TPS65987D: Sometimes won't sleep and power consumption is still large

Hi Maurice,

The device will not go to sleep unless there is no I2C traffic or PD traffic. If there is any I2C traffic, the device will wake up from sleep mode. Please ensure there is no I2C communication or PD communication when you are checking if the device is in sleep mode.

If this answers your question, PLEASE select This resolved my issue

Thank you,
Eric

Hi, Eric-san,


>Please ensure there is no I2C communication or PD communication when you are checking if the device is in sleep mode.


Yes, of course we know it.

We've checked there's no device attached, and there's no I2C (Both of 2 I2C buses) & PD communication,
but power consumpition is still large (like Active or Idle mode, around 19mW).


So I'd appreciate if you TI engineers reproduce by TPS65987-90EVM or other validation environment.


Best Regards,

Hi, Eric-san,


Today, we've reproduced this phenomenon by using TPS65987D-90EVM.


<Reproduce Procedure by TPS65987-90EVM>
　(1) Attach barrel jack AC adaptor to EVM.
　　-> TPS65987DDH become “Sleep” state.
　　　　And you will see about 25.5mA in P3V3 line.
　　　　(We measured P3V3 current by L3.)
　(2) Attach type-C source device, for example, PD adaptor.
　　-> TPS65987DDH become “Active” state.
　(3) Remove type-C source device.
　　-> TPS65987DDH become “Sleep” state.
　(4) Repeat (1) & (2) again & again, and next remove type-C source device.
　　-> You will see about 31.0mA in P3V3 line, so 3.3V x 5.5mA = 18.15mW increased,
　　　　even though there?s no Type-C device and I2C communication & PD negotiation.
　　　　This phenomenon means TPS65987DDH won?t become “Sleep” state.

And I'd appreciate if you tell us the countermeasure against this problem.
and if countermeasure is related with base firmware,
please revise the base firmware by the begining of March 2019.


Note:
Unfortunately, as we cannot send you private message,
we cannot share project file & detail explanation.
But, for detail, we've told your colleague Yone-san,
so please ask Yone-san.


Best Regards,

Hi Maurice,

I am working to reproduce the issue on my end. Can you test to see the following:
- Once the issue is reproduced, disable sleep mode and re-enable it.
- Does the PD controller return to sleep mode?

Thank you,
Eric

Hi, Eric-san,


>Can you test to see the following:
>- Once the issue is reproduced, disable sleep >mode and re-enable it.
>- Does the PD controller return to sleep mode?

We tried changing these registers
System Power State 0x20: 00<--> 05
Sleep Configuration 0x70: 00<--> 05
but TPS65987DDH won't enter "Sleep" state.


And also, we found TPS65987DDH could recover and enter "Sleep" state
if we did (1) or (2)
(1) HW reset by pushing S2 switch.
(2) SW reset, for example 4CC command "Gaid".


But, in our product design,
both (1) and (2) aren't useful as the countermeasure,
because some of our products contains Thunderbolt controller.
And TPS65987DDH & Thunderbolt controller use common SPI ROM.
(Namely, their firmwares are merged.)

So when we reset TPS65987DDH,
there's a possibility that SPI communication between ROM and TPS65987DDH / Thunderbolt controller conflicts,
which leads to system hangup.


Best Regards,

Hi, Eric-san,


I'm very sorry, but there's typo in reproduce procedure by using TPS65987-90EVM.
Let me correct it.

Incorrect: (4) Repeat (1) & (2) again
Correct: (4) Repeat (2) & (3) again


<Reproduce Procedure by TPS65987-90EVM>
　(1) Attach barrel jack AC adaptor to EVM.
　　-> TPS65987DDH become “Sleep” state.
　　　　And you will see about 25.5mA in P3V3 line.
　　　　(We measured P3V3 current by L3.)
　(2) Attach type-C source device, for example, PD adaptor.
　　-> TPS65987DDH become “Active” state.
　(3) Remove type-C source device.
　　-> TPS65987DDH become “Sleep” state.
　(4) Repeat (2) & (3) again & again, and next remove type-C source device.
　　-> You will see about 31.0mA in P3V3 line, so 3.3V x 5.5mA = 18.15mW increased,
　　　　even though there?s no Type-C device and I2C communication & PD negotiation.
　　　　This phenomenon means TPS65987DDH won?t become “Sleep” state.


Best Regards,

Hi Maurice,

Thank you for the instructions on how to reproduce. I am still working to reproduce this issue on my side. Did you make modifications to the EVM in order to measure the P3V3 current? If so, can you explain those modifications to me?
I am opening an internal investigation into this as well to determine the cause.

Thank you,
Eric

Hi, Eric-san,

Thank you for your reply.

> Did you make modifications to the EVM in order to measure the P3V3 current?

> If so, can you explain those modifications to me?

Yes.

We connect L3 ( in 3.3V output DC/DC)  and P3V3 output with multimeter

so that we can measure P3V3 current.

First, we have to remove L3 from EVM, and next we have to replace L3 upward from EVM.

I'll attach picture, so I'd appreciate if you confirm it.

Best Regards,