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.

TMP006 Calibration

Hi Ian,

I have built a test platform which consisted of TMP006 PCB board and transparent plastic box. The board was wrapped by plastic box. I did the calibration and posted the calibration result in another thread.

After calibration, S0 seemed to be 3.0E-14. So in new test case, 3.0E-14 was used as S0. I put also a cup of hot water on top of box, i.e. hot water was in front of TMP006. The cup was also made of plastic.

I observed a strange symptom. When the water grew colder from 70 degree to 40 degree, the calculation result grew bigger from 14 degree to 38 degree. What was wrong with this test case?

BRs,

Luo 

  • Luo,

    I have a few questions/comments:

    1. Can you post a link to the other thread where you discuss the calibration? I think you have a signal measurement issue if your calculated S0 is 3.0E-14. 
    2. The plastic box is transparent to visible light, but it's actually opaque to IR. This means that all the TMP006 can "see" is the plastic box. So, when you place a hot or cold cup of water on the box, the only thing that the TMP006 is measuring is how the water cup conducts heat into or out of the plastic box. In order to directly measure the water temperature, anything in between the water and the TMP006 would need to be made of IR-transparent material such as polyethylene, polypropylene, or more expensive materials such as sapphire.

    Best regards,

    Ian Williams
    Linear Applications Engineer
    Precision Analog 

  • Hi Ian,

    1. I am sorry that I did not remind you that I have removed the plastic above the TMP006 on the box. The hole diameter is about 4cm. It can guarantee IR transparent. Pls find the link below. I agree with you that it has something wrong with the plastic box or test case.

    http://e2e.ti.com/support/other_analog/temperature_sensors/f/243/t/240447.aspx

     2.  I have done more tests afterwards. I have three findings. Some cases looked better than before.

    a)  If hot water was a little far from TMP006, about 21mm, the calculated result looked well. The calculated result approximated temperature of hot water. I will do further calibration and inform you any progress later on.

    b) If hot water was closer to TMP006, 6mm, the calculated result seemed bad. It seemed that the TMP006 was warmed by hot water and the object temperature increased gradually.

    c) In some cases, the calculated result became 118 degree because of some unknown reason when no object was placed in front of TMP006. I was puzzled.

     This morning, I have found the root cause of (c). It resulted from wrong SW code. (c) was fixed now.

    BRs,

    Luo

     

  • Hi Ian,

    I have tried many times to find best result. Unfortunately, I have not.

    In lastest test case, I decreased the distance from water to TMP006. It was about 13mm.  The hole's diameter was 40mm.

    The calibration was done. You can find the calibration data below. 

    3858.TMP006 Calibration(luo).xls

    It had two problems. Firstly, S0 was not concentrated between 5.0E-14 and 7.0E-14. Secondly,  if water became cooler, S0 became smaller so far as to negtive. 

    I guess, the calculated result is temperature of cup bottom surface. So far, what I measured is hot water. Right?

    Could you please look into the excel sheet and give me some proposals how to calculate water temperature? Thanks.

    BRs,

    Luo

  • Hi Luo,

    Your math in the spreadsheet is incorrect, so your results for the calibration function and (TOBJ4 - TDIE4) are also incorrect.

    Compare the math in your spreadsheet to the one I've attached, which automatically calculates a calibrated S0 for two points. I used the minimum and maximum temperatures from your table and calculated S0 = 7.064E-14. Run the test again with this S0 value and your results should improve significantly.

    Best regards,

    Ian Williams

    TMP006 S0 Two-point Calibration_Luo.xls
  • Hi Ian,

    Sorry, I was spending Chinese new year vocation so that I could not reply it fast. I have looked into your spreadsheet. I have some questions.

    1) In my spreadsheet, first row is physical temperature of water. 16 and 85 are minimum and maximum respectively. Tref is fixed to 298.15 K (25 degree) which is stated in TMP006 SBOU017. But in your spreadsheet,  Tref is set to 16 and 85 degree. Why? What does Tref mean?

    2) In your spreadsheet, C2 is not taken into account. What is impact? I guess it can be ignored.

    3) If I use 7.0E-14 as S0, the calculated result is far smaller than physical temperature. For example, in last column, the calculated result is 42 degree. But physical temperature of water is 86 degree. What is the problem?

    4) Could you please point out what is wrong with my calculation? How to correct it?

    Best Regards,

    Luo Sheng

     

  • Hi Ian,

    Any comments and suggestions are helpful for me to make the project volume production. Thanks.

    Best regards,

    Luo Sheng

     

  • I am looking for the answer here as well. The equations you use in the spreadsheet are significantly different than those provided in SBOU107, Dated May, 2011. Is there a more recent version?

    Here are the differences, as I see them:

    1) Vos - there is a difference in using temperature units in C versus temperature units in K. So it appears that per SBOU107 that Tdie is in Kelvin??)

    Per SBOU107 = b0 + b1(Tdie - Tref) + b2 (Tdie -Tref)^2      where Tref is 298.15

    Per Spreadsheet = b0 + b1(Tdie -25) + b2(Tdie-25)^2

    2) f(vobj) - missing term

    Per SBOU107 = (Vobj- Vos) + c2(Vobj - Vos)^2 (I may have read on another blog that the c2 term is negligible, but I don't think so)

    Per Spreadsheet = Vobj - Vos

    3) Tobj (once again, difference in temperature units. It appears that SBOU107 assumes Tdie is in Kelvin)

    Per SBOU107 = (Tdie^4 + (f(vobj)/S)^0.25

    Per Spreadsheet = ((Tdie+273.15)^4 + (f(vobj)/S)^0.25 - 273.15

    But the real issue I have is with the calibration process. Per SBOU107 Tobj refers to the "actual" temperature of the device, recording with an accurate temperature probe. But, per your spreadsheet, Tobj is the calculated value (per equation 3) above), which is based upon the S0 of the mid-range value of 6E-14. So, we are using an estimated object temperature to calculate a gain coefficient, to then turn around and calibrate the object temperature calculation? Does not make sense to me.

    Also, per SBOU107, the Calibration function is f(vobj) / [1 + a1(Tdie-Tref) + a2(Tdie- Tref)^2)], where Tref is a constant 298.15. But, as the Lou pointed out above, in your spreadsheet Tref is the actual temperature measured at the low and high points by our acurate temperature measurement device.

    Please clarify what is correct. There is much confusion, and I am losing faith in the ability to use these sensors because I cannot calibrate them.

    Thank you and appreciation for a quick resoponse.

  • Hello Joseph,

    Lets go over the answer to your questions one by one...

    1) Vos - there is a difference in using temperature units in C versus temperature units in K. So it appears that per SBOU107 that Tdie is in Kelvin??)

    Per SBOU107 = b0 + b1(Tdie - Tref) + b2 (Tdie -Tref)^2      where Tref is 298.15

    Per Spreadsheet = b0 + b1(Tdie -25) + b2(Tdie-25)^2

    You can use either one of the above equations and it will not change your result because Tref =  298.15K = 25degC. Since you are only concerned with the difference between Tdie - Tref. However I recommend you to stick to Kelvin because we use Kelvin scale to calculate Tobj eventually (equation 4 of SBOU107 document)

    2) f(vobj) - missing term

    Per SBOU107 = (Vobj- Vos) + c2(Vobj - Vos)^2 (I may have read on another blog that the c2 term is negligible, but I don't think so)

    Per Spreadsheet = Vobj - Vos

    As per our experience with the part the C2 could be negligible. However in your application if accuracy is a concern you can surely analyse the effect of C2 with the data you obtain.

    3) Tobj (once again, difference in temperature units. It appears that SBOU107 assumes Tdie is in Kelvin)

    Per SBOU107 = (Tdie^4 + (f(vobj)/S)^0.25

    Per Spreadsheet = ((Tdie+273.15)^4 + (f(vobj)/S)^0.25 - 273.15

    I recommend you to stick to Kelvin scale. Please stick to the Spreadsheet example and we will implement the corrections in the SBOU107 document.

    But the real issue I have is with the calibration process. Per SBOU107 Tobj refers to the "actual" temperature of the device, recording with an accurate temperature probe. But, per your spreadsheet, Tobj is the calculated value (per equation 3) above), which is based upon the S0 of the mid-range value of 6E-14.  

    You are absolutely correct, I apologize for the confusion and please note the following correction. I have changed the TOBJ term marked to TRTD (reference device object temp).

    We will also implement the corrections to the document. Please also see attached for the corrected excel sheet.

    So, we are using an estimated object temperature to calculate a gain coefficient, to then turn around and calibrate the object temperature calculation? Does not make sense to me.

    After the above corrections I guessing its clear to you now, because we are using TRTD (reference device object temp ) to find the S coeff so that we can eventually substitite this into equation 4 of SBOU107 in order to find the Object temperature.

    Also, per SBOU107, the Calibration function is f(vobj) / [1 + a1(Tdie-Tref) + a2(Tdie- Tref)^2)], where Tref is a constant 298.15. But, as the Lou pointed out above, in your spreadsheet Tref is the actual temperature measured at the low and high points by our acurate temperature measurement device.

    You are correct. Please see the attached spreadsheet that I have corrected for you.

    Please clarify what is correct. There is much confusion, and I am losing faith in the ability to use these sensors because I cannot calibrate them.

    My apologies for the confusion and I hope my answers helped you with the understanding. I would like to thank you for bringing this to our notice and we will make the necessary changes to the document.

    1200.TMP006 S0 Two-point Calibration_Luo_corrected.xls

    Best Regards,

    Abhi Muppiri

    Applications Engineer

    AIP- Sensing Products

    Texas Instruments



  • Thank you for your very quick response.

    Please answer another question. Why is it that when we calculate Vos, we subtract 25 from Tdie. So, we are offsetting Tdie, which is in Celcius (as read from register 0x01), by 25 degrees (for whatever reason). So, units are Celcius for Vos. When we calculate Tobj we add 273.15 to Tdie to convert it to Kelvin. Don't we have a unit mismatch here? f(Vobj) is used to calculate Tobj, and f(Vobj) is based on Vos, which is in Celcius. If we use 2 different units in the same equation, it does not make sense to me that we get valid information.

    Your updated spreadsheet does not contain any corrected equation (or any equation) for Tobj. Does that need to be updated as well?

    Thanks

  • Hello Joseph,

    The IR sensor inside TMP006 produces a constant voltage when the ambient/local/die temperature is at 25degC. This offset effect is taken care of by subtracting the Tref term in the Vos calculation equation. Regarding the temperature scale, as I already explained to you please stick to one scale for both Vos and Tobj calculation. I recommend you to you kelvin though.

    As far as calculating VOS is concerned it doesn't matter which scale you use because  b0 + b1(Tdie - Tref) + b2 (Tdie -Tref)^2  gives you the same answer because Tdie - Tref value is same immaterial of the scale you choose.

    The only correction made in the attached sheet is that Trtd is added in place of Tobj to avoid any further confusion. Can you provide more details of your project? This will help me better understand your requirements. Does this go into some sort of end user application or is it like a research/university  project?

    Best Regards,

    Abhi Muppiri

    Applications Engineer

    AIP- Sensing Products

    Texas Instruments

  • Abhi,

    I would be very happy to discuss our goals with you further, in an offline, non-public forum. Here is my email address, please contact me if you can help.

    joseph.waclawski@ge.com

    Thanks

  • Joseph,

    Did you ever get these issues resolved?

    In the last spreadsheet the TI guy sent the final S0 value "RESULT OF CALIBRATION" was obviously in error since it was 2.25e-14.  From the datasheet we find that that value isn't within the range of the device.


    TI never updated the App note since the 2011 version is still on the web site.


    The equations are still missing terms. 

    Vos still has a term of 1000000 that is not explained.

    FVobj still has a term of 1000000 that isn't explained.

    I've made a manufacturing measurement system that works except for the final calibration and my project is on hold because of this.  Any help or insight into this issue would be greatly appreciated.

  • Will,

           I just seen your post and went back trough all the related post.  I think for the most part that the conversaiton was about TMP006.  Are you concerned with TMP006 or TMP007?  If you can give me an idea of what your question is I may be able to help. 

    Regards,

    Tommy Santoyo

    Texas  Instruments

    Sensing Apps Team

  • Dear Abhi,

    You've said: "As far as calculating VOS is concerned it doesn't matter which scale you use because b0 + b1(Tdie - Tref) + b2 (Tdie -Tref)^2 gives you the same answer because Tdie - Tref value is same immaterial of the scale you choose."

    This is correct as long *both* Tdie and Tref have the same measuring units, either celsius, either kelvin.

    However, both excel sheets (yours, named 1200.TMP006 S0 Two-point Calibration_Luo_corrected.xls) and Ian Williams, (TMP006 S0 Two-point Calibration_Luo.xls) are wrong as Joseph Waclawski observed very well.

    On Ian Williams sheet, H11 name has been modified in "Tobj" and the content H12 and H13 fields computed in celsius + kelvin, the same A17 and A18 fields.
    On your sheet, H11 becomes "Cal" and H11, H12 fields where computed in celsius, this is correct, but A17 and A18 fields are still celsius+kelvin.


    Can you provide a true corrected excel for this computation?

    Perhaps a complete document or dataset describing how to calibrate the TMP006?

    I need an example how the numbers from the "no name" table at page 13 of SBOU142-August 2014 (after the table 5 and before the table6....) have been computed?

    I'm waiting for your answer and appreciate your help...

    Vasile