4

I'm a lowly mechanical engineer so bear with me. This problem has been the bane of my existence.

I'm using a temperature-compensated barometric pressure sensor in a fluid-contacting application. Specifically, TE's MS5803-02BA. This sensor is hermetically sealed (via silicone gel coating) and commonly used in dive watches. I'm using it in SPI mode, where it outputs two 24 bit values: "D1" for pressure and "D2" for temperature.

datasheet

Failure Mode: In ~1/25 devices after less than 12 hours of continuous use (fluid (saline) contact): D1 and D2 start steadily decreasing, always moving together. When I dry the sensor, the outputs slowly increase back to their expected value, and there appears to be no long-term damage to the sensor. Once I apply fluid again, the values again decrease steadily.

It seems obvious that fluid plays a role, but my question is: why would the output drift in such a predictable, repeatable, and reversible way?

To further complicate things, I've also tried chemically and mechanically removing the silicone (or most of it) and applying saline directly to the sensor. This does not induce the failure. I'm pretty lost at this point.

sensor

block diagram

mere engineer
  • 41
  • 1
  • 2
    Test it with distilled water. Note that water can permeate through silicone and these sensors are very sensitive. – DKNguyen Mar 20 '23 at 23:06
  • mere engineer - Hi, That's an interesting problem. I won't have time to get into detailed troubleshooting, but here's a couple of points: (a) The sensor is marketed on its datasheet as an altitude sensor, not a saline pressure sensor. Despite the silicone gel etc. I didn't see anything in the sensor's datasheet saying that saline pressure sensing *with constant saline contact* is a qualified usage. (As you say, the test results indicate that the fluid is part of the issue.) (b) I would definitely recommend contacting the manufacturer (e.g. their FAE) to discuss your application. Good luck! – SamGibson Mar 21 '23 at 02:08

1 Answers1

0

why would the output drift in such a predictable, repeatable, and reversible way?

This sounds like a high-resistance short between Vdd and GND - one of the type that might be introduced by very slow water incursion.

The sensor's voltage output is referenced to Vdd, and the operating range of Vdd is 1.8-3.6V; if you normally supply it with 3.3V, fluid incursion could cause Vdd to drift down as the high-resistance short between Vdd and GND slowly decreases in resistance, and the outputs will drift down with it. It's likely the sensor would completely shut off given more time under these conditions, once Vdd drops below 1.8V.

This would also explain why the fault takes a while to manifest, but manifests more quickly once it's occurred once. Water may have saturated something, or left salt deposits which get quickly re-dissolved by future water.

The datasheet has this to say about waterproofing:

In products like outdoor watches the electronics must be protected against direct water or humidity. For those products the MS5803-02BA provides the possibility to seal with an O-ring. The protective cap of the MS5803-02BA is made of special anticorrosive stainless steel with a polished surface. In addition to this the MS5803-02BA is filled with silicone gel covering the sensor and the bonding wires. The O-ring (or O-rings) shall be placed at the outer diameter of the metal cap. This method avoids mechanical stress because the sensor can move in vertical direction.

Are you following their design guide here - provide your own robust waterproofing for the PCB side of the sensor, use an appropriately-rated O-ring to seal the housing exit, and only let the metal cap contact water? Water could be getting in through the silicone inside the sensor's metal cylinder, but it could more simply just be getting in through your own waterproofing around the sensor.

If your waterproofing is perfect, you still might be slightly out of spec, as the datasheet does not seem to indicate any rating for sustained immersion. But if you have an example of the part being used successfully in this manner (dive watches), it may still work. That said, do these dive watches actually expose the metal cap to water, or do they use a rubber membrane to transmit pressure to the sensor without immersing it? If the latter, you're probably out of luck for trying to immerse this sensor.

Willa
  • 361
  • 3
  • 5