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.

UCC27288: Circuit design guidance for UCC27288DR Gate Driver

Part Number: UCC27288
Good afternoon,
I have been struggling for a couple of weeks now implementing the UCC27288DR Gate Driver to drive IPP200N15N3 Mosfets for use in a 3 phase motor driver. The control works and the motor operates successfully, however the gate drivers get hot very quickly (~15 seconds) and quickly destroy themselves after a few uses.
I have compared my circuit (image attached below) to the UCC27288EVM development board that TI offers and I do not see any major difference that might explain the circuit failing so prematurely.
A couple of notes:
The motor is a FANUC servo motor. It is rated for 200W (54V at approx. 3-4 amps). As noted in the attached photo of the circuit, I am running it at 12V and on the same power supply as the gate driver.
Regarding the operating frequency of the driver, I am not doing any high speed PWM control currently. There are 4 commutation cycles per revolution of the motor and the drivers are only changed when a feedback encoder registers an angular change of 15 degrees. From my observation the motor spins at approx. 300rpm, so each gate driver sees approx. a 40hz input signal.
I have this currently wired on a breadboard, and I considered it acceptable since it is operating at a low frequency.
The power supply is a PSU from a computer. The 12V rail handles up to 4.2 amps per line. I have tried to supply separate rails to the gate driver and mosfets and it does not help.
I have probed HO/LO pins and I do see the HO pin is not a sharp transition from off to on. However, I cannot seem to figure out why it is that way. I thought it might be a power supply issue, but the amperage rating should be more than enough for the drivers.
If technical support would rather expedite this issue, I would be happy to settle for a correct circuit to go forward with. In the attached photo of the circuit I am currently using, I have only shown 1 phase of the motor driver. The circuit is duplicated 2 more times for the 3 total phases and is identical to the attached photo.
Regards,
Robert 
  • Robert,

    Sorry you are having difficulty with your circuit. 

    Do the FETs get hot as well? Or only the driver itself?

    Best regards,

    Don

  • Only the drivers. I did swap to all new drivers and FETs just to be sure and the problem persists. And all 3 drivers get hot. It isn't just one.

  • Hi Robert, 

    Thanks for the information. 40Hz switching frequency is very slow and this would mean that the bootstrap capacitor needs to hold the charge above HB UVLO for 12.5ms (assuming 50% duty cycle). The HB quiescent current will slowly discharge the capacitor during long times and in this case would totally deplete the capacitor charge.

    Using IT=CV, the voltage drop due to the HB quiescent current in a specific time interval is V=IT/C, where:  

    • “I” is the HB quiescent current of the driver
    • “T” is the time it takes for the capacitor to discharge to “V” level
    • “V” is the voltage drop in the capacitor after “T” time interval
    • “C” is the capacitance of the bootstrap capacitor

    For this slow frequency, you might want to consider using a floating/isolated bias supply to supply the high side driver (from HB to HS) instead of the bootstrap circuit. 

    Best regards,

    Leslie 

  • Thank you Leslie. I did check the high side gate output on the oscilloscope and its waveform makes a lot more sense regarding your analysis that the capacitor is discharging. It looks like it lacks the "force" to drive the gate quickly.

    Could I offset this by paralleling multiple capacitors to maintain a charge for longer? At least for testing? Regarding the equation you provided, should I set V to be my final voltage after driving the gate?

    For the floating/isolated bias supply, are you suggesting I remove the diode and capacitor and add in something like a 12V battery between HB and HS?

    Is there anything inherently wrong with my circuit that would prevent me from running at a higher frequency and voltage? I'm aware the diode would need to be replaced for a higher voltage rating. Based on your post, I'm assuming the solution is just to run at a higher frequency. What would be the minimum/maximum frequency? How would I go about calculating this? Is it based on the "T" time for the capacitor to discharge to "V" level or are they not related?

    Frankly, this is new to me. I don't mind experimenting, but I am running low on gate drivers (and they are not in stock anymore), so anything I could do to minimize destructive testing would be very helpful.

  • Hi Robert, 

    I understand how frustrating it can be to keep damaging devices... while you are running experiments I would recommend to follow these recommendations to minimize this:

    • First, test the driver without a load and with Vbus OFF. VDD would be powered and the inputs switching so you can observe the outputs switching as well. In this step, make sure that both HO and LO are never ON at the same time since this would cause a shoot through event (a short from Vbus to GND). For example, when LO turns off, wait 100ns before turning HO on. 
    • Once you confirmed there are no issues in step 1, still with the load removed, you can turn on the Vbus but start at low voltages (e.g. 10V) and increase slowly from there. In this step, you can check that none of the recommended operating conditions of the driver pins are being exceeded. From example, the HS voltage (including transients) shouldn't exceed the maximum recommended voltage. 
    • Lastly, you can go ahead and connect the load and test the full circuit. Still in this step, start with a low Vbus voltage and increase slowly, always checking the supply current of the driver and the bus voltage. 

    Would you mind providing more details about the reason for using 40Hz? Is it a requirement for your system or are you just experimenting at the moment? If currently you are just trying to learn the device behavior, I would recommend starting with 100KHz switching frequency. Additionally, you can find an example of a design implementation in the datasheet in section 8.2.1 "Design Requirements" that you can use as a guide. This section also has guidelines for selecting the bootstrap capacitor components. If running at 100KHz, you don't need the floating supply that I suggested. 

    Best regards,

    Leslie

  • Thank you Leslie. I have soldered another 3 sets and I will go through your steps you suggested.

    The 40hz was just a back-of-hand calculation. The servo motor is rated for 54V but I figured it would be easier to spot issues if I ran at a lower power, i.e. 12V in this case. At this voltage it spins slowly and I just guesstimated the approximate rotational speed, and from the commutation sequence the approximate operating frequency of the gate drivers. The intention is to run the gate drivers at a much higher frequency.

    It's funny now that I think about that last statement. Had I programmed the commutation for the intended operating frequency I likely would not be in this position.

    I did go through the datasheet and I have referenced the design implementation you mentioned, but I have just been picking component values without a clear understanding of the reasoning behind it. As an example, the pinout on page 3 of the datasheet states a recommended capacitor value of 0.1uf on the HB pin. I picked the component but did not understand until you explained it, that there is an assumption that the user isn't operating at 40hz.

    I will take some time to implement your suggestions. For my comfort, is the circuit I provided suitable for the application? It is much less involved than the one you reference in section 8.2.1, but I would like to know if it should suffice going forward (with a 100khz frequency).

    Regards,

    Robert

  • Hi Robert, 

    Our expertise is the gate driver itself, not the specific application, so I'm not able to provide advice regarding the system. However, from the driver point of view, the bootstrap capacitor you selected is sized correctly for the FETs you are using, and the VDD capacitor is also sized correctly (10 times or more the size of the bootstrap capacitor). One thing we also suggest is to add a small bypass capacitor connected as close as possible to VDD and VSS pins to filter high frequency noise (e.g. 100nF).

    We do recommend to use a schottky diode for fast reverse recovery and low forward voltage drop, but the voltage rating should take into account the Vbus voltage including transients. For example, if the Vbus voltage is 54V, and the HS transients during switching reach 70V, your FET, driver and diode should be rated to tolerate that voltage plus some headroom to account for any variations such as temperature. The diode you selected is rated for 40V, which is too low if your bus voltage is higher than that. My suggestion would be to select a diode that has the same voltage rating as the FET's Vdss voltage rating. 

    Another concern I see is the setup. You mentioned that you are running this test on a breadboard. The parasitics of the board play a big role in these applications, since any parasitic inductances will translate in higher transients that can potentially exceed the ratings of the driver. The "Layout" section in the datasheet has our recommendations to minimize this. Additionally, I would recommend to always monitor the signals with the oscilloscope (specially HS to GND) to make sure the transients are within recommended limits. 

    Regarding your comment about testing at lower voltages, I do agree with your approach. It is always good to start at lower voltages and increase the voltage slowly while monitoring the correct behavior of the circuit (when I run tests in the lab, I start at 1V Vbus and slowly turn the knob up the desired voltage). The main concern with you previous test conditions was the switching frequency because the bootstrap circuit won't be able to stay charged long enough to work correctly at 40Hz. 

    Feel free to contact us if you need further support!

    Best regards,

    Leslie

  • Thank you Leslie for your assistance and guiding me along the right path. I understand not being able provide confirmation on specific designs. Like any project I have to make concessions for what is available. I have been made aware of the limitations of breadboarding the circuit, but the time involved in iterating on pcb designs and a small supply of gate drivers necessitated it. As well as the diode being what it is since the schottky diodes I do have are all SMD.

    I have been doing further reading since your suggestions (TI has a video on gate drivers for low frequency applications which I watched), including referencing the datasheet again knowing what I know now. Despite the frustration of these past couple of weeks, I do have a greater understanding of what is actually happening, and how little I understood what I was doing initially.

    I still have a ways to go before I can commit to a final design and implement it into my robot arms, but I appreciate you giving me enough to work through its operation and tune it (hopefully non-destructively).

    Regards,

    Robert