 11 replies
  • Answers 4 answers
  • Subscribers 65 subscribers
  • Views 704 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

UCC21750: PSPISE convergence issue when using two 21750 gate drivers to drive high and low side MOSFETs

A M
Prodigy 20 points
Part Number: UCC21750
Other Parts Discussed in Thread: UCC5350

When we simulate a buck converter using two UCC21750 to drive two SiC MOSFETs, PSPICE gives a convergence error at 27% of the simulation duration. On the high-side driver, we even used a 100 Megaohm resistor between the GND (on the primary side) and COM (on the secondary side). We still get convergence errors, and the simulation would not complete (even if we use the auto-convergence feature in PSPICE). GND and COM on the low-side driver are tied together. We use the exact same setup and replace the gate drivers with two UCC5350, the simulation runs fine, and we can see input/output voltages/currents. When using UCC5350, we don't have to include a 100 Megaohm between the primary and secondary side GNDs. We wonder if there are any issues with the UCC21750 PSPICE model because the PSPICE model for UCC5350 works fine. Would you please provide support in this regard?

  • 0
    TI__Expert 3760 points

    Hi,

    There is an issue with the PSPice model for the UCC21750 part where the pins of the part cannot be left floating (for example, if the CLMPI pin is left floating, floating COM for half-bridge configurations, etc). We are working on a solution but this will take some time. In the meantime, if you use just one gate driver in LTSpice and use the built-in COM node for the COM pin it should work. 

    Best,

    Nancy

  • 0 in reply to Nancy Hidalgo
    Prodigy 20 points

    Thanks for the information, Nancy! We are not leaving any pins floating and still getting convergence issues. Is that normal with the current model? We have not encountered convergence issues in PSpice when using a single UCC21750. Since you mentioned "LTSpice," are you suggesting that your model works better with LTSpice than PSpice?

  • 0
    TI__Expert 3760 points

    Hi,

    While using this model I found that PSpice did not like having a "floating" COM node at all (I would have to ground the COM node to be able to run a simulation successfully). Whereas in LTSpice, there is a built-in support for a separate ground node (pictured below). 

    Best,

    Nancy

  • 0 in reply to Nancy Hidalgo
    Prodigy 20 points

    Thank you, Nancy! A few more questions:

    - Looks like PSpice can simulate the 5350 model successfully because this driver does not have COM on the secondary side. Is that correct?

    - Is there any trick in PSpice to use the current 21750 model and still simulate a few circuits with reasonable accuracy? 

    - When will the current 21750 model be fixed?

    - Is there any gate driver's model similar to 21750 with no COM-related issue where we can successfully simulate it in PSpice?

    - Do you have any LTspice-based demo circuit for 21750 that we can evaluate?

  • 0
    TI__Expert 3760 points

    Hi,

    The lack of a COM node is likely why the 5350 model works with more ease.

    I was able to run some tests with the 21750 model in pspice if i connected the COM node to the GND node. 

    As far as I know, all of the UCC217xx models have this issue with the COM node unfortunately. 

    As for an improved model, I would have to check and get back to you. Once I have a better idea I will update this thread. 

    I've uploaded a LTSpice simulation that you can try out:

     https://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/196/UCC21750_5F00_model-evaluation.asc

    Best,

    Nancy

  • 0 in reply to Nancy Hidalgo
    Prodigy 20 points

    Thanks, Nancy! Have you had any chance to find out about the improved model's availability? Do you mind including the library file along with the schematic? The symbol gets detached from the schematic, and our model does not run properly with your schematic. 

  • 0 in reply to A M
    TI__Expert 7513 points

    Hi AM,

    Thanks for the follow up. Our improved Pspice model is still work in progress. We'll post it in the "ti.com" website of UCC21750 once it is ready to go. 

    Our expert would get back to you in terms of sharing the library file by the end of this week.

    Best,

    Pratik

  • 0
    TI__Expert 3760 points

    Hi,

    Here is the lib file I used:

     UCC21750_TRANS.lib

    (same one from ti.com)

    I don't believe this model came with a symbol but I used the instructions here to create the symbol and be able to use the pspice model in LTSpice: LTspice: Simple Steps to Import Third-Party Models | Analog Devices

    As for the new model, development is wrapping up but it will take some time to approve before it's on TI.com. I would check back in another month or so. Thanks for reaching out again and for your patience!

    Best,

    Nancy 

  • 0 in reply to Nancy Hidalgo
    Prodigy 20 points

    Thank you, Nancy! Using the current 21750 model, we can have two gate drivers in a half-bridge configuration (with the high-side com-pin connected to the high-side kelvin source, not to GND) and run simulations in LTspice successfully. Is that correct, or will the simulation fail even in LTspice?

  • 0 in reply to A M
    TI__Mastermind 23215 points

    Hello A.M.

    Nancy may be out of the office at the moment, she should respond to your latest questions within the next business day.

    Regards,

  • 0
    TI__Expert 3760 points

    Hi,

    Thanks for your patience. Unfortunately, the currently released model of the UCC217xx devices cannot support half-bridge architecture (the high side floating ground and the floating switch node will cause convergence problems even in LTSpice). 

    Best,

    Nancy