[ This is a mess.  I'll clean it up :) ]

Sensors are at once usually reliable, and prone to a multitude of failures.  In other words, there are lots of things that can go wrong, but in practice, often don't.

• What do they do?
• How do they work?
• How do you know when a sensor fails? (and why they fail, look for articles on water sealing the membrane, low output, temp, humidity/flooding, linearity issues, etc)
• What do you do when they fail?
• How are the failure modes dealt with from a design standpoint?

Date: Mon, 20 Jan 1997 17:16:17 -0500
From: ski@dciem.dnd.ca (Dave Eaton)
Subject: Re: BioMarine O2 sensors

As stated before we use the Teledyne B5 and R1 sensors. We've used the B5 sensors since 1987 and we just started to use the R1 last year so our experience is not extensive yet.

Oxygen sensors and wet environments was a problem back in the early 80's when we were first investigating the use of an oxygen analyzer in Siva+. We avoided the problem by looking at the dry gas coming from the gas supply circuit--not the wet gas in the breathing loop. So we have not had any reliability problems as a result of high humidity or liquid water.

As far as service life goes, I have numbers on actual exposure times but they are not handy right now. I will dig them up tomorrow.

We have had finger-trouble earlier with people touching the contacts on the back of the cell. This lead to corrosion on the contacts and unreliable readings. [Tech's were telling us that they never touched the terminals but when the corrosion was in a pattern just like their finger prints they didn't argue too much.] Once we explained the importance of cleanliness when handling the cells this problem disappeared.

One thing we have to remember about these cells is that they are very similar to a primary battery. The main difference is that oxygen is providing an external fuel source to produce the electrochemical reaction. Like batteries, they don't like being too wet, they occassionally break and leak, there is inter-cell variation in the output, and they're affected by temperature--cold reduces the rate of reaction, heat speeds it up. I remember when a new set of batteries leaked inside my camera and toasted the electronics. Batteries are supposed to be reliable, but even they screw up.

Determining the eventual failure of a cell is a bit of a black art. Speaking of black arts, I like the black spot indicator on the Biomarine sensors. If it works that's great. Unfortunately, we don't have something as convenient.

In the Siva+ we are able to test the gas control circuit and diagnostic unit (O2 cell) to maximum operating pressure to calibrate the gas mix. Prior to calibrating the gas mixing circuit we calibrate the oxygen analyzer. Our analyzer consists of the fuel cell who's current output passes through a temperature compensation PCB to produce a voltage signal (mV level). In the B5 the temperatre compensation circuit is not integral with the cell body while in the R1 it is. The raw voltage signal is amplified and calibrated to display PO2 (1 volt = 1ATA). We calibrate the amplifier section using a single point calibration, i.e., 1.5 ATA on pure oxygen. We then check the analyzer at 1.2 ATA and 1.8 ATA on pure oxygen to see if the amplified output is within +/- 5% of these endpoints. If the analyzer output is not within these ranges then we consider the linearity of the cell insufficient for our applications and replace it. As I wrote in a previous posting, we are having some tests done now to try and characterize the decay curve of the B5 and R1 cells. This should allow us to get a better prediction of when the cells are going to crash.

Date: Mon, 20 Jan 1997 08:45:36 -1000 (HST)
From: Richard Pyle <deepreef@bishop.bishop.hawaii.org>
Subject: Re: D.E.M.A. SHOW

That's why it's important to have two independent ways of reading the sensors. If there is any sort of error "downstream" of the sensors themselves (such as a glitch with the electronics hardware), the problem will only manifest itself in one of the display systems. This is also why it's important to compare both displays, and why it's important to display the value with a relatively high level of precision, regardless of actual sensor accuracy. (Static readings will be evident much more quickly if the display is high resolution). I know it sounds like a lot of task loading to check these things, but it's really not. A glance every 15 minutes or so would catch this sort of problem LONG before youg got into hypoxia. Also, if your solenoid is audible, and only injects a small amount of O2 per shot, a failure that leads to too much O2 injection will be as obvious as a blown hose on an OC system. Even if you don't notice the audible cues, if O2 is injected faster than your body is burning it up, the increase in buoyancy and the increase in back-pressure from increased volume of gas in the loop will be obvious long before you get into the toxicity red-zone - even in deep water. That's why the CC systems require a lot of in-water time in shallow water - to develop a level of "intimacy" with the unit so that malfunctions are obvious immediately.