• State Resolved
  • Locked Locked
  • Replies 19 replies
  • Subscribers 64 subscribers
  • Views 1026 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.

TPS61288: Device destroyed when boost capacitor is too low even when output is unloaded

Christian Geißinger
Prodigy 40 points
Part Number: TPS61288
Other Parts Discussed in Thread: PMP40909

Tool/software:

Hello,

we build up the reference design https://www.ti.com/tool/PMP40909 to do some tests before we use it in our design. We had quite trouble to bring it up which had nothing to do with the usage of two TPS61288 in parallel. During the first power up we considered the usual precautions: The output was unloaded, the power source was current limited to 100mA. The input voltage was set to 5V, the output voltage to 12V. We only used one of the two TPS61288 on the board for the first tests. But directly during power the TPS61288 died. Before we powered the board, we could measure Vf of the body diode between the SW pin and the VOUT pin. After the powering the board there where zero ohm between SW and VOUT. After replacing the defective TPS61288 we could reproduce this behaviour. Also with another board same happened. To make it short: The root cause on board level was, that 100pF instead of 100nF whre mounted for the boost capacitor.
Now everything works as expected. But we can absolutely not explain ourself why this can destroy the TPS61288 in such way. Do you have any idea what could went wrong here? From our perpective none of the device specifications where violated. Neither the voltage at any of the pins, nor any current limits. When I took waveforms at the switch and boost pin, the voltage was never above VIN (5V). Maybe there is a mechanism, that in this case VGS of the HS MOSFET will exceed its limits? 

Many thanks and best regards,
Christian

  • 0
    TI__Genius 16765 points

    Hi Christian,

    Thank you for reaching out.

    I am a little confused here.

    Christian Geißinger said:
    that 100pF instead of 100nF whre mounted for the boost capacitor.

    It is found that is only the boot capacitors are 100nF. 

    Did you change the Boot capacitors to 100pF to solve the problem?

    Also, when you are using one of the two TPS61288s for first tests, are current sharing circuit like U3 and its external circuit soldered?

    Best Regards,

    Fergus

  • 0 in reply to Fergus He
    Prodigy 40 points

    Hi Fergus,

    thanks for your fast reply. I mean the boot capacitor, sorry for the misleading term. And yes, replacing it with 100nF instead of the accidentally mounted 100pF solved the problem.
    For testing only one of the TPS61822 I removed the shunt resistors and parts of the current sharing circuit. During troubleshooting I used a second board where only one of the TPS61822 and its circuit were mounted, so we can really ignore the current sharing circuit.
    All killed devices have a short between SW and VOUT, but not between SW and GND. Moreover the LS switch of the killed devices is still switching.

    Best regards,
    Christian

  • 0 in reply to Christian Geißinger
    TI__Genius 16765 points

    Hi Christian,

    It seems quite strange.

    Could you show me your schematic and the layout?

    Forgive me that I will be in out of office and reply will be slow during next week.

    Best Regards,

    Fergus

  • 0 in reply to Fergus He
    Prodigy 40 points

    Hi Fergus,

    we directly used the reference design. Schematics, PBC, BOM etc. are all available here:  https://www.ti.com/tool/PMP40909
    We used the Gerber files, that are also provided there, to order PCBs. We did not any modifications.

    Best regards,
    Christian

  • 0 in reply to Christian Geißinger
    TI__Genius 16765 points

    Hi Christian,

    I will assemble one board and test as you did.

    Best Regards,

    Fergus

  • 0 in reply to Fergus He
    Prodigy 40 points

    Hi Fergus,

    that sounds very good. If you want me to do some specific tests or measurements please let me know. I'm quite in practice to solder the TPS61288 :-).

    Best regards,
    Christian

  • 0 in reply to Fergus He
    Prodigy 40 points

    Hi Fergus,

    I expect, that the TPS61288 EVM https://www.ti.com/tool/TPS61288EVM-064 will show the same effect. If you have one available at TI, you can save the effort to assemble one PMP40909 board.

    Best regards,
    Christian

  • 0 in reply to Christian Geißinger
    TI__Genius 16765 points

    Hi Christian,

    Actually, I test the TPS61288 EVM and the output is normal.

    I will move the components to the parallelled one and have a try.

    Best Regards,

    Fergus

  • 0 in reply to Fergus He
    Prodigy 40 points

    Hi Fergus,

    this is the board with the reduced components which I'm using for analyzing the boot capacitor oddity.

    The PCB layout around the TPS61288 ist identically for the PMP40909 and the TPS61288EVM as I can see. Nevertheless I ordered a TPS61288EVM now, and do also a try next week if it behaves the same way.

    Meanwhile I did some further tests. For the tests I reduced the the input voltage from 5V to 3V, and the current limit of the voltage source to 50mA.In the first step I always successfully tested the board with 100nF boot capacitor, then I repeated the test with 100pF. Always with no load. I tried the following things (not simultaneously) :
    1. Removing the 100uF capacitor at the input to reduce a possible surge energy that might harm the device.
    2. Adding a 100 nF capacitor between switch node and GND.
    Every time the TPS61288 showed the same defect after powering it the first time with a 100pF boot capacitor. Up to date I have 15 devices broken in that way.

    Best regards,
    Christian

  • 0 in reply to Christian Geißinger
    TI__Prodigy 550 points

    Hi Christian,

    Our engineer is currently out of office, please expect a reply by next week Wednesday. Thank you for your patience.

    Best regards

    Yezi

  • 0 in reply to Christian Geißinger
    Prodigy 40 points

    Hi Fergus,

    the TPS6288EVM-64 arrived today, and it failed the same way after replacing the boost capacitor (C13) with 100pF.  So don't think it is worth to spend the effort that you build up the PMP40909.

    Best regards,
    Christian

  • 0 in reply to Christian Geißinger
    TI__Mastermind 20040 points

    Hi Christian,

    Our engineer is currently out of office, please expect a reply by next week Tuesday. Thank you for your patience.

    Best regards

    Lei

  • 0 in reply to Christian Geißinger
    TI__Genius 16765 points

    Hi Christian,

    Sorry for reply late.

    The boot capacitor is used for power supply for high-side MOSFET gate driver.

    When it is misused with 100pF, the energy is not enough for a complete on time of high side MOS.

    The current can only flue through its body diode, and it could cause the damage.

    Best Regards,

    Fergus 

  • 0 in reply to Fergus He
    Prodigy 40 points

    Hi Fergus,

    the current from the power supply is limited to 100 mA in my test setup. Do you really think this can damage the body diode? Furthermore the output is unloaded, so that after initial charging of the output capacitors nearly no current will flow. Additionally I added a Schottky diode as bypass for the internal body diode. This does not change anything.
    I just want to understand the failure mode, so that I can assess the robustness of the design.
    Due to the very limited energy the failure can not be caused by excessive current and power losses. IMHO there only remains a violation of VGS limits of the internal FETs. Maybe there is a ringing on the switch node and only a high enough capacity of the boot capacitor ensures that the voltage on the gate follows the voltage on the switch node? I don't know if the voltage at the switch node really looks like shown in the oscillogram below. The measurement is really one-shot because the device is broken afterwards, so I can not repeat this very often.

    What I'm wondering about is, that neither a snubber (1R+2N2) nor a fast schottky diode for clamping negative voltages at the switch node to GND can prevent the fail (I modified the board in such way).

    Best regards,
    Christian

  • 0 in reply to Christian Geißinger
    TI__Genius 16765 points

    Hi Christian,

    1. Your argument about the diode could be right.

    When the upper MOS cannot conduct, the power loss on body diode& upper MOS would be Vf*Io, this is large but not enough to damage the diode or MOS.

    2. 

    Christian Geißinger said:
    he current from the power supply is limited to 100 mA in my test setup.

    The input current limit cannot stop the burn as the input capacitor has enough energy to generate high current to device.

    3.  For the oscillogram, do you test the SW waveform with short grounding wire?

    The long grounding wire waveform and short one is listed below.

    4. I have repeated the device damage.

    The output voltage does not go to higher than 12V so that I guess the snubber is not needed.

    I will do more analysis to the failure moment and tell you later.

    Best Regards,

    Fergus

  • 0 in reply to Fergus He
    Prodigy 40 points

    Hi Fergus,

    thank you for your endurance with this issue.

    Fergus He said:
    1. Your argument about the diode could be right.

    Another argument that I omitted in our discusssion: I also did a test with an external schottky diode in parallel to the body diode. This didn't change anything.

    Fergus He said:
    The input current limit cannot stop the burn as the input capacitor has enough energy to generate high current to device.

    The same thoughts I also had. So I removed the input capacitors for one test. The device still was killed.

    Fergus He said:
    3.  For the oscillogram, do you test the SW waveform with short grounding wire?

    I was someone in between of your two figures. I have also some doubts if the measurement is plausible.

    Fergus He said:
    4. I have repeated the device damage.

    So I'm curious about your next findings.

    Best regards,
    Christian

  • 0 in reply to Christian Geißinger
    TI__Genius 16765 points

    Hi Christian,

    Please see waveforms when device damage.

    CH1 SW CH2 VOUT CH3 IL

    The Inadequate Boot voltage cannot ensure a fast shut down or start up as usually.

    This causes the shooting-through of high side and low side MOSes.(it can be seen that the SW voltage is not 0,which means high side MOS is not completely shut down)

    Then we can see that the high side mos is damaged and the current reversed.

    The damage power comes from output capacitors so decreasing input capacitance or limit input current is useless.

    Best Regards,

    Fergus

  • 0 in reply to Fergus He
    Prodigy 40 points

    Hi Fergus,

    that looks very reasonable, although it's not very obvious, that not only turning on the HS MOSFET is affected by an insufficient boot voltage, but also turning off the MOSFET. The part of the driver circuit that turns off the MOSFET therefore seems to also rely on the boot voltage. So when using the recommended 100nF boot capacitor the design is really safe.
    Thank you very much for your effort. This helped a lot to make my list of worries a bit smaller.

    Best regards,
    Christian

  • 0 in reply to Christian Geißinger
    TI__Genius 16765 points

    Hi Christian,

    You are welcome.

    It is also meaningful research for me.

    Best Regards,

    Fergus