AMI Analyzer Manual
Maintenance and troubleshooting
30
Troubleshooting
Basics of trace oxygen troubleshooting
Most problems are due to either leaks, or a used-up sensor.
Oxygen is of course present in air at about 3psi partial pressure, and the rate at which it diffuses into a gas
line is only dependent on the difference in the partial pressure between the gas in the line, and in air. Since
there is normally (hopefully) no oxygen in the pipe, that full partial pressure differential drives oxygen into
any leak, no matter what the partial pressure of any other gas may be. Any leak therefore will increase the
oxygen in the sample. This is also true for span gas – a leak at the regulator on a span gas tank can allow
oxygen to flow into the tank, changing its oxygen level and making any spans performed with that gas
incorrect.
Typical life expectancy of a sensor is somewhere between one and two years, depending on oxygen levels.
If they are exposed to too much oxygen, they will get used up, but also they may get saturated by being
exposed to too much oxygen and not be able to come down to low levels for a long time, even a matter of
days. We strongly suggest that you keep its air exposure to a maximum of one minute for best downscale
response time, and always keep it shorted when it is not in the analyzer. Leave its shorting clip in until you
have put it into the cell block, and only then pull the shorting tab out of the sensor.
If the sensor is deeply frozen (below about 25
º
F), or cooked (above 115°F) the membrane can be damaged
and if so, it will no longer read correctly. This may or may not happen. Also, the sensor can be poisoned by
excessive levels of H
2
S or other poisonous gases. Any of these things will cause the sensor to read too low.
Reading too high is normally caused by too much oxygen – either in real time due to a leak, or else because
the sensor has been saturated with air. You may be able to detect this latter by looking at the data log. If
you suspect a leak, due to excessive oxygen readings, you can often get an idea of where it is by changing
the sample flow rate. If the oxygen reading decreases as you increase the flow rate, that is a sure sign of a
leak since the intake of oxygen is pretty constant, but the dilution of that oxygen by the sample is of course
greater with a higher flow rate. If you time how long it takes for the reading to decrease, the distance to
the leak source is more or less proportional to the time it takes to fall. If it goes down immediately, the
source is very close to the sensor – possibly the cell cap is loose, or perhaps an O ring in the analyzer valve is
damaged. If it takes a while for the reading to change, the source is further away. It can be very difficult to
pin point the source of the leak, but don’t give up – the analyzer is essentially a leak detector and it is only
doing its job.
Another common problem is a reading that is too low. This is normally due either to a dead sensor, or to an
incorrect calibration. If the span gas is contaminated with oxygen, when you span the analyzer you will turn
the span down too much to make the reading come out to what you thought it should be. It can be hard to
figure this one out: what we suggest when things seem to be confusing is that you perform a rapid air
calibration, taking less than a minute of air exposure, and then let the sensor stabilize on the sample gas (or
nitrogen); then flow the span gas and see what the analyzer reads. The sensor and analyzer are in fact very
linear, so if the analyzer now says that the span gas has a lot more oxygen than the label on the tank states,
