EVAL-SCS001V1 Failure on short circuit

[jandrews]

We are experiencing failures on some of the STUSB4500 IC chips on production boards and from what we can tell, this is due to the power output (VBUS) being shorted to ground. We observed the STUSB4500 chip failing to negotiate PD, and the 1V2 and 2V7 LDO rails are all at 0VDC. This failure also happens on the EVAL-SCS001V1. When we replace the STUSB4500 IC, with a new one, the problem is resolved, which indicates there is no damage to any other component. This issue may be created with an off-the-shelf EVAL-SCS001V1 PCB connected to a 20V/3A source and shorting the output terminals. The datasheet claims to have short-circuit protection, but it seems to not have this.

 

Our design already has output voltage clamping, ESD protection on VBUS, RPP, UVLO, and OVP, and the correct 1uF decoupling caps on the 1V2 and 2V7 rails, which is more protection than the EVAL-SCS001V1 has. What are some design implementations we can make to prevent this?

[Didier HERROUIN]

Dear jandrews,

To avoid any misunderstanding, The STUSB4500 implements protections against over-voltage on CC pins and over-voltage on VBUS, it does not have any current measurement/protection.
And it continuously monitor VBUS so that it opens the switch when VBUS is out of the expected range (after a short-circuit on Vout for instance)

Can you confirm the 2 use cases below ?

1. On the EVAL-SCS001V1, the STUSB4500 is definitively damaged if you short Vsnk to GND whereas it is connected to a 20V/3A source (do you connect a load to sink effectively 3A ?) 
2. Using your own design, some boards are defective, and the problem is solved when you replace the STUSB4500 chipset. Have you applied exactly the same short-circuit as case 1 ? Or did the problem occurs spontaneously during a normal usage ?

Best regards,

[jandrews]

Thank you for your response. I can confirm that the EVAL-SCS001V1 is damaged when connected to a USB-C PD source with the 20V/3A PDO active and when the load (Vsnk) is short circuited. In this case, the PD source rightly identifies that the sink (STUSB4500) has malfunctioned and disables the power delivery, but after reconnecting, the EVAL-SCS001V1 fails to negotiate PDO again because the STUSB4500 chip is damaged and the LDO voltages have collapsed as previously stated. Replacing this chip solves the issue, so it is definitely the STUSB4500 that failed. I have seen failures on large transient loads as well (On initial connection usually) with the EVAL-SCS001V1. Using my own design, we have seen the same failure mode as the EVAL-SCS001V1 when the load is effectively a short-circuit. We are planning to add an inductor on the source side of the MOSFET downstream of Vsnk in order to help with current transients, and give the STUSB4500 a chance to detect VBUS drop, but is there something better that we could implement?

 

To answer your questions more directly,

1. On the EVAL-SCS001V, the STUSB4500 is definitely damaged when the load sinks effectively 3A. We are not shorting VBUS directly, but rather the source side of the MOSFET (This would be Vsnk in your schematic shown below).
2. Using our design, the boards do not seem to be defective, but we can see the same failure mode from (1) on the eval kit if we sink 3A fast. (Same short circuit case.)

 

[Didier HERROUIN]

Dear jandrews,

This is surprising, as the short-circuit on Vsnk is more likely to damage the MOS FET instead of the STUSB4500. 
On my side, I was not able to reproduce the issue on ISC005V1 eval kit.

When you apply the short, could it be possible that you generate some spurious (for instance, by bringing another GND to the circuit) which may affect the chipset ?

You can raised a support case here: https://ols.st.com/s/ so that we can answer directly to your use case.

You may also want to open a FAR (Failure Analysis Request) with your Distributor if you think it is a defective chipsets lot.

Best regards,

[jandrews]

I agree. At first, I assumed the MOSFET was damaged, but after replacing the STUSB4500 the issue was resolved. Even with the original MOSFET. I was able to reproduce this on two different SCS001V1 boards by shorting Vsnk with a screwdriver across the J10 connector (again, with 20V/3A), but I have not tried with the ISC005V1. The chip lot codes on the eval boards and mine are different, so I assume this is a design limitation rather than a manufacturing defect. Can you please try to recreate this on a SCS001V1 board and share your results?

 

Thank you,