LPS22 sensor suddenly behaving weird

[mbenkmann]

I've had this problem with an ILPS22QS and now I'm having it with an LPS22HH. It has me completely mystified. The issue is like this: The pressure sensor works as normal for a week or even multiple weeks and then all of a sudden it enters a mode where it no longer reacts to pressure changes properly. Instead it looks more like a capacitor where pressure peaks increase the value of the sensor and if there are multiple in quick succession they add up. Once the pressure goes back to environmental pressure the value slowly goes down. A picture illustrates this better. The picture is a recording of pressure values from 2 sensors of type LPS22HH on the same PCB. They've both been working fine for about a week now. Today one of them entered the weird-mode. Power-cycling does nothing. I've unplugged and replugged the PCB multiple times. The firmware of the PCB has not been changed, nor does it have non-volatile memory that could change its behavior. For both pressure sensors I take a drinking straw and blow on the sensor a few times in quick succession. The lower trace is what I expect. It's what I've seen every time I've done this test with LPS22 type sensors. This trace is from the sensor on the PCB that still works. The upper trace is the reading from the weird-mode sensor. As you can see the individual blows stack up and then slowly drain out.

What could be responsible for this behavior?

[TDK]

What environment is this in? Could the sensing element or hole to it be contaminated with dirt or something else? Can you pull the chip off and inspect?

[mbenkmann]

Normal air. I don't see anything on the chip. Besides, I do not see how dirt could cause this weird behavior where the chip starts summing up pressure. Also blowing on it while I tested with the straw would certainly have removed or at least changed any dust accumulation on the chip. I'm confident dirt is not the explanation.

[mbenkmann]

I may have spoken too soon about it not being related to dirt. By pure coincidence when I was assembling a new unit from a PCB that I had verified to be working, immediately after assembly, one of the sensors malfunctioned in the above manner. I had a hunch that it might be related to me touching the sensor with a silicone seal during assembly. I brushed the silicone seal over a working sensor and was able to induce the weird behavior. I didn't see anything on the sensor under a magnifying glass except for some dullness. I was able to restore sensor function by polishing the sensor area with a foam Q-tip. So the problem seems to be related to some form of residue. Since one sensor failed after working for days after assembly I am now wondering if the problem is related to some specific chemical that might gas out of the silicone seal or if the LPS22 sensor is so sensitive that just normal air from a human living environment may cause it to malfunction. From the documents available I did not get the impression that the LPS22 requires extreme filtering (HEPA, activated charcoal). The fields of application named in the datasheet certainly make it sound like the sensor would work fine with normal unfiltered air. Am I wrong about that?
Let's pick one example application from the datasheet: e-cigarettes. How would the sensor be used in an e-cigarette if it can't stand either human breath or things like vegetable glycerin that would be contained in the vapors from the e-cigarette. What kind of filter would one have to use to protect the sensor?

[ajanky]

exactly same behaviour here: LPS22HH in a developer board, working fine, suddenly room pressure at 3000mbar. 2nd sensor on the same board reading normal range values of 1020mbar.
after two days fiddling found this post, cleaned sensor surface with Qtip+IPA - everything normal again, same reading as the 2nd sensor.
Since this cannot be done by users: what filtering / precautions must be taken to exclude such contamination. Otherwise this could be a killer argument against the use of the LPS22. Any advice is welcome!

[mbenkmann]

I've tried various things and come to the conclusion that the only sensors suitable for normal air are the waterproof versions ("W" types like the LPS22HHW). They're more expensive but work reliably without any filtering.