Analox analyzer

[tarponchik]

Going back to the advice to seal the analyzer in a sandwich bag, I do not see how this can help to prolong sensor's life. There are 2 events that shorten sensor's life, lead oxidation and electrolyte evaporation. The former only happens when the measurements are taken, and the latter will not be stopped by isolating the analyzer in a bag because the volume of the sensor cell is too small compared to the bag's volume. UNLESS you add some water into the bag, of course

 

[tbone1004]

@tarponchik

I misspoke about that first comment. O2 is much less variable than atmospheric. Was a long day..

Linearity occurs regardless of what you calibrate with. The farther from the reference gas, especially when using single calibration points, the higher risk of the variability being there. The example that @victorzamora gave is one of cell limiting, but cell limiting is lack of linearity taken to the extreme

 

[victorzamora]

Atmosphere and 100% are the only two gases with oxygen content that is a more or less constant. A bottle of 50% doesn't mean it is 50.0%. That is all I was getting at with the two gasses.

I agree completely, and you're certainly thinking the right way. One thing to mention is you always want to calibrate with a known-gas close to what you're actually wanting to measure. Calibrating on pure O2 isn't going to do squat if you're trying to measure 6% oxygen content. Rebreather divers want their cells reading near the 1.0 range, so they calibrate on pure oxygen. Many divers then check the linearity at a pO2 of 1.6 to validate that it's not limited when on pure oxygen at 20ft (another known point).

Just like everyone should: They calibrate at a known-value near the range of their measurements, and then go high to check for limited cells.

I just find, checking against 50% a little off

I think what tbone meant was to calibrate on Air, validate it with pure O2, and then you'll be more confident in your measurement of the EAN50 analysis. Like I said above, you want to calibrate near where you want to measure....and EAN50 is far from both known gasses. I agree with tbone's statement of needing an O2-linearity check when measuring something like EAN50 or EAN70. At EAN80, I'd likely just calibrate on oxygen.

 

[tarponchik]

@tbone1004
Speaking theoretically, you are correct. In linear approximation (Y=a*X+b) the slope (a) is the major contributor to the error, regardless of b being equal to zero or not. Thus, when you are building a straight line using just 2 points and one of these points is fixed at 0,0 and does not contribute to the error, the further away on X axis the 2nd point is, the better. An error of 1% in X will give you higher error in the line slope at, say, X=50% than at X=100%, etc.

However, in this case we know that when the sensor is aged, the greatest deviation from the straight line occurs at high O2 percentage. So using high O2 works better in linearity testing but using high O2 as a calibration standard will give you higher error for lower O2 mix if your sensor is aged. Makes sense?

 

[doctormike]

I don't know enough science to be sure about this. But another interesting point is that looking for extreme accuracy of one variable in any decompression model may have little real world implications unless you are at extreme depths. What I mean is that decompression theory is "a bright line through a grey area", and even the oxtox limits are exposure over time, it's not like there is some line where you are totally safe above it and instantly seize if you go below it. So assuming you aren't pushing O2 CNS exposure, I looked at some numbers.

Looking at the Analox chart, a huge swing due to humidity and temperature would be comparing 32° F and 10% humidity to 100°F and 100% humidity. The chart says to calibrate to 20.9 with the former and 19.5 with the latter. 7% difference. Assuming linearity, that would mean approximately the difference between EAN 32 and EAN 30.

So plugging in those two mixes to MultiDeco for a 30 minute dive to 90 feet, and setting the model to VPM with +2 conservatism, I get 2 minutes of deco for EAN30 and 1 minute for EAN32. VPM+3 is 4 minutes vs. 2 minutes, VPM+4 is 8 vs. 4, GF 30/70 is 8 vs 7, and GF 10/90 is 5 vs 4. I'm sure that you would see similar differences comparing a conservative vs. liberal recreational dive computer.

The point is that the variability in your dive planning is affected much more by what model you choose than whether or not you carry around a hygrometer and measure the humidity and then adjust the calibration factor. And all of these models are reasonable choices in some context. I get the sense that if the OP had asked if he should use VPM+2 or GF 30/70, he would have been told that it's a personal decision and that there is no right answer, even though that has far more impact on his dive plan than even the greatest swing in temperature and humidity.

Reminds me of the joke - a tour guide at the Museum of Natural History tells his group that a dinosaur fossil is 100 million and three years old. When someone asked him how he could be so accurate, he said "When I started working here three years ago they told me that it was 100 million years old".

 

[victorzamora]

Mike, you're quite right. Deco has been described as measuring with a micrometer, marking with chalk, and cutting with an axe. A few percentage points really isn't important for 99.999% of dives occurring globally. Heck, even most technical dives. Dives where things like that start mattering often send samples of their gases to labs for proper analysis.

 

[tbone1004]

@tbone1004
Speaking theoretically, you are correct. In linear approximation (Y=a*X+b) the slope (a) is the major contributor to the error, regardless of b being equal to zero or not. Thus, when you are building a straight line using just 2 points and one of these points is fixed at 0,0 and does not contribute to the error, the further away on X axis the 2nd point is, the better. An error of 1% in X will give you higher error in the line slope at, say, X=50% than at X=100%, etc.

However, in this case we know that when the sensor is aged, the greatest deviation from the straight line occurs at high O2 percentage. So using high O2 works better in linearity testing but using high O2 as a calibration standard will give you higher error for lower O2 mix if your sensor is aged. Makes sense?

that's why you use two point calibration. So I ask again, how do you determine any of the values to calibrate by using a 0% cal gas? I'm 100% OK with using ambient air for calibration gas, I do it on a regular basis, including all of the next two weeks since I won't have O2 easily accessible at the dive sites. Use of O2 was as an ideal second calibration gas to verify linearity as well as a cell being limited or not. You don't know if a cell is limited without that knowledge and as @victorzamora pointed out, his analyzer would calibrate at air, but would not read EAN32. You have no way of knowing that without doing a multipoint check. Calibrating against EAN50 would probably be more accurate for normal nitrox mixes, but how accurate can you guarantee that gas to be without purchasing dedicated calibration gases? I know 100% O2 out of the bottles is going to be 100% because of the way they are filled.

So the question again, what benefit does using 0% as a calibration gas provide when 0 always equals 0, and how can you be sure that when calibrating to ambient, that the cell is not limited when you do not test it at a value higher than where you need it to function?

 

[victorzamora]

So the question again, what benefit does using 0% as a calibration gas provide when 0 always equals 0, and how can you be sure that when calibrating to ambient, that the cell is not limited when you do not test it at a value higher than where you need it to function?

To add to this: Because of how the galvanic cells work, pO2=0 means mV=0. So that point is always a known/given. That's the only way you can determine the slope with a single-point calibration anyway (two points are required to define a line).

 

[tarponchik]

@tbone1004 and @victorzamora
Don't you guys read my replies at all? I'd already agreed with you on this one. The neutral gas calibration would be useless because the 0,0 point is already used by the analyzer.

And why do I think the 0,0 point is used by the analyzer? Because there is no other way it can work the way it works.