January
2005, Issue 174
Microcontroller-Based
Nitrox Analyer
TEST
RESULTS
Atmel’s
AVR Studio simulator is a valuable aid for testing firmware.
I used it for the verification of the fixed-point code
to step through the assembly. I also used the AVR Studio
to verify the calculations for a wide range of input
values. Simulation can take you only so far though.
To truly test the device, I needed access to various
samples of Nitrox. Thanks to Dive Connections in Charlottesville,
Virginia and the Olympus Dive Center and Discovery Diving
Company in Morehead, North Carolina, I was able to test
my device’s results against those produced by the shops’
commercial analyzers.
I
tested 17 different tanks of Nitrox. It’s evident that
the worst-case deviation between the Nitrox analyzer
and the commercial analyzers is 0.4%, with the average
deviation being on the order of 0.1% (see Figure 3 on
page 26). The Professional Association of Dive Instructors
(PADI) recommends that a Nitrox dive be planned using
a percentage of oxygen within 1% of the actual value
in the tank.[1] So, the accuracy is acceptable.
|
(Click
here to enlarge)
|
Figure
3—I put 17 different Nitrox tanks to the test. Here
you see the percentages of oxygen in each tank tested
with the analyzer versus the percentages found with
a commercial analyzer. |
Although
the results look encouraging, it’s important to consider
the following factors. Commercial analyzers are also
susceptible to errors in measurement and calibration,
which means that the test results illustrated in Figure
3 only provide an indication of accuracy relative to
the commercial device being used as the reference. In
other words, it isn’t a good indicator of absolute accuracy.
To measure absolute accuracy, precision Nitrox reference
samples are required. Another factor to consider is
that the available samples were all between 29% and
32% oxygen. More exhaustive testing would require samples
across the 21% to 40% operable range of the analyzer.
