Were you trained on Oxygen sensors with this level of detail?

[tbone1004]

@tarponchik I see no reason why on a CCR you would ever want to calibrate with air. The ppO2 of air at the surface is so incredibly far from where you will ever run the setpoint of a rebreather that have that as the calibration point is completely useless.

Your argument about wearing the cells out faster has little validity as well because the rebreather is going to be run at a ppO2 of .8 to 1.6 for hours upon hours upon hours for the life of that cell so exposing it to the ppO2 for calibration is comparable to diving for any extra 5 minutes on the cell life? If that is something that you care about, you probably need not be on a rebreather. Obviously you are consuming more electrolyte, but it is the same as doing an extra 5 minutes on a decompression stop. Not something that is going to matter to the actual life of the rebreather cell.

I think this is best thought of as a graph
Cell controllers are simple devices. We will assume that air has a ppO2 of .2 and is 20/80 for easier math. To follow @Bobby 's math, we will say that the cell shows 10mV when exposed to air, and 45mV when exposed to pure O2 at atmospheric pressure.

In single point calibration for air, the device assumes a straight line of what amounts to .02x. I.e. it takes the mV reading, multiplies it by .02, and that gives your ppO2 reading.
In single point calibration, ALL it can do is take the mV reading and multiply it by .02 and that's it.
When you are actually breathing 1.6ppO2 which is 65mV, it will be showing you 65*.02=1.3
If you bring your setpoint up to where it shows 1.6, then you will be at 80mV and obviously breathing significantly higher than 1.6 ppO2 somewhere in the 1.7-1.8 ish range

If however we do an oxygen calibration, then the slope is now .22 because it takes 45mV and tries to get 1.0.
When this calibration sees 65mV it shows 1.43ppO2 on the controller and would only try to bring it up to 73mV if you had a setpoint of 1.6 which would be slightly less bad than the air calibration.

If I have time tomorrow I'll try to plot this so you can see it visually.
TLDR is that single point calibration uses one formula
y=mx where y=ppO2, m=some multiplier, and x=mV from the cell
The closer to y that you calibrate, the more accurate M will be in the area that you are reading

2 point calibration like the old Meg, creates a formula of y=mx+b which gives something a bit closer to reality.

The only way to be 100% sure what a cell is going to do is to be able to pressurize the head. Best way to do that is to be able to pressurize it to 4.76ata on air which is a known gas and easiest to be sure about, and have it show 1.0, then pressurize it to 7.62ata on air and have it show 1.6
No rebreather I am aware has a head calibration kit that can be pressurized to 100psi to do that, so we have to do a lot of math to attempt to predict the way a cell will hopefully behave when diving

 

[tarponchik]

@tbone1004 You do not have to explain this math to be because I know this math. What did I say? I said that calibrating with O2 will be more accurate (but only when your sensor is new).

The reason is that non-linearity caused by sensor aging is higher at higher O2 values. Too long to explain the chemistry, but the deviation from linearity in the aged sensor will look something like this: -0.5% at 60%, -3% at 90%, and -10% at 100% O2. So if you calibrate this sensor with 100% O2 you'll get the error higher than if you calibrate with air.

 

[tbone1004]

@tbone1004 You do not have to explain this math to be because I know this math. What did I say? I said that calibrating with O2 will be more accurate (but only when your sensor is new).

The reason is that non-linearity caused by sensor aging is higher at higher O2 values. Too long to explain the chemistry, but the deviation from linearity in the aged sensor will look something like this: -0.5% at 60%, -3% at 90%, and -10% at 100% O2. So if you calibrate this sensor with 100% O2 you'll get the error higher than if you calibrate with air.

Why do you care what the accuracy of the cell is outside of the ppO2 range you intend to dive it?
I personally want to know what the cell does in the area that I'm diving it, which is 0.8 to 1.6 depending on the circumstance. The deviation at a ppO2 of 1.0 is what I want to know, because what the sensor does below a ppO2 of 0.8 is completely irrelevant unless you are trying to analyze nitrox. A CCR with three galvanic O2 sensors is a VERY expensive nitrox analyzer

 

[_Slacker_]

Again, please remember the paper is a draft that was done years ago and hasn't been formally reviewed, so the actual numbers given may be a touch off, and in this case, they are by a bit.

Right, I've reviewing it and saying it is in error.

 

[paulraymaekers]

@tbone1004
The reason is that non-linearity caused by sensor aging is higher at higher O2 values. Too long to explain the chemistry, but the deviation from linearity in the aged sensor will look something like this: -0.5% at 60%, -3% at 90%, and -10% at 100% O2. So if you calibrate this sensor with 100% O2 you'll get the error higher than if you calibrate with air.

can you show me the graph of that sensor please?

 

[tbone1004]

here is a video of the CCR Liberty doing cell checking to validate current limiting as an fyi. Certainly a kit and a computer feature that all of the ccr manufacturers should be looking at as it removes a lot of the user error from the process.

You can see that it stops and checks the mV at 1.5bar, 2.0 bar, 2.5bar, 3 bar, and 3.5bar, to make sure that everything is behaving properly and nothing is limited.
It plots linearity for you to show that the cells are behaving properly.
You can see when it shows the results of the O2 cell test that the air values are all way off because the cells aren't perfectly linear at the bottom of the range, but they stabilize up in the top section where it counts, and for us that is that 1.0 to 2.0 range.
It also gives you a nice chart that you can write into your wetnotes that has the mV values for air, 1.0, and the specific ATA values that it tested at.
In this chart that it shows, you can see how cell 3 has more linear drift than the others since it starts out with a higher mV value on air, but then falls off more than the others as it gets higher.

tagging @Bobby so he sees the video

 

[sea_ledford]

I think there is a significant disconnect between understanding mV and "always know your PO2", and part of the problem is reading your mV is a bit of a task, rather than glancing at your controller or HUD (at least on Shearwater controlled units).

I spoke with the Shearwater rep at a recent conference and said it sure would be handy if they were able to have the mV reading always displayed as an option. He said it was on the wish list, but the more people request it, the more likely it is to happen. So send an email to shearwater and ask for an option for the mV to always be displayed!

-Chris

 

[wedivebc]

I think there is a significant disconnect between understanding mV and "always know your PO2", and part of the problem is reading your mV is a bit of a task, rather than glancing at your controller or HUD (at least on Shearwater controlled units).

I spoke with the Shearwater rep at a recent conference and said it sure would be handy if they were able to have the mV reading always displayed as an option. He said it was on the wish list, but the more people request it, the more likely it is to happen. So send an email to shearwater and ask for an option for the mV to always be displayed!

-Chris

What's the benefit?

 

[sea_ledford]

You are seeing the actual output of the cell rather than the output adjusted to a calibration and displayed as PO2.
My po2 might be 1.23, 1.20, and 1.17, but that doesn't tell what the mV readings are, or that one might be significantly different than the other two.

 

[joshk]

You are seeing the actual output of the cell rather than the output adjusted to a calibration and displayed as PO2.
My po2 might be 1.23, 1.20, and 1.17, but that doesn't tell what the mV readings are, or that one might be significantly different than the other two.

But that's exactly what the mV readings are only divided by whatever you calibrated at 1.0 with. If each sensor gave out 60mV at 1.0 then you'd have 73.8, 72, and 70.2 mV out for the display in your example.