AD7708 strange behaviour at cold temperatures

 

Hello there,

With the AD7708, we are experiencing a strange behavior with sudden changes of temperature.

If there is a rapid cooling of the device down to -40 Celsius, it happens that a reading is completely erroneous.

This behavior happened when verifying the robustness of our product in thermal shock.

We have been able to reproduce the problem with a can of spray freeze.

 

I have read this in the datasheet for the calibration: 

 

 

 

“It should also be noted that to optimize@@@

calibration accuracy, all AD7708/AD7718 ADC calibrations

are carried out automatically at the slowest update rate with

chop enabled.”

 

Does it mean that the SF must be set to 255 to avoid this error ? 

Please advise,

Teodor

 

  • 0
    •  Analog Employees 
    on Oct 5, 2015 9:20 AM over 5 years ago

    Hi Teodor,

    To be able to help debug this I'm going to need a bit more details of the issue.

    1. What was AD7708's mode of operation during this thermal shock test? (e.g. continuous read back?)
    2. Did calibration took place during this test?
    3. when you said "reading is completely erroneous", do reading stay erroneous after the temperature has settled? or do data recover?
    4. did you check the Status Register after experiencing the the erroneous data?

    thank you.

    Ren

  •  

      1. What was AD7708's mode of operation during this thermal shock test? (e.g. continuous read back?)

        Single conversion mode since we are reading all channels one after the other.

      2. Did calibration took place during this test?
        1. No.
      3. when you said "reading is completely erroneous", do reading stay erroneous after the temperature has settled? or do data recover?

      

    Very rarely, a sample is erroneous.  It recovers quickly (the next sample).  This is a problem when looking at a peak value from the sampling.

     

      1. did you check the Status Register after experiencing the the erroneous data?

        No, we don’t need to because, in the conversion mode, no status bits are useful: we use the RDY pin to know when the data is available instead of the RDY bit; the CAL bit is always zero, the ERR bit is only set high if the data is all zeros or ones which we easily see in the data itself, and we don’t care about exact sampling frequencies so the LOCK bit is not useful.

      

  • 0
    •  Analog Employees 
    on Nov 2, 2015 11:33 AM over 5 years ago

    Hi Guys,

    This type of tests may be too stringent for most of the semiconductor devices.

    may things can happen during thermal shock, including mechanical stress from package/PCB expansion contraption.

    some of the automotive and military test standards has the thermal shock test requirement, and these standards only require for the device to recover after the temperature has stabilized, rather than maintain its performance during it.

    do you have a specific standard that you want to meet with this product?

    Ren

  • 0
    •  Analog Employees 
    on Aug 2, 2018 3:11 PM over 2 years ago
    This question has been assumed as answered either offline via email or with a multi-part answer. This question has now been closed out. If you have an inquiry related to this topic please post a new question in the applicable product forum.

    Thank you,
    EZ Admin