topic Re: Internal Temperature sensor measurement offset in Kinetis Microcontrollers

Internal Temperature sensor measurement offset

abhishek_sinha — Fri, 20 Sep 2019 07:01:03 GMT

Hello all,

I have been trying to use the internal temperature sensor for the family Kinetis K26. Though i was able to make it work successfully, yet, the measurement that I get has a huge offset from the actual temperature. I am also using an external temperature sensor, and the results from both of them are utilised. My question is :

1. Does calibration play any role in the behaviour that i see? Because the i got the same results with and without the calibration.

2. What could be the actual reason behind this? I know by changing the offset i can get a workaround, but logically speaking, internal temperature sensor monitors the core thus should be at a higher temp compared to external one.

Note: this external temp sensor is on the same board, and that too in close proximity of the processor.

I have attached an image file, showing the behaviour.

Thanks in advance.

Re: Internal Temperature sensor measurement offset

jingpan — Mon, 23 Sep 2019 07:28:00 GMT

Hi,

I tested on my TWR-K65F board, it seems work fine. There isn't such a big deviation. Please see my attached code. If it is not a software problem, please check your Vref and VDDA. A small input offset may cause a big deviation.

If you want a higher precision, please check AN3031. The slope value and Vtemp25 is get at VDDA=3v. You should do some calibration.

Regards,

Jing

Re: Internal Temperature sensor measurement offset

abhishek_sinha — Wed, 25 Sep 2019 09:26:45 GMT

Hello Jing,

I did nearly the same thing but using my functions which directly interacted with the register. Yet I see the same behavior as described in the original problem.The functionalities that I have changed are the following :

1. Used the calibration during the initialization

2. Used 16 bit conversion

3. Used 16 samples hardware enabled average.

Re: Internal Temperature sensor measurement offset

jingpan — Fri, 27 Sep 2019 07:45:47 GMT

Hi,

I guess this may because the real VDDA is not 3.3v, for example, 3.26V. This value is very sensitive. A very slight change will cause several degree deviate.

Regards,

Jing