• State
  • Locked Locked
  • Replies 4 replies
  • Subscribers 83 subscribers
  • Views 253 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.

TMS320F2808: VADCREFM rises 10mV when ambient temperature rises from 10C to 50C

Howard Zou
Genius 12365 points
Part Number: TMS320F2808

Hi,

my customer have been using TMS320F2808 for long.

Recently they've got a batch of board whose ADCRESULT will change for a fixed sampled voltage when temperature changes.

And we have found out that it's caused by the change of VADCREFM when temperature changes.

VADCREFM is 0.56V at 10C, and 0.57V at 50C.

VADCREFP is 1.31V at 10C, and 1.31V at 50C.

They want to do failure analysis since only this recently batch has such problem, but I'm not sure if the voltage we measured is out of spec since we only have typical value for internal voltage reference temperature coefficient in the datasheet.

  • 0
    TI__Guru* 80080 points

    Howard,

    Can the customer clarify the change in the ADCRESULT value from 10C to 50C?  It would also help to give the ideal input voltage/ADC code(i.e. if there was no gain/offset error)

    We expect their to be some change in gain error(and smaller movement with offset error) with changing temperature, and voltage on ADCREFP/M will move accordingly.  

    We need to understand if the device is out of spec for the gain or offset error listed in the DS at +/-60LSBs each.  

    The TYP temp co-efficient is listed at 50ppm/degC.  This would mean that for 40C movement we should expect 2000ppm movement.  This would translate into about 0.2% potential change in gain error or about 8 codes at 3.0V input.

    Agree that it is pre-mature to submit for failure analysis until we determine if the device is still in specification.

    Best,

    Matthew

  • 0 in reply to MatthewPate
    TI__Genius 12365 points

    Matt,

    For example, if we have VIN=1.5V with internal reference, the ideal ADCRESULT=2048.

    If we got the real ADCRESULT=2000.

    Then the error is 48.

    But the error is composed of gain error, offset error, INL and DNL.

    1. Then how can I know if the gain error is out of spec?

    2. Besides, since the gain error is as large as 60LSB, it means that if we input 3V, the ADCRESULT=4036 (4096-60) is still acceptable. The error is too big. How can we reduce the error?

  • 0 in reply to Howard Zou
    TI__Guru* 80080 points

    Howard,

    Your point #1 is correct, we would need to do a 2 point calculation to determine what error is gain and offset.  The INL is capped at 1.5LSBs, so it will be a very minor contributor.

    In your example the ideal output code is 2048LSB and measured is 2000.  In this case we can take another point, perhaps 0.5V which ideal is 683LSB and get the measured value.

    Once we have this we can substitute into the standard slope/intercept formula y = mx+b, where m will be gain and b the offset.

    In terms of temperature movement, do we know how many codes the customer observes from 10C - 50C in change?  Or is both the 10C and 50C values un-acceptable to them.

    Best,
    Matthew

  • 0 in reply to MatthewPate
    TI__Genius 12365 points

    Matt,

    it moves from 2040 to 2047. But they haven't tell me the exact input voltage at the ADC input pin because they haven't measured it, we only know that 7LSB correspond to 700mA, they are measuring the current in their power stage, and this 700mA change is unacceptable.

    They are concerned if this change with temperature is within F2808's spec, then they may be using an inappropriate controller for their equipment for long.