TPS65988: Unwanted Attach/Detach events prior to Power Negotiation

Hello,

It seems that VBUS is going overvoltage causing the voltage to crash. Do you have any protection for overvoltage events?

In addition, the following Application Note, "TPS65987DDH Power Path Performance and Protection," may be helpful to understand the overvoltage events.

Thanks,

Mike Emanuel

Please click "Resolved" if this answered your question.

Hi Mike,

Thanks for sharing that App Note. It is very useful. From our measurements it seems like there should be no overvoltage event (VBUS < 5.1V). I also configured a 24V OVTrip point and its usage for testing but seeing the same outcome. I also have voltage warns activated at 10% but those do not fire also. Here is a pic with a negotiation to 20V. What is interesting is that there are multiple changes on the CC line (some accompanied by the VBUS transitions from 0 to 5V back to 0, but some not). These happen for about 4-5s before the PD comm actually starts and the voltage transitions to 20V. Can there be any other root causes?

Hello,

I will address the two captures separately.

First Capture

     It looks like after the 9 V contract is negotiated there is an overvoltage event causing the voltage crash. This is shown by the sudden increase from 9 V on VBUS. This needs to be investigated because the voltage is crashing.

Second Capture

    Because this is a DRP, toggling of the CC lines is indeed normal. Does the voltage remain stable after it comes up to 20 V?

Please capture PD logs for both use cases to further assist in debug.

Thanks,

Mike Emanuel

Please click "Resolved" if this answered your question.

Hi Mike,

It is my mistake for trying to cram 2 things in capture 1. There is no issue once the PD is negotiated to 9V (or 15V or 20V). The drop in capture 1 is an actual physical disconnect followed by a physical connection of the cable to illustrate the variability of the issue. So let's put capture 1 aside.

My question is mostly related to the events prior to any PD packet being sent on the CC lines. It is better reflected in capture 2 where there is a 4 seconds interval until I see the first PD packet on the sniffer. I agree there may be some delays due to the CC lines toggling but 4 seconds seem a lot. Cannot see the same behaviour on the EVM though. 

I can capture the log but not sure how much it will help as it looks clean, apart from the VBUS attach/detach at the start.

Thanks,

Alex

Dear Alex,

Can you program the EVM with the same project that you are using on your system? The EVM may not be a DRP showing the different behavior. Please let me know if the behavior then becomes the same on both the EVM and your design.

Thanks,

Mike Emanuel

Please click "Resolved" if this answers your question.

Hi Mike,

I have flashed the exact project and come to this result: the DK has rarely 1 attach/detach event before starting the PD exchange, so it does not match the same behaviour. Our PCB shows this good behaviour only on the first plugin of the cable. Successive cable plugins show the same 3-4 s period where the CC line toggles until in the end it settles, PD exchange starts and all works well from that point on. 

I attached pictures and log traces showcasing the 2 scenarios (DK and our PCB). Could this be design related? Does TI offer design review?

ti pd.zip

Hi Alex,

I will take a look at your files to see what the issue could be and provide you with feedback next week. If it is not happening on the EVM, it could be a schematic/layout issue. I recommend reviewing Section 11 of the datasheet.

Thank you,

Hari

Hi Alex,

I think you may have to change some configurations in your project if both devices are DRP. Could you please provide your project file that you used as well as the schematic for me to review?

Thank you,

Hari

Hi Hari. That sounds good! Let me send them to you via private message

Thanks,

Alex

Hi Alex,

Sure, I've accepted your friend request.

Thank you,

Hari

Hi Alex,

I have a few suggestions for your design:

• For the OVP, I noticed that you have the limit set to 24V, but in addition you also need to set the usage field in Port Config to be based on the OVPTripPoint as shown below, I think you had it on the 15%. This way the OVP will be set to the TripPoint which is 24V in this case.

• Also, since this is a DRP, you may also need to select the fields, Process Swap To Sink, and Process Swap to DFP in Port Control if you will be supporting those.
• Assuming you have the schematic properly configured as the datasheet recommends, such as having the right amount of bypass caps and such, this could also be a result of the layout of your board.

I would recommend referencing the layout section of the datasheet and making sure your layout also follows those guidelines.

Thank you,

Hari