• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 30 subscribers
  • Views 1521 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.

TMS570LS20216 ADC Self-Test Vu=Vn=Vd?

Michael Yeh2
Prodigy 30 points
Other Parts Discussed in Thread: TMS570LS20216

Hi,

My workplace has developed a subsystem based around the TMS570LS20216 MCU. When testing on our hardware, I am finding that when I attempt a self test on the ADC modules, my Vu, Vn and Vd are all approximately equal to each other for ADC input channels that are connected to a device (in our case, a differential chip). They are approximately equal to the voltage being applied to the input of each ADC channel from the differential chips. Because of this, Vu is sometimes slightly less than Vn or Vd and Vd is sometimes slightly greater than Vn or Vu and as a result, we fail the self-test.

I found this post that has some relevance to the problem I am experiencing: http://e2e.ti.com/support/microcontrollers/hercules/f/312/t/200346.aspx

I have also tested our ADC self-test code on a TMS570 MCU Development Kit, specifically on the ADC channels connected to the temperature and light sensors, and see the correct results (Vu slightly less than AD_REFHI, Vd slightly greater than AD_REFLO and Vn in the middle), so I am convinced there is likely a hardware issue with our subsystem board.

I am wondering if there is a minimum output impedance for the driving buffer on the ADC input channels, in order to achieve appropriate ADC self-test results? As I understand it, the lower the impedance of the driving buffer, the less effect AD_REFHI and ADREF_LO have on Vin that is converted by the ADC core.

Please advise on any suggestions or requests for additional information.

Thanks,

Mike  

 

  • 0
    TI__Genius 11645 points

    Hi Mike,

    I will try to get one of our ADC experts to help you with.

    In the meanwhile can you pls help with what's your Vn=? when Vu=Vd=Vn

  • 0 in reply to KARTHIKEYAN RAJAMANICKAM
    Prodigy 30 points

    Hi Karthik,

    Thank you, I'd greatly appreciate your help. Typically, Vn (along with Vu and Vd) is equal to the voltage that is being applied to the ADC input from an external device. I've included a few samples I reported while troubleshooting the self-test:

    Trial 1: Vn = 0x054A Vu = 0x0548 Vd = 0x0542

    Trial 2: Vn = 0x054C Vu = 0x0545 Vd = 0x054C

    Trial 3: Vn = 0x054F Vu = 0x054F Vd = 0x054C

    Trial 4: Vn = 0x054F Vu = 0x054C Vd = 0x054D

    Trial 5: Vn = 0x0547 Vu = 0x0553  Vd = 0x054C

    Please note, I've tested many more than just 5 trials (on various ADC channels) of this and all appear to reflect the same results, wherein Vn, Vu and Vd are all approximately equal to the voltage on the input pin. I've also re-run the self-test with with different voltages and Vn, Vu and Vd continue to be approximately the same and equal the new voltage being applied.

    Thanks,

    Mike   

     

  • 0 in reply to Michael Yeh2
    TI__Genius 11645 points

    I'm sure you have made sure all the following four pins are driven appropriately

    VCCAD, VSSAD, ADREFHI, ADREFLO

  • 0 in reply to KARTHIKEYAN RAJAMANICKAM
    Prodigy 30 points

    Yes, these pins were checked and are driven appropriately.

  • 0 in reply to Michael Yeh2
    TI__Mastermind 49120 points

    Hi Mike,

    Sorry for the delay in getting to this question.

    Consider the ADC input switch equivalent circuit:

    When the ADC input channel selftest mode is enabled, a reference voltage (VREFHI or VREFLO) is superimposed on the input signal. The above diagram shows when VREFHI is selected as the reference to be used. Depending on the reference voltage selected, a partial charge/discharge of the external input capacitance Cext will occur through internal calibration resistors Rcal (7K || 5K).

    If the bond-wire is intact, you'd expect a small change from normal input conversion results, as you have identified. The change between conversion results Vn, Vu and Vd depends on how long the input channel is sampled. Also, the change in conversion result depends on the large Rcal * Cext time constant.

    If the bond-wire is broken, you'd expect a large change between Vn, Vu and Vd. This change again depends on how long the input is sampled, and is now dependent on the very small Rcal * Cmux time constant.

    Note that the tolerance of the input calibration resistors (+/- 70%) as well as that of the external Cext are critical in determining the actual change between Vn, Vu and Vd.

    Regards, Sunil

  • 0 in reply to Sunil Oak
    TI__Genius 11645 points

    Mike,

    If this answers your question, can you pls close this thread. Thanks.