TPS65981: Variable sink requests operating current greater than both available current and requested maximum current

When using a variable PDO with a TPS65981 configured to negotiate across a range of currents the chip has been observed to request an operating current greater than both the current the source advertises it can supply and the maximum operating current requested in the same message. The GiveBack flag is not set so these values should be the same. Because the requested operating current is too high the source (correctly) rejects the request and advertises it’s capabilities again causing the whole process of incorrect requests and rejects to continue ad infinitum.

My test setup is using a TPS65981 EVM as a power sink (firmware configures it as DRP but only the sink functionality is relevant to this issue) and a TPS65987EVM as a power source. I’ve included the firmware files used for both boards. PP2 of the TPS65987 is connected to the internal DC-DC  converter and PP1 is connected to a bench supply being used to emulate the variable supply. I’m configuring both boards with the TPS65981_2_7_8 Application Customization Tool, GUI Version :  6.1.3.

The source PDOs being made available from the TPS65987 are as follows:

The sink PDOs used by the TPS65981 are as follows (all are configured to “Ask For Max” and the system is configured to negotiate for maximum power):

In this circumstance I’d expect the TPS65981 to negotiate to use the battery PDO with either PDO5 or PDO6 (preferably PDO6 as this would ensure the highest minimum power availability). I’d expect the chip to request both a maximum operating current and operating current of 3A. This behaviour of limiting both values to the maximum available from the supply is shown when negotiating for a fixed supply with a source current limit below the maximum current of the sink PDO. The maximum operating current negotiated for is the correct 3A but the operating current requested is 3.32A.

The source of the 3.32A number is unclear. Modifying the every current value across all three variable PDOs does not impact it and neither does changing the “Capabilities Mismatch Determination” (under autonegotiate sink) from automatic to 30W.

Any guidance on how I can modify the configuration to ensure the chip operates in the expected manner or troubleshooting advice would be appreciated. I’d also be grateful if anyone could pin down the source of the duff 3320mA number, it’s really got me scratching my head.

I’m aware that this chip in not recommended for use in new products and cannot be certified, but this change to functionality is necessary to support some existing equipment. A hardware change is unfortunately not viable at this time.

VariablePDOSink1.pjtVariablePDOSource1.pjtVariablePowerNegotiation.csv