LPS25: Extrem temperature dependant pressure output

Uwe Bonnes · ‎2017-08-15

Posted on August 15, 2017 at 14:57

Uwe Bonnes · ‎2017-08-15

Posted on August 15, 2017 at 15:07

Sorry, text should have gone with first text.

I am reading pressure and temperatur on a IKS01A1 board. Fifo mean with decimation and highest averaging is used. Heating the IKS01 from below with my finger for about 1 degree changed pressure reading for about 4 hPa. As the data sheet has no min or max values or curves with typical values for 'Absolute accuracy pressure

over temperature' the device could still be in specs, but is not usefull...

Find appended a CSV list of time in milliseconds, pressure in hPa and temperature in Celsius. At the top is a register dump after setup.

Any comments?

________________

Attachments :

log.zip : https://st--c.eu10.content.force.com/sfc/dist/version/download/?oid=00Db0000000YtG6&ids=0680X000006HyVE&d=%2Fa%2F0X0000000b8m%2FKeRxy4NBizaLMH4FM0b8frRbW7jsLN9AFxgihqfdevY&asPdf=false

Miroslav BATEK · ‎2017-08-16

Posted on August 16, 2017 at 11:14

If you want to evaluate temperature dependency of pressure value over temperature please use some heating element or temperature chamber. If you use your finger which push on the sensor you create significant mechanical stress and this stress influence the measurement.

Uwe Bonnes · ‎2017-08-16

Posted on August 16, 2017 at 11:24

In another test I put the board in a box, 'closed' by some paper. The same pressure/temperature dependancy is seen: As heated air can't easily escape, the air inside the box and so the board heat up and the measured pressure goes up. When removing from the board, board cools down again and pressure goes down also.

So I am sure the stress applied by my soft finger touch had only minor impact on the measurement above.

Uwe Bonnes · ‎2017-08-28

Posted on August 28, 2017 at 10:25

I found another IKS01A1 board, that one with an LPS25H with the old packaging. The LPS25HB on the board showing the large variations was a LPS25HB in 'Bastille' packaging mode. I ran both board stacked up on a Nucleo, one board jumpered with the other set of I2C addresses. The LPS25HB again showed that large variations, whereas the LPS25B has sensible output.

As soldering stress may heavily influence the results, I reflowed the LPS25HB with a hot air pencil and cleaned it with denaturated alcohol. Before there was some layer visible on the exposed chip backside, probably solder resin.

At first the re-reflowed LPS25HB was off about 50 hPa, but gradually creeping versus the value of the LPS25H. After about 6 hours it has reached about the value of the LPS25H and tracked the values of the LPS25H afterwards. Noise on the pressure values of the LPS25HB seems somehow larger, while the LPS25H has more noise in the temperatuure values.

It seems that the soldering process for the IKS01A1 boards was inappropriate to keep the accuracy of the LPS25HB.

If your IKS01 with an LPS25 has large offset or temperature variation, another reflow and cleaning might bring it back to accuracy...

Uwe Bonnes · ‎2017-08-29

Posted on August 29, 2017 at 10:17

More observations:

During the weekend, there was nobody in the lab, so no fast temerature changes. On Monday, lab was busy, with windows open and so fast temperature changes. This again resulted in extreme pressure changes with the LPS25HB while the LPS25H was not disturbed by the temperature changes. Does anybody have good results with a LPS25HB?