TPS65070 current differs on AC versus USB pin

Mike Looijmans said:

.

When I power the system via the USB pin, using a source capable of supplying at least 1500mA, and set the current limit for the USB to 1300 mA via I2C, the resulting charge to the battery - as reported by the gauge - is about 490mA. The board itself consumes around 280mA from the TPS chip.

If i connect the same source to the AC input, and also set the AC current limit to 1300mA via I2C, the resulting charge to the battery is 600mA.

Why? This does not make any sense to me, the charge current should be equal, regardless of whether the USB or AC input is being used. That is what I understand from the datasheet.

[D.A.] Have you confirmed the current measurements with a multimeter? Also note that the TPS6507x input current limit has a range from 1000mA to 1300mA. Also, charge current will automatically throttle back if the SYS voltage drops toward under-voltage lockout (UVLO). Please measure the voltage at SYS when troubleshooting. The charging current also depends on the voltage of the battery and the state of the charger per the description on page 25 of the datasheet. Also, verify that your Riset is for your desired charging current and that the battery spec can support.

Other observations:

- Changing the current limit to 500mA via I2C limits the battery current to ~200mA, as is to be expected. Both AC and USB behave similar then.

- Applying both AC and USB with a limit of 500mA also results in only 200mA to the battery. It appears that the chip does not "mix" or "add" both AC and USB inputs, is that correct? The datasheet appeared to suggest the opposite

[D.A.] The TPS6507x will use either AC or USB, not both. 

- Setting a 500mA on the USB input, and enabling a few periferals to increase the load will make the system fail much earlier than expected. Measuring the input with a digital multimeter (on the cable) reveals that the current into the system is only 420mA. Raising the USB input limit to 800mA resolves the system failure, even though the input current does not exceed 500mA at all. Is this a bug in the chip (we have a dozen boards that behave the same) or is there some other factor that we need to take into account? 

[D.A.] Again, there is a range for the input current limit, Riset is a factor and the SYS voltage going to UVLO will throttle charge current. There are no known bugs with the IC. When you mention system failure, you should define failure and support with analog oscilloscope screen shots. Additionally, there are speical support forms for the Logic PD boards: http://tdg.logicpd.com/