• State
  • Locked Locked
  • Replies 7 replies
  • Subscribers 85 subscribers
  • Views 575 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TMS320F28P550SJ: inconsistent ADC results with preproduction material

Tom Schultz
Expert 4170 points

Part Number: TMS320F28P550SJ

Tool/software:

Team,

Posting on behalf of a customer:

"We noticed a discrepancy in the ADC counts for the same input between our TI F28P55X control cards.  Both use "X" prefix devices.  One card is labeled MCU132E1 and the other MCU132A.

 The boards are running the exact same code on them and yet report different ADC count values for the same input. The ADC analog reference is configured for VDDA. Both boards measure the same VDDA at 3.38V (powered via USB-C). The calibration factors are loaded by calling Device_cal().

 Below is a table showing the difference in ADC counts for the same ADC input between the old board and new board.

 

ADC input

Old board ADC counts (MCU132E1 )

New board ADC counts (MCU132A)

1VDC

1333

1258

2VDC

2657

2509

3VDC

3983

3762

 

I used 'old' vs 'new' above because all the boards from an older PO behave in the same way and all from a newer PO behave in the same way.

 

We found some physical differences between the boards as well – pointing to a revision change,  MCU132E1 vs MCU132A.

 We are hoping to understand this discrepancy among the same micros as we will be using these in the future."

Thanks,

Tom

  • 0
    TI__Mastermind 44830 points

    Hi Tom,

    Talked to the control card designer and learned that there was a routing improvement to avoid crosstalk on some of the channels that was implemented on Rev A.  Would you be able to get ADC channels used by the customer during the conversion?

    Thanks,

    Joseph

  • 0 in reply to Joseph Casuga
    TI__Expert 4170 points

    Hi Joe, it is ADCA.IN0 and ADCB.IN7.  Please confirm why this issue is occuring and if there's a way to correct it on the silicon the customer already has.

    Thanks,

    Tom

  • 0 in reply to Tom Schultz
    TI__Mastermind 24811 points

    Hi Tom,

    On the F28P55X controlCARD, there were some adjustments made to the analog signal routing to improve the performance on some analog signals with Revision A. Specifically, this is relating to signals A2, A7, and A15 (and other ADC signals routed on those same pins). So this may be why you are seeing differences in the analog performance across the two controlCARD revisions. Do you have access to an additional controlCARD device to compare to? As well, how are you delivering the analog signal to the controlCARD?

    Regards,

    Peter

  • 0 in reply to Peter Luong
    TI__Expert 4170 points

    Hi Peter,

    Here is the response from the customer.

    For my ADC test, I simply provide 1V, 2V and 3V DC output from a function generator to the same pin on the Dock board and the GNDs being common among the micro, function generator and scope. Then I swap out the control card from the dock. I performed this test on two boards, rev E1 vs rev A, one of each.

     

    This table is from the first email in this thread:

    Input to ADC

    Old board (rev E1)

    ADC counts

    New board (rev A)

    ADC counts

    1VDC

    1333

    1258

    2VDC

    2657

    2509

    3VDC

    3983

    3762

     

     As for comparing to additional boards – about a month ago, we developed an application code that takes analog input from ADC and calculates a Vrms value. Across 6 – 8 boards between the team members, the boards which were rev E1 reported a value with a consistent ~5% error, whereas rev A boards reported the expected Vrms value.

     

    In both scenarios we observed a discrepancy of ~5% between the two board revisions.

    Does this align with what the C2000 is expecting?

    Thanks,

    Tom

  • 0 in reply to Tom Schultz
    TI__Mastermind 24811 points

    Hi Tom,

    Thanks for the information from the customer. This information seems like a reiteration of the previous data but I still don't see which exact ADC channels they are using in their system. The E1 are engineering revision units with preproduction silicon so it is expected that ADC performance would not be as good as the A units which are first revision . Do note that if customer wants additional rev A units, please have them put a note to procure the units directly from our Singapore warehouse and that should be sufficient.

    Regards,

    Peter

  • 0 in reply to Peter Luong
    TI__Expert 4170 points

    Hi Peter,

    I believe I had already mentioned the ADC channels previously in the post unless I'm missing something.

    "ADCA.IN0 and ADCB.IN7 on the micros."

    Would these channels mean this is a known issue or something new? 

    -Tom

  • 0 in reply to Tom Schultz
    TI__Mastermind 24811 points

    Hi Tom,

    The ADC channels you mentioned don't have any alterations in the board signal routing between Revs E1 and A, but the test coverage between the two devices in terms of the actual c2000 device is different, with the Rev A having greater test coverage. This would mean that the device is likelier to have better ADC performance. Recommend to order samples from Singapore PDC if customer is needing additional boards of Rev A

    Regards,

    Peter