TPS2482: Large offset in shunt measurement from Vin

Hi Erik,

Welcome to E2E!

I do not expect offset to have dependency on the input voltage level. Can you check if there are any leakage paths to OUT. 

Can you remove MOSFET, short OUT to SENSE and then retest ?

Best Regards,

Rakesh

Hello Rakesh, I’ve retested the circuit. First with the pass FET removed.

Vin: 24.006V
ADC Measurement: 0.183A [Shunt register: 00000000 01101110]
Vin: 11.996V
ADC Measurement: 0.090A [Shunt register: 00000000 00110110]
Vin: 09.196V
ADC Measurement: 0.068A [Shunt register: 00000000 00101001]
Vin: 05.996
ADC Measurement: 0.045A [Shunt register: 00000000 00011011]
Vin: 03.019V
ADC Measurement: 0.020A [Shunt register: 00000000 00001100]

Next with the resistor shorted.

Vin: 24.006V
ADC Measurement: 0.183A [Shunt register: 00000000 01101110]
Vin: 11.996V
ADC Measurement: 0.090A [Shunt register: 00000000 00110110]
Vin: 09.197V
ADC Measurement: 0.068A [Shunt register: 00000000 00101001]
Vin: 05.996
ADC Measurement: 0.045A [Shunt register: 00000000 00011011]
Vin: 03.018V
ADC Measurement: 0.020A [Shunt register: 00000000 00001100]

Neither modification seems to have had any effect on the register values.

Here is the implemented circuit:

Hi Erik,

I expect some system offset but what I am wondering is its dependency on the Vin voltage, which it should not. At hand I don't have EVM to verify because this device is low running item. I have placed order for EVM and will take this up as soon as I get it (tentatively a week).

BR, Rakesh

Hi Erik,

I will receive EVM tomorrow after that I will verify and update you by end of this week.

Thanks!

BR, Rakesh

Thank you Rakesh, I will look forward to your results.

Hi Erik,

I have connectivity issue with the TPS2482 GUI which I am debugging now. I will update you in next couple of days. Apologies for the delay,

Best regards, Rakesh

Hi Erik,

One quick question.

Are you using "TPS2482 and TPS2483 EVM GUI" (at http://www.ti.com/product/TPS2483/toolssoftware) to interface with TPS2482 device ? Did it worked ?

Best regards, Rakesh

Hello Rakesh,

Unfortunately I am using my own code to read from the registers directly. Best of luck with the GUI.

Erik

Hi Erik,

Apologies.. I would need some more time to work on this. There is an issue in using GUI, so, I need to interface with USB-I2C to read the registers value.

I will update you by the end of next week.

Best regards, Rakesh

Hi Erik,

Apart from this, do you have any other challenges with this device.

Can you please confirm - Are you in proto-type build phase or validation or production stage ?

Best regards, Rakesh

Hello Rakesh,

Sorry for the delay. I replied to this thread last week but I don't see that reply on the site so I will repost the response.

This issue (ADC readings having a significant offset based on VIN potential) is the only fundamental issue I have observed so far, but I have not been using the device very long. There is an error in the datasheet calculations, but that is a tangential issue.

I am still in the prototyping phase, but will be updating the design soon. I was hoping to get feedback on my implementation in case I missed something that could cause this behavior. If no remedy can be found then I will have to move to a competing device since the measurement error is significant enough to impact the functionality of the board.

Have you had any luck checking the accuracy of the development board?

All the best,

Erik

Hi Erik,

I was trying to get hold of original developer for help but so far no success. Since you are in the initial stage, can you please consider LM5066i.

Please have a look and let me know your opinion.

BR, Rakesh

Hello Rakesh,

I have previous experience using the LTC4282 controller. If we're unable to diagnose the problem then I will use that IC instead.

Thanks again,

Erik

Hi Erik,

Understood your concern. Can you let me know which parameters you care for your system design so that I can pick the right device and provide comparison with LTC4282 

BR, Rakesh

Hi Erik,

As you are in initial design phase, I recommend to consider LM5066i over TPS248x devices. Please let me know if you have any questions on LM5066i

Thank you!

BR, Rakesh

Hello Rakesh,

I appreciate the attempt to help, but as we were never able to determine the root cause for the buggy behavior I don't feel confident switching to another TI device. I will return to the Linear/Analog part (LTC4282) that I have experience using and that I know works as advertised.

All the best,

Erik

Hi Erik,

I used TPS2482 EVM and the TPS2482-3 EVM GUI for verification. The offset of shunt voltage is only +/-2.5uV. I did not see the large offset issue you are observing.

Please find the attached file with results at various Vin voltages..TPS2483 shunt registers.docx

Have you configured the current register as mentioned in the EVM user guide http://www.ti.com/lit/ug/slvu795/slvu795.pdf

Please check and let me know if you have any questions.

BR, Rakesh

Hi Erik,

Can you check at your end and let me know if any thing needs to be verified

Best regards, Rakesh

Thanks for continuing to chase this down Rakesh. I will examine the demo board documentation and see if I can spot any differences.

Thanks Erik.

Hello Rakesh,

I looked at the difference between my implementation and the development board. Removing the filter resistors on the VINP/VINM pins seems to removed the voltage scaling offset. This is slightly odd as the datasheet suggests using the resistors in some cases. However it seems like the sample/leakage current on these pins is significant enough to be a problem.

I still had a 60mA/A scaling non-linearity after fixing the offset but that might be due to a mistake in the layout. I will confirm after testing some additional boards.

Thanks again,

Erik

Ok Erik