• State TI Thinks Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 3 answers
  • Subscribers 64 subscribers
  • Views 612 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

UCC27212: can't make UCC27212 working, dead after few seconds with low side getting shorten internally

florent NS
Expert 1080 points
Part Number: UCC27212
Other Parts Discussed in Thread: UCC27712, UCC27712-Q1

Hello, I have designed motor control power stage for 20 years. Recently I had to deisgn a 1.5KW stage for a brushless DC motor as per I did before. This is the first time I use the UCC27212 driver. For the last 2 weeks I burnt 15 pieces just by starting my power stage while the current involved is as low as 0.5 A under 20VDC and the dead time high enough to make sure there is no cross conduction (that I checked with isolated probes). After 2 weeks of intensive unsuccessful trials I must admit there may be something wrong in the inner part of this device. As an example, when I even output a fixed duty on a light load, after 1-2 trials (start-stop-start...) the device get burnt : the low side is shorten internally and I have to replace it. I am sure of my layout (that I can share with you), I am sure of the connections, I am sure of the wiring. I tried different decoupling capas -> no change, even I tried to change the bootstrap diodes (and without)  -> no change, I changed series gate resistors  with higher values to slow down dv/dt -> no change.  

today I replaced the ucc27212 with legacy drivers from competition (l6386), all good. 

I am keen on using the UCC27212  as the datasheet looks outstanding, but to me there is a design or quality issue. Have you suffered any production issues on these devices? I could see that some distributors dont hold any more stock of these devices (Farnell for example). Is the 10 pin package with pin to pin spacing the issue? Do you have a reference design with proven motor control stage working at 2-3kW with these devices?

Thnaks/Florent

  • 0
    TI__Mastermind 20950 points

    Hello Florent,

    Thank you for the interest in the UCC27712. I am an applications engineer with high power drivers and will work to help resolve your concerns.

    Without seeing the schematic, and layout will help, and importantly some scope waveforms I am not able to give detailed advice.

    I do have some questions and comments.

    When the drivers fail do the power MOSFET’s fail also, or is it only the drivers?

    There are some common application related concerns when the UCC27712 is used where there were other driver devices used previously.

    The UCC27712 has higher gate drive current of 1.8A/2.8A than many 600V half bridge drivers. The L6386 is a 400mA/650mA driver so that is the case here. The higher gate drive current in many applications will result in higher dV/dt on the power train switch node, resulting in more ringing and noise. This ringing can result in voltage spikes exceeding the driver voltage ratings, or cause false triggering if noise is coupled in the driver input pins.

    I see you commented  the gate resistor values where changed, but to limit the current to the L6386 values the gate resistance would need to be at least 33 Ohms or more.

    The propagation delay of the UCC27712 is shorter than many 600V drivers, which may result in the UCC27712 responding to input signal noise that is rejected by slower drivers. Many times adding an R/C filter to the driver input pins, or increasing the filter capacitance if a filter is in place can resolve false triggering issues, if this is the case.

    There is a 4KW BLDC motor drive reference design based on the UCC27712-Q1, since this is for an automotive HVAC compressor. The UCC27712 has the same parameters as the Q1 device. The reference design can be found at: http://www.ti.com/tool/TIDA-01418

    So the UCC27712 has been used in high power motor designs previously.

    Can you provide the schematics of at least the gate driver section and MOSFET half bridge sections? Please include MOSFET/IGBT part number if not shown in the schematic. Also show how the HB bias is generated in the schematic.

    Waveforms, or scope plots showing the driver HO, LO, and switch node will be helpful. If the HO-HS signal can be recorded with a differential probe that is the most helpful. Show several switching cycles and a zoom in version of the turn on and turn off events of the high side and low side power devices.

    Also waveforms showing the driver LI, HI, LO and HO outputs will be helpful.

    Regards,

  • 0 in reply to Richard Herring
    Expert 1080 points

    Hello Richard, thanks for your reply. What I can see as the main difference is the active miller clamp which is not part of my design as I have never had to do so with any legacy drivers I ve been using so far. From what I can see here on my side s the driver low side going into short circuit in no time after a smooth start/stop of my motor (which I guess explains why the low side burns when stopping as the peak current shall increase), and this happen only when drawing 0.3A under 20VDC which is nearly nothing in term of power consumption. In term of gate drive I use the same kind of schematics, series resistors with shottky diode to turn off faster etc etc. Also I even increased the series gate drive to few hundreds ohm (100 ohm + diode for turn off, 220 ohm turn on, 3us dead time which is high as I expect max 800ns in the end) -> exactly the same happened, ucc27212 low side damaged internally (when I force manually the low side with 3.3V logic level the device goes into short circuit and gets very hot with current increasing drastically). 

    I believe that without the miller clamp you will get into the same trouble (could you give it a try on your side by just removing one FMMT720A on your ref design?). It is nowhere mentioned that this should be mandatory or at least suggested to use (it is not shown nor suggested in the user guide http://www.ti.com/lit/ug/sluubj7/sluubj7.pdf). From design point of view it is a matter of an extra transistor which is also quite expensive and takes space onto the PCB. The FMMT720TA is an added cost, the clamp should be part of the IC which is the case on the other drivers I am using (note ST gives for example dV/dt immunity ± 50 V/ns in full temperature range). 

    - Please confirm there is no miller clamp into the UCC27212?

    - For 60V design, what is the simplest BOM you recommend as an external active clamp?

    Thanks/Florent

  • 0 in reply to florent NS
    Expert 1080 points

    Also one more thing, I see you added also the shottky protection diodes on the gates (like d314/d317) on the 4Kw ref design schematics, I suppose also there is no embedded protection inside the chip correct? Thanks/Florent

  • 0 in reply to florent NS
    TI__Mastermind 20950 points

    Hello Florent,

    Thank you for the comments on the testing results and review of the reference design. On the reference design there is shown and explained the use of the PNP pull down device for driving the IGBT. The UCC27712 has good current sink capability within the driver of 2.8A. In the application usually the benefit of the PNP pull down device is that it can be placed very close to the power device gate and emitter to minimize the trace inductance.

    I many motor drive applications, and others it is not always practical to achieve a very short gate drive trace loop for both high side and low side FET's to the same driver IC. I looked to find the FMMT720A datasheet and did not have success. I see very low cost devices such as 2N2907(A) used in this application quite often.

    The driver IC does have capability to withstand driver output negative undershoot as shown in the datasheet of -2V for <100ns. The gate drive loop trace inductance can result in gate driver output overshoot or undershoot, especially if the trace lengths are long. The schottky diodes are shown for the purpose of clamping these voltage spikes to the bias and ground reference.

    Can you provide scope plots which will allow me to provide better advice?  Waveforms of LO, HO (with respect to HS), and HS showing several switching cycles. Also zoom in to show turn on and turn off details.

    Confirm if this addresses your concerns, or you can post additional questions on this thread.

    Regards,

  • 0 in reply to Richard Herring
    Expert 1080 points

    Hello Richard,, from  my last tests, I cannot see much difference by placing  protection shottky diodes as per your schematic on each mosfet gate. So seems there  is no issue with negative voltages (my shottky diodes have very low Vforward).-> the drivers still get burnt very quickly. As I had no suitable PNP last week I just placed a small capacitor of 1nF on each gate/source to decrease voltage spikes if any, then I could see some improvement but still the drivers get burnt by increasing a bit the overall power involved in the 3phase stage (20V/300mA <>  6W only). The dV/dT immunity may be the issue on this driver.  I am in the stage of making a new layout so that  I can test different drivers, that will be the best way to compare and come to a conclusion.  

    Another question : if I want to connect my power stage to one of your instaspin MCU, i can see theree is a need for a voltage sense on each U/V/W phase, like R403/R404/R405 on your schematics, correct? What are these sense for indeed? (that is extra lines not needed generally, are you sensing the back EMF when the MOSFET are in a suitable configuration?).

    B.Rgds/Florent

  • 0 in reply to florent NS
    TI__Mastermind 20950 points

    Hello Florent,

    Typically the schottkey diodes should help with possible driver output undershoot issues that can appear in certain applications.

    There are a couple of things that has helped with dV/dt switch node related issues, which can cause disturbance on the driver output due to the miller charge conducted into the gate. Slowing the dV/dt down by having a larger gate resistance for turn on, the turn off resistance can be lower with a diode/resistor for turn off. Adding capacitance to the Mosfet gate to source very close to the MOSFET is used often in motor drive applications. This provides two benefits, the Vgs rise and fall time will be longer for a given gate resistance, which allows lower resistance to the driver compared to not having Vgs capacitance. The added capacitance reduces the voltage deviation from the Cgd charge during switching.

    I would suggest increasing the gate to source capacitance to see if that continues to improve the operation. If there are long traces from the driver to the MOSFET gate and source terminals, this capacitance can help significantly since it can be close to the power device.

    For the instaspin MCU question, I would assume the separate current sensing is required for compatibility. If there is a single current sense point, there would need to be the logic/code in place to determine the timing of the CS signal relative to each phase. I am not sure about the back EMF question. I primarily support drivers and power train related issues.

    Please confirm if this helps address your questions, or you can post additional questions on this thread.

    Regards,

  • 0 in reply to Richard Herring
    Expert 1080 points

    Hello Richard,

    The shottkies dont help on my side, I guess the only issue is the dv/dt. Note I have also already increased the resistor gate drive to larger values without any improvement (100ohm turn off, 220 ohm turn on).

    For the instaspin question, I refer to the extra voltage senses (it is not current senses as you shall notice on your schematics, it is voltage sense  phase U V W). You may forward to one of your instaspin expert?

    Thanks

    Florent

  • 0 in reply to florent NS
    TI__Guru** 109810 points

    Yes, the 3-phase voltage sensing circuit is necessary to instaSPIN-FOC, the BEMF value is calculated by the sensing phase voltage of the motor, it does not depend on the power inverter, is just related to the motor.

    You might refer to chapter 5.2 (Hardware Prerequisites) of InstaSPIN user's guide (SPRUHJ1G) to understand how to design the voltage feedback circuit.