MSP430FR2433: Integrated Temperature sensor is not expcted

Hisao Uchikoshi

Hi Experts,

We are using the integrated temperature sensor within ADC. The blow is the result.

	C	PCB１	PCB2
Measured ADC data	23	664	622
calib30 @ 0x1A1A	30	694	642
calib85 @ 0x1A1C	85	812	768
Calculated temperature [deg C]		16.0	21.3

We followed the sample code and used Calibration data on the TLV. As you can see, even though measured at 23 degree, it shows 16 degree for PCB1 and 21 degree for PCB2, the diff is larger than expected, especially for PCB1.

Are these data expected one or any other better way to get more accurate temperature data?

Regards,

Uchikoshi

over 6 years ago

0 Wei.Jetim Zhao over 6 years ago

TI__Genius 13215 points

Hi Hisao,

The temperature sensor offset can vary significantly. TI recommends a single-point calibration to minimize the offset error of the built-in temperature sensor.

So, what's the application for temperature sensing? If it requires for a high accuracy temp data, I would suggest to add an external temp sensor.

0 Hisao Uchikoshi over 6 years ago in reply to Wei.Jetim Zhao

TI__Genius 15370 points

Hi Wei,

Even though we are using TLV data, does customer need to calibrate it by themselves? The data on TLV can be used to estimate the slope and the offset should be calibrate at a single-point, correct?

Please see note.

Regards,

Uchikoshi

0 Wei.Jetim Zhao over 6 years ago in reply to Hisao Uchikoshi

TI__Genius 13215 points

Hi Hisao,

Please ignore the TLV data if a high accuracy is required. Users should do the single-point calibration it by themselves.

I would give you a reference flow for the calibration.

0 Hisao Uchikoshi over 6 years ago in reply to Wei.Jetim Zhao

TI__Genius 15370 points

Hi Wei,

Thank you for your explanation.

Which document did you get the flowchart?

I want to confirm where the value "0.00355" is coming from.

Regards,

Uchikoshi

0 Wei.Jetim Zhao over 6 years ago in reply to Hisao Uchikoshi

TI__Genius 13215 points

Hi Hisao,

I created the flow chart for you. The parameter "0,00355" is from the section2.2.9 of the device User's Guide.

0 Wei.Jetim Zhao over 6 years ago in reply to Hisao Uchikoshi

TI__Genius 13215 points

Hi Hisao,

Please measure the signle-piont Vcali for each chip instead of reading the TLV data directly.

0 Hisao Uchikoshi over 6 years ago in reply to Wei.Jetim Zhao

TI__Genius 15370 points

Hi Wei,

Sorry again, but I have a question regarding the parameter "0.00355". I understood the below equation (11) meaning. But not sure how to use it. For example, what value should we use for VT? AD read data?

Regards,

Uchikoshi

0 Hisao Uchikoshi over 6 years ago in reply to Hisao Uchikoshi

TI__Genius 15370 points

Hi Wei,

Is there any update?

Regards,

Uchikoshi

0 Wei.Jetim Zhao over 6 years ago in reply to Hisao Uchikoshi

TI__Genius 13215 points

Hi Hisao,

Sorry for reply late since I'm on vacation.

Yes, Vt is the ADC measured data. You could see it on the flow chart I shared with you.

0 Hisao Uchikoshi over 6 years ago in reply to Wei.Jetim Zhao

TI__Genius 15370 points

Hi Wei,

Thank you for your answer during on your vacation.

I have asked one more question about "0.00355" which is described on the datasheet. Where does this value come from? Can every device use this "0.00355"?

For example, when we read TLV data, it shows as below.

CALADC_15V_85C = 760

CALADC_15V_30C = 638

In this case, the proportional constant becomes 0.45, which is (85 - 30)/(760 - 638). It does not match with "0.00355" This is the reason why we are asking.　

Regards,

Uchikoshi

0 Wei.Jetim Zhao over 6 years ago in reply to Hisao Uchikoshi

TI__Genius 13215 points

Hi Uchikoshi,

Sorry for confusing you by the data from User's Guide VS TLV data. Please ignore the TLV data for the single-point calibration. You should measure the data device by device.

Then, the "0.00355" is from User's Guide as shown to you. You should confirm the data for each part you are using, although there should be the same data for most of MSP430FRxx devices.

0 Hisao Uchikoshi over 6 years ago in reply to Wei.Jetim Zhao

TI__Genius 15370 points

Hi Wei,

Thank you for your answer during your hard days.

Sorry for asking you again but if we ignore TLV data, we cannot get the proportional constant by only single-point calibration.

Does customer need double-point calibration?

Regards,

Uchikoshi

0 Wei.Jetim Zhao over 6 years ago in reply to Hisao Uchikoshi

TI__Genius 13215 points

Hi Hisao,

Thanks a lot for you and we are doing well here.

You could use the single-point calibration data plus with the proportional constant of "0.00355". Also double-point calibration with gives you proportional constant device by device could also work for you.

But always keep in mind that the on-chip temp sensor is not designed for real temp sensing application which requires accuracy. TI can't guarantee any accuracy for the temp sensor.

0 Hisao Uchikoshi over 6 years ago in reply to Wei.Jetim Zhao

TI__Genius 15370 points

Hi Wei,

OK, so we should ignore the TLV data and following the equation on the UG.

In this case, we need to normalize raw ADC data to fit this equation and to use the proportional constant of "0.00355".

Could you please show us how to do that?

Regards,

Uchikoshi

0 Wei.Jetim Zhao over 6 years ago in reply to Hisao Uchikoshi

TI__Genius 13215 points

Hi Uchikoshi,

The flow chart I shared you previously is about this.

**Attention** This is a public forum

MSP low-power microcontrollers

MSP low-power microcontroller forum

MSP430FR2433: Integrated Temperature sensor is not expcted