• State TI Thinks Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 1 answer
  • Subscribers 32 subscribers
  • Views 243 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.

TMP119EVM: Accuracy, or Bad Sensor?

David Elsbernd
Prodigy 10 points
Part Number: TMP119EVM
Other Parts Discussed in Thread: TMP119,

Tool/software:

We have two TMP119 sensors from the evaluation module, but with the sensor detached at the perforations and wired to our own interface.  I was comparing them to several J/K thermometers.  The three thermometers are in general agreement, whereas one TMP119 is 0.3C below the thermometers and the other is 0.6C above the thermometers.  I know the J/K thermometers are not too accurate, but expected the TMP119 sensors to be nearly identical in their readings.  Do we have a bad sensor?  If so, it is impossible to tell which one is bad from my data, if not both.  

  • 0
    TI__Expert 3585 points

    Hi David, 

    It's possible that the difference in sensor readings could be caused by a real temperature gradient in the testing environment. Could you share any picture of the test setup, or describe it?

    Temperature gradients can occur in ambient conditions, ovens, and environmental chambers, and may be caused by localized hot/cold spots (such as an oven's heating element, a board hot spot such as LDO or MCU, even a human). Our R&D team uses an oil bath which is ~0.005mC stable and uniform, this allows us to properly test the temperature sensor accuracy.

    Best,

    Sakeenah 

  • 0 in reply to Sakeenah Khan
    Prodigy 10 points

    Hello Sakeenah,

     I’d like to have an oil bath!  As you can see by the picture, I taped the two sensors and the two temperature probes to a cardboard box, with a piece of paper underneath the probes to ensure they were sitting off the cardboard in free air as are the TMP119s.  The fifth thermometer has the sensor buried in the white case, so it has a very slow response, but it seems to mirror the other thermometers.

     Here are some of my readings:

    Comparison to External Meters

    Sensor/Interface

    TMP119

    Govee

    Fluke 52

    Agilent U1272

    1

    1

    25.43

    24.7

    24.8

    24.6

    1

    2

    25.41

    24.8

    24.6

    24.7

    2

    1

    24.47

    24.7

    24.7

    24.7

    2

    2

    24.48

    24.8

    24.7

    24.6

     

    There are extra readings as I swapped interface cables to make absolutely sure there was no external cause for the unexplained readings.

     My initial data collection over a period of 2 minutes each sensor show a dropping temperature on sensor 1 but I think that was due to the room air conditioner turning on.

     .

    You can see the four sensors attached to the cardboard box, all in close proximity.  We need to know if the TMP119 is accurate as specified, can you explain the discrepancy?

    David Elsbernd

  • 0 in reply to David Elsbernd
    TI__Expert 3585 points

    Hi David,

    Thank you for sharing the test setup and data. 

    All TMP119 units are tested on a NIST traceable production setup and calibrated to ensure that the temperature accuracy performs as specified in the datasheet (0.08C max temp error from 0-45C).

    The discrepancies are likely caused by the reference thermometer accuracy/resolution and ambient fluctuations/gradients. The TMP119 has higher accuracy than the 3 references, so it is isn't possible to verify the TMP119 accuracy with those references; whether they match ends up being arbitrary. In addition, since the test environment is open and uncontrolled, this will cause the ambient fluctuations/gradients which would further mismatch measurements between the sensors (especially if there is AC on and such). To test the accuracy properly, you'll need a stable and uniform (controlled) environment and a reference sensor that is at least more accurate than the TMP119. You could even try just using a sealed box or something to cover the sensors and let them reach equilibrium for some time to see if that helps the measurements settle (make sure there are no heating sources nearby as well). 

    Good luck! 

    Best regards,

    Sakeenah 

  • 0 in reply to Sakeenah Khan
    Prodigy 10 points

    As you suggested I put the sensors in a box, internally wrapped in foam, and all sensors within 1" of each other.  Here are the pics:

    Sensor # Soak Time Hours Temperature TMP119 (°C) Fluke 1504 Error from Fluke Meas. (°C)
    1 4 24.7 24.5791 0.121
    2 4 23.67 24.7255 -1.056

    Maximum allowable error - 0.08C for TMP119, 0.02C for Flule 1504 plus 0.02C for drift over time = 0.12C.

    Sensor 1 is acceptable; the thermometer is at the end of its calibration cycle, but it seems to confirm the TMP119.

    Sensor 2 is clearly 'off' by a degree Centigrade, and that measurement is repeatable.

    Question 1 -- will you replace the bad sensor?  We can send the bad part back for your analysis.

    Question 2 -- If a bad sensor can slip through, what assurance do we have that the final yield on TMP119 installed on our modules will be acceptable?  Do you have any data on that?

  • 0 in reply to David Elsbernd
    TI__Expert 3585 points

    Hi David, 

    1C error between the 2 sensors in the same box is definitely quite high, I wonder if that's caused by the foam or a heat spot close tot he box. I'm not sure the sensors should be wrapped in foam, I would just let them sit in stable air inside the box. 

    The production test fails/discards any unit that does not conform. This device is also NIST traceable, so you can share the Unique ID for your reported bad unit with me, and our team can double-check the production data.

    You may submit a customer product return (CPR) so that TI can evaluate the reported issue. Please follow the instructions here to submit a return: https://www.ti.com/support-quality/additional-information/customer-returns.html. In the case of a real quality issue, TI can offer a refund or replacement. 

    Best,

    Sakeenah 

  • 0
    TI__Prodigy 20 points

    Hi David,

    I am a Field Quality Engineer. I received your return request CPR251125891. The CPR form is for device returns, so the Quality team cannot help with issues on the TMP119EVM. Sakeenah's team will need to help with next steps.

  • 0 in reply to Maaha Sakhia
    TI__Expert 3585 points

    Hi David,

    Sorry for the internal confusion regarding CPR. The Temperature & Environmental Sensing product line's application team can help with the issue. I will reach out to you shortly via email for the next steps.

    Best,

    Sakeenah

  • 0 in reply to Sakeenah Khan
    Prodigy 10 points

    Thanks, Sakeenah, I'll look for that email.