• State Resolved
  • Locked Locked
  • Replies 28 replies
  • Subscribers 62 subscribers
  • Views 2930 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.

TPS650864: TPS65086401RSKR for my Zynq ultrascale+

sam chen
Prodigy 160 points
Part Number: TPS650864
Other Parts Discussed in Thread: CSD87331Q3D

I use TPS65086401RSKR for my Zynq ultrascale+ board.

The Voltage list below are stable when power on and CTL1 is 3.3V.

VSYS=12V

LDO5=5V

LDO3P3=3.3V

BUCK2=0.85V

BUCK1=1.8V

LDOA2=1.2V

LDOA1=1.8V

BUCK6=1.5V (CTL2=1, CTL3=1)

But GPO1 becomes a 300us pulse, Voh=3.3V, Vol=0V, (CTL6 shorted to GPO1).

When the GPO1 pulse goes high, in the meantime BUCK4 voltage rising to 1V only. 

When GPO1 pulse goes low, in the meantime BUCK4 voltage falling to 0V slowly.

I would like to know if the capacitance of BUCK4 is large, will the PMIC treat it as power fault after 300us?

I check the resistance is about 1.7k ohms from BUCK4 to GND.

Thanks!

  • 0
    TI__Expert 4155 points
    Sam,

    According to section 6.10.7 the PMIC will not powerfault unless it takes 10ms for a rail to come up.

    How much capacitance do you have on BUCK4? We don't recommend anything above 400uF.

    Also, if the part is shutting down after GPO1 is high for 300us, you can try reading the SHUTDNSRC, PWR_FAULT_STATUS1, and PWR_FAULT_STATUS2 registers to see why the PMIC is shutting down.

    Thank you,
    Nick
  • 0 in reply to Nick Jakse49
    Prodigy 160 points

    Because the PMIC I2C is directly connected to Zynq ultrascale+, I can not load firmware without 0.85V and 3.3V

    Above image is BUCK2 (green), GPO1 (Yellow) at boot up time,  I am sure the VSYS(12V) and PVIN3,4,5(5V) are stable, CTL1 is always 3.3V.

    My BUCK4 capacitance is more than 500uF, I remove them all, the situation is still the same.

    Above image is BUCK4 (green), GPO1 (Yellow) at boot up time.

    Below image is BUCK4 (green), GPO1 (Yellow) after 1s.

    Could you please advise me which power rail is fail?

    Thanks!

  • 0 in reply to sam chen
    Prodigy 160 points

    Please see the 5V waveform, 5V source input (green), 5V input at pvin3,4,5 pins (yellow), is the 5V for pvin3,4,5 pins not good?  

    Thanks!

  • 0 in reply to sam chen
    TI__Expert 4155 points

    Sam,

    From what you have shared I think your issue is with BUCK2. You can see it starts to drop at the same time GPO1 goes low. If something else was causing the part to shut down I would expect BUCK2 to wait 4ms after GPO1 goes low before shutting down, but since it isn't waiting 4ms I think that drop is the cause of the shutdown. 

    The waveform you shared of BUCK4 is what I would expect in this case. It looks like it is starting to turn on, but then the part shuts off, so it just keeps repeating this process (like in your 2nd picture), until the output cap is fully charged, and you end up with the waveform like in your 3rd picture.

    It is hard to say whether your 5V input is bad, but I think it may just be noise on the scope. If you measure this with the "tip and barrel" method across the input cap I would expect this noise to decrease.

    Is there any way you can measure the current through BUCK2?

    Also, if you want to share your schematic I can see if I notice any issues. 

    Thank you,

    Nick

  • 0 in reply to Nick Jakse49
    Prodigy 160 points

    Thank you very much.

    This image is my schematics of BUCK2, FET device is CSD87331Q3D_DQZ_8, is the current limited by ILIM2 resistor? 

    I also found the impedance of VCCPSINTFP or VCCPSINTLP(Zynq ultrascale+) is only 5~6 ohms. it that right?

    Sam Chen

  • 0 in reply to sam chen
    Prodigy 160 points
    I measure the BUCK2 current with current meter, it is maximum current is 36mA.
  • 0 in reply to sam chen
    Prodigy 160 points
    Update:
    The BUCK2 average current is 108mA, BUCK4 average current is 5mA
  • 0 in reply to sam chen
    TI__Expert 4155 points
    Sam,

    I don't think the ILIM resistor is an issue. From my calculations it shouldn't current limit until about 9A.

    How much capacitance do you have on BUCK4? Could you probe the LX4 pin? I realized that the rails will all shut down simultaneously during an emergency shutdown so BUCK2 might not be the issue. Do all of the other rails look fine?

    As well if you want to send the rest of the schematics I could look those over as well, just in case there is something that looks suspicious on one of the other rails.

    Thanks,
    Nick
  • 0 in reply to Nick Jakse49
    Prodigy 160 points

    Thank you.

    I remove some capacitors of BUCK4, I think it is only 300uF, now.

    I also measure the current of BUCK4, it is only 5mA.

    The other rails are as beow:

    LDOA1

    LDOA2

    LDOA3

    SWA1

    SWB1/SWB2

    Thank you.

    Sam Chen

  • 0 in reply to sam chen
    Prodigy 160 points

    Hi Nick,

    It looks LDOA3 is strange, it should be 1.2V (default) for TPS65086401, its input is BUCK1 (1.8V)

    BUCK1

    Sam Chem

  • 0 in reply to sam chen
    TI__Expert 4155 points
    Sam,

    I agree something is definitely not right with LDOA3.

    Can you check your schematic to make sure there isn't anything else connected to the LDOA3 rail that may be pulling it up to 1.8V?

    Thanks,
    Nick
  • 0 in reply to Nick Jakse49
    Prodigy 160 points

    Hi Nick,

    Is there any settings for BUCK4 (TPS65086401) to make its current is limited (100 mA)?

    I do many tests, if the BUCK4 output current < 100 mA then the 3.3V is stable (PGOOD is always on), even BUCK1, BUCK2 and BUCK6 are stable, too.

    But, if the BUCK4 output current >100 mA, the GPO1 output is repeat between 0V and 3.3V.

    The BUCK4 decoupling capacitance is about 300uF now.

    Thanks,

    Sam Chen

  • 0 in reply to sam chen
    TI__Expert 4155 points
    Sam,

    Unfortunately there is not a way to change the BUCK4 current limit.

    When you keep the BUCK4 current under 100mA do you still see that jump up in voltage on LDOA3? Does LDOA3 still do that if it is disconnected from the system?
  • 0 in reply to Nick Jakse49
    Prodigy 160 points

    Hi Nick

    When I do the test for BUCK4, I temporarily pull CTRL4 pin to 0V.

    The LDOA3 is 0V also.

    Thank you

    Sam Chen

  • 0 in reply to sam chen
    TI__Expert 4155 points
    Sam,

    If keeping CTL4 low limits the load current on BUCK4, but also turns off LDOA3, then I am curious whether the issues goes away because the BUCK4 current is lower, or if it goes away because that jump up in LDOA3 voltage never happens, because LDOA3 never turns on.

    Maybe you can try connecting a diode with a 1.2V drop on the LDOA3 pin to see if that fixes the issue?

    Thanks,
    Nick
  • 0 in reply to Nick Jakse49
    Prodigy 160 points

    Hi Nick,

    I first keep CTRL4 low to turn off LDOA3, but the PMIC is still repeatedly shut down.

    Then I just reduce BUCK4 loading, I find the BUCK4 is stable 3.3V and the current of BUCK4 is 25mA.

    Then I just increase BUCK4 loading, I find the BUCK4 is 2.36V and the current of BUCK4 is only 125mA.

    Thanks,

    Sam Chen

  • 0 in reply to sam chen
    TI__Expert 4155 points
    Sam,

    Can you try probing the switch node for BUCK4?

    Thanks,
    Nick
  • 0 in reply to Nick Jakse49
    Prodigy 160 points

    Hi Nick

    image#1 and #2 are BUCK4 3.3V stable state, LX4 waveforms (Yellow)

    image#3 is BUCK4 3.3V stable state, 5V decoupling capacitor waveforms (Yellow), PVIN4 pin waveforms (green)

    image#1:

    image#2:

    image#3:

    image#4 is BUCK4 unstable state, LX4 waveforms (Yellow), GPO1 (green)

    image#5 is BUCK4 unstable state, 5V decoupling capacitor waveforms (Yellow), PVIN4 pin waveforms (green)

    image#4:

    image#5

    Thanks,

    Sam Chen

  • 0 in reply to sam chen
    TI__Expert 4155 points
    Sam,

    I would try switching the BUCK3 inductor to a 470nH.

    The converters are optimized for 0.47uH or 1uH inductors so this should increase stability and may fix your issue.
  • 0 in reply to Nick Jakse49
    Prodigy 160 points

    Hi Nick,

    The issue is still exist with 0.47uH or 1uH inductor.

    If I pull down Zynq Ultrascale+'s PS_POR_B and PMIC's CTRL4 to 0V, the PMIC is stable, the voltage of  BUCK1, BUCK2, BUCK6, LDOA2, LDOA1, VTT and BUCK4 are correct.

    But, when I pull up Zynq Ultrascale+'s PS_POR_B to 1.8V after BUCK4 is stable, then the PMIC is shutdown again.

    I find the LDO3P3 drops to 2.7V when PS_POR_B is 1.8V for 1.8ms, then the GPO1 drops to 0V. 

    Is the PMIC shutdown when POR of Zynq Ultrascale+?

    Channel#1 is LDO3P3 (Yellow), Channel#2 is PS_POR_B (Green)

    Channel#1 is LDO3P3 (Yellow), Channel#2 is GPO1 (Green)

    Thanks!

    Sam Chen

  • 0 in reply to sam chen
    TI__Expert 4155 points
    Sam,

    I am not sure why pulling PS_POR_B high would cause the 3.3V LDO to start to drop. That is an output pin of the PMIC, so maybe if the PMIC was trying to pull it low, and you pulled it high it would cause an issue.

    Can you try putting BUCK4 in forced PWM mode and see if that helps at all?

    Also, if you have a layout you want to send me I can take a look at that and let you know if there is anything that can be improved.
  • 0 in reply to Nick Jakse49
    TI__Expert 4155 points
    Sam,

    I just wanted to check in on this. Has there been any updates?
  • 0 in reply to Nick Jakse49
    Prodigy 160 points

    Hi Nick,

    Yes, I decrease external 5V input for PVIN3/PVIN4/PVIN5/V5ANA to 4.2V, then the CPU boot up successfully. (all voltage of BUCK1~6, LDOA1~A3 and VTT are correct)

    But, LDO5P0 becomes only 3.9~4.1V, LDO3P3 is 3.3V, VSYS is 12V.

    The LDO5P0 is only connected to DRV5V_2_A1 and DRV5V_1_6 in my circuit.

    Do you know why using 4.2V for PVIN3/PVIN4/PVIN5/V5ANA is better than 5V? Is the internal load-switch problem for selecting LDO5P0 output?

    The I2C read for DEVICEID1 and DEVICEID2 are always "0x00", for some registers are "0xaa".

    The I2C write is correct, because I can change BUCK6 and LDOA1 voltage to 1.2V and 2.5V.

    Thanks!

    Sam Chen

  • 0 in reply to sam chen
    TI__Expert 4155 points
    Sam,

    LDO5P0 dropping in voltage when you decrease V5ANA isn't surprising. Once V5ANA drops below 4.7V it will turn on the load switch to connect it to LDO5P0, but if you are forcing V5ANA to 4.2V, then it would also be forcing LDO5P0 to that voltage as well.

    My only thought on why lowering the BUCK input voltage fixes your issue is that maybe the lower voltage causes less ripple/switch node ringing? Did you ever try putting this in forced PWM mode to see if it fixes the issue?
  • 0 in reply to Nick Jakse49
    Prodigy 160 points

    Hi Nick,

    My default power on sequence is "12V-->4.2V-->0.85V (BUCK2)-->1.8V (BUCK1)-->3.3V (BUCK4) -->load I2C F/W-->I2C F/W force PWM".

    I can not do the sequence "load I2C F/W-->I2C F/W force PWM-->change 4.2 to 5V".

    Thanks!

    Sam Chen

  • 0 in reply to sam chen
    TI__Expert 4155 points
    Sam,

    I was looking at your schematic again, and I just realized you have the VTT LDO set up as if it is a BUCK. Can you try replacing the inductor with a short and see if that fixes your issue?
  • 0 in reply to Nick Jakse49
    TI__Expert 4155 points

    Sam,

    Any updates on this?