TPS65987EVM: Damaged when down-stream device requests moving to 9V

Hi Matt,

When you say that the EVM gets damaged, could you describe how so? Does the EVM just not negotiate a voltage or does it not power on when you connect a barrel jack to power it? I may need some PD logs to look into it and also how you are programming it.

Thank you,

Hari

Hi team,

I designed a board using the TPS65987. I figured the best way to proceed, to minimize any chance of my damaging anything as a first time TPS65987 operator, was to set my design aside and test the known good TPS65987EVM. Once I demonstrated I could get proper operation from the TPS65987EVM, then I would apply this learning to my similar design.

I am operating the TPS65987EVM as I received it out of the TI box. I made no jumper changes to it. I wrote no registers. I loaded no firmware. I used the TPS65987EVM is it shipped to me. I have yet to involve the TI Configuration Tool on the TPS65987EVM.

Here is my test procedure

1. I read the TPS65987EVM user manual and verified that the jumpers were (A) set up in the documented default state, and (B) where the state that I needed, with the TPS65987EVM as the upstream device with a down-stream facing port.
2. I connected the AC to DC power supply specified in the TI documentation, a Dell VJCH5 (19.5V 130W).
3. I connected a Passmark PM110 USB tester to the TPS65987EVM with a USB-C to USB-C cable.
4. The PM110 asked for 5V and received it.
5. I noticed right away that VBUS had high ripple, going outside the USB VBUS spec (I changed this part of the design in my design, in part to reduce the VBUS ripple.) The PM110 alarmed intermittent VBUS slightly above limit. I confirmed this with an oscilloscope.
6. The PM110 performed sweep tests 0 to 3A. Worked. No anomalies.
7. I selected 9V from the PM110 control software, and the TPS65987EVM 5V LED went off and the 9V LED went on.
8. A portion of the U2 package became orange hot, popped, a gas jet with small flame shot out of U2, and the TPS65987EVM refused thereafter to communicate with the tester. It had prior.
9. I checked the power supplies and VBUS at each step with an oscilloscope, never saw anything anomalous.
10. I ran the tests over again with an Apple 87W USB-C power adapter, and I was able to test at 5V, 9V, and 20V with no issues, no 15V was offered by the Apple 87W, no power of over 20V and 3 A was offered also - no anomalies. Also, no anomalies with an Apple 30W and Apple 61W USB-C chargers.
11. I ran the tests over again with a second TPS65987EVM, except with a current limited power supply, and the procedure proceeded in the same way with the same failure of the TPS65987EVM at the same point.

Best regards,

Michael

Hi Michael,

Yes I don't think that should happen on the EVM from negotiating a different voltage. It may be a defect. Did the U2 get damaged on the second EVM test as well or it just showed the 9V LED and didn't negotiate? Do you happen to have any PD logs you could share and the scope captures as well? 

Thank you,

Hari

Hi Hari,

Thank you for your help. U2 was damaged on each of the 2 instances of TPS65987EVM that I tested. The PM110 could no longer post the TPS65987EVM "port detected capabilities" after U2 failed.  Many combinations of power down and re-start of all of the devices involved in the test were tried. In every case, the PM110 was never again able to post the TPS65987EVM port detected capabilities. With 19.5V applied to theTPS65987EVM DC power connector, D20 VAR_DCDC green LED remained always illuminated, D19 SYS_PWR blue LED remained always illuminated, with no other LEDs illuminated. When the PM110 was connected, D22 5V, D14 SSMX_FLIP, and D15 SSMX_USB3 white LEDs illuminated with approximately a 2Hz approximately 20% duty flash, all flashing entirely in unison.  I searched the TI documentation for PD logs and see nothing significant. What PD logs do you refer to? What device would produce such PD logs? If any TI component produces PD logs in this test, please let me know how to get these logs. Please suggest how to capture PD logs in this case. Pardon any lack of knowledge I may exhibit regarding the TI PD hardware and software. The Passmark (PM) USB Power Delivery Test software does provide a log window, but the log window merely indicated that 9V had been selected. This simple PM test software has no PD traffic debug or sniff capability.  The tester was able to report the port detected capabilities before U2 failed on the TPS65987EVM. The tester was unable to report the port detected capabilities after U2 failed. The PM110 produced the following to log window.

Voltage:5.00V, Max Current:3000mA.

Voltage:9.00V, Max Current:3000mA. [TPS65987EVM U2 fails] [no further logs are displayed]

Best regards,

Michael

I Hari,

TPS65987EVM D16 VBUS white LED never illuminates after U2 fails. Oscilloscope shows VBUS remains with a few mV of GND while PM110 asks for 5V while D20, D19, and D22 pulse.

Best regards,

Michael 650-862-5163

Hi Hari,

What PD tester do you have success with when operating the TPS65987EVM as I have described?

Best regards,

Michael 650-862-5163

Hi Michael,

For PD logs, I would recommend something like the Total Phase PD Analyzer as one option. It will show the communications between the two devices.

I will look into this a little more and provide feedback.

Thank you,

Hari

Hi Hari,

I appreciate your help. I have requisitioned a Total Phase USB Power Delivery Analyzer to provide you with a log. However, in a central sense your request for a PD log is tangential to my forum post, as the TPS65987EVM should not fail with a catastrophic burn-out no matter what symbols are signaled.  I was hoping for guidance on how to modify the state of the TPS65987EVM so as to avoid it's catastrophic burn-out failure such that I could make progress without spending thousands of dollars buying new TPS65987EVMs. If the tester sourced too much Voltage potential it could trigger a catastrophic burn-out of the TI U2 on the TI TPS65987EVM, but this is orthogonal to your request for PD logs, and not something that this USB-powered USB PD tester is capable of as far as I can determine so far, as it always operates as a current sink, and I have tested the purity of its current sink and have not been able to find it sourcing power. Nor does it provide high Voltage potentials on its CC leads as far as I have been able to determine.

Best regards,

Michael, 650-862-5163

Hi Michael,

I am working with my team to see what could be causing this behavior. I will provide some feedback by the end of this week. If you have scope captures of VBUS and CC lines, please upload them here.

Thank you,

Hari

Hi Hari,

Thank you for considering this post repeatedly. I may have provided a false data point. The Passmark test software may have transitioned from 5V to 9V while at 3A burden, while I had stated 0A. I found a bug in the Passmark test software that occurs when certain circumstances arise, where sometimes it remains at 3A while at other times it moves to 0A when requested to go to 0A at the completion of a load sweep. Questions.

(1) Is the TPS65987D and the TPS65987EVM rated to switch VBUS Voltage potentials simultaneous with full rated port burden (3A)?

(2) Is the TPS65987D and the TPS65987EVM rated to start (power-up) VBUS with up to full rated port burden (3A) at 5V, 9V, 15V, and 20V?

Best regards, Michael, 650-862-5163

Hi Hari,

I have zero remaining TPS65987EVM to test. They all failed as I previously described while being signaled to switch from 5V to 9V. Regarding the down-stream test device, the Passmark PM110 tester current sink, I do not observe it pulling below 0V GND, it implements a first quadrant only current sink, it pulls down to ground only in my testing. I tested a Samsung-based PD power supply and a Microchip-based PD power supply with the PM110, in addition to the 3 Apple PD sources, and they all work properly and avoid failure when the PM110 asks to move from 5V to 9V, even when the PM110 remains at 3A when it requests a move to 9V.

Best regards, Michael

Hi Michael,

The TI devices all follow the USB PD Specification and are compliant and therefore, require the current to decrease when the voltage needs to be changed. Here's the diagram showing the flow from the PD Spec that you may reference to, it is in section 7.3.2:

Thank you,

Hari

Hi Hari,

Thank you for the chart and your help. (1) What is the maximum VBUS capacitance that the TPS65987EVM is rated to switch, for use in the equation at the bottom of the chart that you provided here, for fast role reversal and for other than a fast role reversal? [The USB spec may require less than 10uF on VBUS for fast role reversal applications..] (2) How can TI's customers protect their USB PD designs based on the TPS65987EVM from failure when encountering generally compliant devices with bugs such as too much capacitance or failure to reduce current draw to below the specified level during a VBUS Voltage change?

Best regards,

Michael