Oxygen Sensor Fundamentals

[tbone1004]

That is likely very true. I was trained as a diver and instructor by Leon Scammahorn and I promise you he is very thorough about cell principals and I can assure you I cover it in detail in my classes as well. I would accept your claim regarding instructors and divers you know but since we run in different circles my experience is much different than yours.

Leon certainly knows what he's talking about and this type of instruction is starting to grow as the years go on, but I am still willing to bet that at the global level it is a very small fraction that know about linearity, how it works, what it means, and actually track it...

Of note, linear drift is one of the reasons that the meg has 2-point calibration and the reason that the Liberty has multi point calibration as well

 

[Jack Hammer]

you'd be surprised... Cell linearity is not taught by any agency that I'm aware of. There are many prominent instructors who deny it even exists. Cave country in Florida has a lot of CCR divers and instructors, the VAST majority down there have no idea...

It's taught in GUE CCR1 class and verifying cell linearity is part of both their standard CCR assembly and predive checklists.

 

[telemonster]

Interesting presentation. I knew a good bit of the info but did pick up some interesting stuff.

With regards to these solid state sensors from Poseidon:

Poseidon

It says a dye reacts to the O2 and there is color change. Are they sampling this with a CCD and a light source or something? How does the dye not dry out?

 

[tbone1004]

Interesting presentation. I knew a good bit of the info but did pick up some interesting stuff.

With regards to these solid state sensors from Poseidon:

Poseidon

It says a dye reacts to the O2 and there is color change. Are they sampling this with a CCD and a light source or something? How does the dye not dry out?

Poseidon Fixes Closed Circuit Rebreather Diving Weakest Link - DeeperBlue.com

dye is hit with an LED and is reflected back onto the sensor. It interprets the color coming back as ppO2. Standard methods for optical analysis of other gases, but oxygen is particularly tricky, especially when considerable moisture is present. Since CCR loops are always at 100% relative humidity, it's particularly annoying.

 

[stuartv]

Joe Citelli is awesome. Thanks for posting this presentation.

I have a couple of questions:

In the middle part, where he first talks about linearity, he goes through an example of a cell being non-linear by 10%. So, in his example, your computer is telling you you're at a ppO2 of 1.0 and you're really at 1.1.

So, I don't know about any other CCRs, but the Shearwaters on my rEvo all calibrate to O2. So, no matter what the linear deviance is of my sensor, if the computer says I'm at 1.0, then I'm at 1.0. Right? And if it says 1.1, then I'm really at 1.11, and if it says 1.3, then I'm really at 1.33. Right?

Second, during the summation at the end, Joe talks about putting a piece of tape on your controller and writing down the expected millivolts for each sensor at a ppO2 of 1.0 (and possibly other ppO2 values). He didn't go into great detail on how to use that, but if I understood him correctly, the idea would be that during your dive, you can check your displayed ppO2 against the millivolt readings.

For example, if your expected millivolts are 47 for a ppO2 of 1.0, and you're going along on your dive, you adjust your loop until the ppO2 reading (of the sensor you're checking) is showing 1.0, then you (if it's a Shearwater) tap the right button a few times to check the millivolt reading and confirm that it is showing 47.

I don't understand how that makes sense. Isn't the displayed ppO2 simply the current millivolt reading multiplied times whatever factor was set during calibration? So, after calibration, if it shows ppO2 of 1.0 and 47 millivolts, then won't it ALWAYS show ppO2 of 1.0 for a reading of 47 mV, until it gets calibrated again?

I don't understand how noting the expected millivolt readings and checking them during a dive is of any use at all. If I understand how these things work correctly, then the ppO2 reading will always directly relate to the mV reading, so you can just check that (the ppO2). And, the only way to check it (the ppO2) for accuracy is to either do a dil flush and compare against the Dil pO2, or do an O2 flush (if you are shallow enough). Anything else is just looking at an unknown gas and comparing the displayed ppO2 against the displayed mV reading (which will always match, for any given calibration).

I feel like I must be missing something here. :-(

This is all presuming single-point calibration, as that is the only thing I have experience with.

 

[RainPilot]

I don't understand how noting the expected millivolt readings and checking them during a dive is of any use at all.

The main issue here is that the operating conditions for cells get worse during the dive, thus the likelihood of a new failure (or an existing small error becoming dangerous) increases during the dive. A cell may be current limited at 1.2 for example, which you won't pick up at the 6m flush but will be noticeable by mV readings on deco for example. Linearity deviations may not be linear at depth either, so knowing what mV to expect for a given PO2 and given sensor on that day, gives an early diagnostic tool to hopefully notice the deviations before they become life-threatening.

 

[tbone1004]

@stuartv
The computers are programmed for y=mx, only the Meg and Liberty have the ability to put "+b" into the formula with multi point calibration to adjust the pitch.

Noting the expected mV value for 1.6 is important because when you do the 1.6 check for current limiting, you can also validate your linear deviation. Depending on your chosen ppO2, that may or may not be important. I don't run a 1.6 until I get to my O2 stops and at that point the sensors are essentially irrelevant. If you run a 1.6 at deeper stops, then you really want to make sure that you are actually at 1.6 instead of 1.7 or worse. I run a setpoint of 1.1 on most all of my diving because the HUD on the Meg is obnoxious at any other ppO2. If I'm doing nitrox cave diving then I really couldn't care less what my linear deviation is because my Meg has 2-point calibration that has it factored in, I calibrated at .21 and 1.0, so the deviation at 1.1 is insignificant. I don't hit 1.6 until I get to my 20ft stops, so it doesn't matter in that specific circumstance. On my Meg, I have a high tolerance for linear deviation because of the 2-point calibration and the setpoints close to calibration.
My current sidemount breather does not have 2-point calibration, and because I don't have a blinky HUD, I can run whatever setpoint I want to. Now the linear deviation can start to add up, especially if you have a higher setpoint if you're trying to minimize deco. You want to check those ppO2's vs. mV's when you are on your descent with a known O2 loop.
Like @RainPilot said, you also want to validate the cells at various points during the dive. You can do this with a dil flush if you have a hot dil since the computers will tell you your dil ppO2 and you can validate that during the flush instead of the mV, but not all divers run hot dil mixes. A dil flush to check for cells at the dil ppO2 is pretty useless if the dil ppO2 is below 1.0. If you have a rich dil mix though, you do want to check that if you can. On ocean dives this may not be that significant due to shorter run times, but the cells can get pretty strange when you're on really long dives.

 

[stuartv]

The main issue here is that the operating conditions for cells get worse during the dive, thus the likelihood of a new failure (or an existing small error becoming dangerous) increases during the dive. A cell may be current limited at 1.2 for example, which you won't pick up at the 6m flush but will be noticeable by mV readings on deco for example. Linearity deviations may not be linear at depth either, so knowing what mV to expect for a given PO2 and given sensor on that day, gives an early diagnostic tool to hopefully notice the deviations before they become life-threatening.

Okay, now I feel REALLY lost.

I get to 6m and do an O2 flush and you're telling me that I won't be able to detect a sensor that is current limited to the equivalent of 1.2 ATA ppO2?

I thought that was the whole point of the O2 flush at 6m. If a sensor doesn't read around 1.55 ATA ppO2 or better, then it is current limited.

@stuartv
The computers are programmed for y=x+b, only the Meg and Liberty have the ability to put M into the formula with multi point calibration to adjust the pitch.

Noting the expected mV value for 1.6 is important because when you do the 1.6 check for current limiting, you can also validate your linear deviation.
[snip]
Like @RainPilot said, you also want to validate the cells at various points during the dive. You can do this with a dil flush if you have a hot dil since the computers will tell you your dil ppO2 and you can validate that during the flush instead of the mV, but not all divers run hot dil mixes. A dil flush to check for cells at the dil ppO2 is pretty useless if the dil ppO2 is below 1.0. If you have a rich dil mix though, you do want to check that if you can. On ocean dives this may not be that significant due to shorter run times, but the cells can get pretty strange when you're on really long dives.

I'm not getting this and I apologize if I'm just having a huge brain fart.

You said computers are programmed for y=x+b.

I'm taking Y as ppO2 and X as the mV reading. So, if my sensor is good and it reads 11mV in air, then I would expect it to read 52mV when I calibrate it in O2.

Y=X+B means: 1.0 = 52 + B. So, B = -51? That is obviously not right. Otherwise, when it's back in air, it would be saying my ppO2 is 11 + (-51).

It seems to me that computers would be programmed with Y=MX+B, and only the Meg and Liberty could determine B. All other computers would assume 0 = 0 as one point and be unable to calculate B. Those single-point computers would use 0,0 and the calibration point to determine M.

So, if calibrated at 1.0 ATA in O2, with the sensor giving 52mV, it would assume that 11mV means 0.21 ATA ppO2.

However, I could be wrong. But, if so, can you please give me a couple of real numbers as an example (for a single-point computer) of the Y=X+B calibration?

Anyway, moving on to the REAL problem I'm struggling with:

You calibrate your computer however you calibrate it. 1 point, 2 point. I don't care. When you're done calibrating, the computer reads 68mV (we'll just say for sensor #1) when your sensor is good and you're at 1.3.

So now, during your dive, sensor #1 is showing 1.2 ATA ppO2. I'm thinking that that means the mV reading for that sensor is going to show as 63mV. Am I correct, so far?

Assuming so, how does looking at the mV help me? I have one sensor reading 1.2 and two sensors reading 1.3. I can see that it's reading low. If I could do the math in my head, I would know without looking that it is showing 63mV on that sensor - because the two numbers are mathematically tied together. So, I know sensor 1 is limited just by seeing the ppO2 readings. What does looking at the mV # tell me that I don't already know?

The way you're talking about it makes it sound like the computer might still be showing me 1.3, but if I checked the mV I would see 63, instead of 68. I don't think that is possible. That would imply that the computer's calibration has changed during the dive.

 

[stuartv]

I apologize for being so wordy.

Here is the short version.

When my sensor is good, it has 68mV at 1.3 ppO2 and my computer shows 1.3 ATA ppO2.

When in O2 at 6m, and the sensor is good, it has 84mv and the computer shows 1.6.

Now my sensor has gone bad and is limited to 63mV. I get to 6m and do an O2 flush.

What is my computer going to show me? I think it's going to show 1.2 ppO2 on that sensor. And if I check, it will show 63mV.

So, what does the mV reading tell me that I didn't already know?

Is there some other time that the mV reading would tell me something that I didn't already know from the ppO2 reading?

 

[tbone1004]

@stuartv
sorry formula was corrected after you quoted it, not enough coffee this morning.
single point calibration is y=mx, 2-point is mx+b

If your ppO2 is displayed in hundredths, then the mV isn't going to necessarily give you anything more. Many only spit out in tenths and that isn't always "enough" to validate cell function depending on how they round.
What is not really explained well is that the original author of this thing is basing his experience off of a Meg. Since the Meg uses 2-point calibration, it is shockingly easy to get a bad calibration in air. By charting the mV's, you are checking the cell function directly as opposed to checking the cells as a function of your calibration.
I think it's less critical with single point calibration, but where the authors are coming from is thousands of hours with a Meg where the 2-point calibration can add some complexity and you are validating the cells outside of that likely imperfect calibration.