TMP006 Data at 41C

Hi Russell,

Let me try to dig in if I can find any data. Will get back to you soon.

Aaron

Hi Russell,

My colleague is able to get a preliminary data for human body temperature range. I am not sure if you have looking for die temperature or object temperature, but below is the data that we collected for one device.

Coefficient calculated for Tobj = 36°C to 40°C, 55mm distance, 10 samples per point, 3.3V

Hope this helps!

Aaron

Thanks Aaron! I just want to make sure I am reading this right. In the top left corner, the temperature of the die is 10 C, and the temperature of the object is 42 C, and the accuracy of the reading is off by 1.48%. Is that right?

Thanks!

Hi Russell,

The temperature in blue is the die temperature, and the temperature in red is the target object temperature. The number inside the box are the temperature error. For example, if the object temperature 42°C, and the die temperature is 10°C then your temp error is 1.48°C.

Note: The data is collected only for one device.
 
Hope this clarify! Let me know if you have any question.

Aaron

Thank you for the clarification! My customer has a few more questions...

1) If die temperature is 45, object temperature 39, what kind of accuracy can we expect?

2) If this sensor is 1mm away from the object, will it have the same accuracy?

3) How does the material properties of the object effect accuracy? Example, any difference if object is shiny plastic or matt plastic? Does the color of the object make a difference?


Thank you very much!

Russell

Hi Russell,

Unfortunately, I do not have any data for die temp = 45°C and object temp = 39°C. Let me discuss with my colleague if he can collect this data, and I will get back to you tomorrow. For best performance we normally recommended the diameter opening of an object is at least 4 times the distance of the sensor. So, if the sensor is 1mm away from the object, this shouldn't be a problem.

The material properties does effect the object accuracy depending on the thickness too. If you know a particular material, I can look it up for you.

Aaron

Hi Russell,

Here is more information for you to question 2 and 3 after discussed my colleague. I would like to correct some of my statement on previous posted.

If TMP00x is 1mm from the target object that customer is trying to measure, the majority of the FOV will be taken up by the object; therefore, this is good. However, the object will radiate heat into the PCB and Die due to the close distance; therefore, the TMP00x local temp will track closely to the device. The closer object temp and die temp are, the better the accuracy. If the customer calibrates one system at 1 time, as long as all the systems will be similar, the new coefficients can be produce greater accuracy across all systems without recalibrate for each system again.

The material affects on the accuracy are all in FOV % and emissivity of the object. If they are trying to measure the temperature of a shiny metal object with emissivity below 0.8 it’s really not going to work. Our coefficients and formula count on a high emissivity material. This can be fixed by painting the surface flat black. Matte finish is better than shiny for sure. Color doesn’t really matter, it’s all in knowing the emissivity of the material.

Will need to get back with you on question 1.

Aaron

Hi Russell,

Regarding to your temperature data for die temp = 45°C and object temp = 39°C, we can add the die temp into the measurement, but we are unable to add the object temp of 39°C because bench characterization is setting up for a 2°C temperature increment starting from 32°C. We hope to start the data collection sometimes next week. Will keep you posted on the progress.

What is the application you are planning to use with the distance of 1mm? Note that 1mm distance with skin measurement might be ended up with a close contact to the die because when placing the distance of 1mm, the skin might create a bump touching the sensor then the distance now may become zero.

Aaron

Hi Federico,

TI recommends designing a FOV restriction into all TMP00x applications. You can read more in the IR FAQ. e2e.ti.com/.../548824

TMP006 and TMP007 utilize the same thermopile design and packaging. Their FOV should be identical.

Thanks,
Ren