• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 31 subscribers
  • Views 426 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.

TMP117: possible sources for measurement non-linearity

Reinis Veips
Prodigy 30 points
Part Number: TMP117

Hi,

I'm using TMP117 as temperature sensor that measures product's housing temperature to calibrate the temperature-dependence of another sensor (there are actually 4x TMP117 currently in various parts of the housing to determine the best place to measure housing temperature from.

The calibration occurs in a climatic chamber which is stepped from 0..60 deg C (every step is 5 deg C and 90 minutes in duration).

There are three sources of temperature data being logged: climatic chamber's built-in (which I believe is RTD), temperature sensor built-in the sensor being calibrated and 4 of the TMP117's.

The issue I'm seeing is: at the end of the test, measurements of all the TMP117s differ by approximately 4 deg. C from both other temperature sources (chamber and the sensor being calibrated).

In fact, the relationship between the {climatic chamber, sensor being calibrated} and TMP117 is very linear- as if the TMP117 readings were non-linear.

Here is the measurement data (TMP117 temp. is higher than ambient because the data acquisition starts after 1 hour of operation of product, and the housing heats up significantly above ambient temperature):

Tsens[start]=282.96K, Tsens[end]=342.92K, delta=59.96K
TMP117_mean[start]=293.20K, TMP117_mean[end]=349.36K, delta=56.16K
Tamb[0]=273.15K, Tamb[-1]=333.05K, delta=59.90K
sensor 0, TMP117[start]=293.37K, TMP117[end]=349.47K, delta=56.10K
sensor 1, TMP117[start]=292.99K, TMP117[end]=349.34K, delta=56.35K
sensor 2, TMP117[start]=293.31K, TMP117[end]=349.36K, delta=56.05K
sensor 3, TMP117[start]=293.15K, TMP117[end]=349.28K, delta=56.13K


Here is the plot of sensor vs tmp117 temperature and chamber's temperature vs tmp117.
 

TMP117 in BGA package is being used. The PCB it's on has a ground plane with no solder mask, the sensor itself sits in a recessed pocket in the aluminum housing, 
with the ground plane of the sensor being thermally coupled to housing by means of silicone thermal paste (electrically non-conductive).


Now, if it was just the difference between sensor being calibrated and TMP117, I would just assume that other sensor has non-linear behaviour and that's it.
If the housing would indeed increase it's heat dissipation properties (i.e. assuming a constant heat input from electronics)- the other sensor would also register it- but it's registering exactly the same temperature difference than
climatic chamber is.



  • 0
    TI__Mastermind 43815 points

    Hi Reinis,

    I'm not sure I see the problem. How can you be sure the housing is not 4C different from the other locations? Have you considered TMP117 could be better than your other sensors? If you discovered that your other sensors were both NTC based, it could explain the lack of linearity at higher temperature that seems to track. In my experience, though, chambers are designed with thermocouple rated for +/-2.2C. It could be a coincidence they agree.

    thanks,

    ren

  • 0 in reply to Ren Schackmann
    Prodigy 30 points

    The absolute temperature of housing would definitely differ from other locations: but what I am looking at is the steady-state temperature rise (i.e. place the product in chamber at 0 deg C, let it stabilize for 90 minutes, set chamber to 60 deg C, let it stabilize for 90 minutes).

    I would expect the temperature rise of housing match the temperature rise of ambient/operating temperature, given that the products heat dissipation does not change and there is no active cooling happening. What I am seeing is that temperature measured by TMP117 rises by 56 deg C, when the ambient temp. rises by 60 deg C.

    I have now instrumented the same calibration process with another, free standing TMP117 just to make sure that the climatic chamber temperature does indeed rise by 60C, but I'm fairly sure it will (chamber uses PT100 DIN class A RTD and still has a valid calibration certificate).

  • 0 in reply to Reinis Veips
    TI__Guru* 91206 points

    Reinis - 

    PT100 / Class A RTDs are less accurate than TMP117. See Figure 5 here: https://www.ti.com/lit/an/snoaa16a/snoaa16a.pdf  For your reference sensor you need something that is as good as or better than TMP117.