Question O2 sensor calibration

[lermontov]

Don't you need to know what MV the cells calibrate to in O2 to do this? Say you have 3 cells and they read 42 mv, 50 mv and 57 mv to start with (at surface) duing o2 calibration.

How do you remember on the dive what the cells started at and which cells had the assigned value? How do you know during the dive that the cells don't drift with temperature change. How can a combination of cell drift and poor memory, make an real sense of the during a dive?

My mental maths level approximates that of a 3 year old, whilst diving at any depth below 6m.

im assuming you check your MV reading at calibration so youll know what RANGE you have and if that range is acceptable. Your mv reading as above are only an example Im assuming as 42mv to 57mv is large difference, so with a top end PO2 of 1.6m, even 10% variation is significant

Im not sure how much of an impact temperature will make but were taking about checking the cells to find anomalies

If you want to check linearity using your mv readings then you can do it with a depth that doesnt necessarily demand your reach po2 of 1.6 so Im not using any math under water - just checking on where the linearity if any starts to flatten off


Ive put in a couple of links for you to look at they will add further info

J

Thread 'Cell linearity checks'

Dec 4, 2019

I have recently been reviewing my knowledge of cells and have read several useful threads on here. This prompted me to resurrect a thread on a UK board that I started some time ago.

I am now a bit concerned that my understanding of the concept is flawed and would appreciate a sense check.

It's probably easiest to understand by reading the following thread if you have time:

Current Limiting and Linearity - Page 8

• JonG1
• Replies: 16
• Forum: Rebreather Diving

 

[inquis]

While Bobby's blog brings up many good points (especially the in-water test), it does neglect the impact of calibration. The example data he gives are

• 1.0 PO2 gives 43 mV (10% error from linear) -- 0% error after single-point calibration
• 1.6 PO2 gives 65 mV (15% error from linear) -- 5.5% error after calibration, showing a PO2 of 1.51

That's one of the reasons to compare to expected/calibrated values (either mV or PO2 works) -- error is reduced for this type of cell degradation. Certainly you can write down the measured/expected mVs (43 mV=1.0 bar, and 69 mV=1.6 bar) for each cell, I just don't see the point when displayed PO2s are far more relatable. (That said, I would have binned that sensor after seeing that 10% mV error.)

 

[LFMarm]

I am not good at doing math underwater so really appreciate the ideas on how to do quick checks during a dive.

For real testing, I prefer to do periodic dry linearity tests across the full range 0.3-2.0

Post in thread 'Oxygen Sensor Fundamentals'

May 5, 2025

Very true. This is why I started to periodically test sensors across the whole pO2 range.

• LFMarm

• 

 

[piranhaman00]

While Bobby's blog brings up many good points (especially the in-water test), it does neglect the impact of calibration. The example data he gives are

• 1.0 PO2 gives 43 mV (10% error from linear) -- 0% error after single-point calibration
• 1.6 PO2 gives 65 mV (15% error from linear) -- 5.5% error after calibration, showing a PO2 of 1.51

That's one of the reasons to compare to expected/calibrated values (either mV or PO2 works) -- error is reduced for this type of cell degradation. Certainly you can write down the measured/expected mVs (43 mV=1.0 bar, and 69 mV=1.6 bar) for each cell, I just don't see the point when displayed PO2s are far more relatable. (That said, I would have binned that sensor after seeing that 10% mV error.)

How did you calculate the 5.5% and PO2 of 1.51?

 

[inquis]

How did you calculate the 5.5% and PO2 of 1.51?

When calibrating, the computer locks in a conversion factor of actual/measured, so C=1.0 bar/43 mV. When later measuring 65 mV, that is multiplied by the conversion factor, yielding 1.51 bar for the displayed PO2. The relative error is (displayed/actual - 1) = (1.51/1.6 - 1) = -5.5%.

 

[mr_v]

Mods, please move this to Basic Scuba. (Edit: I started with a sarcastic remark and thought I deleted it. Sorry, the original remark was rude.)

I am not good at doing math underwater so really appreciate the ideas on how to do quick checks during a dive.

For real testing, I prefer to do periodic dry linearity tests across the full range 0.3-2.0

To check cell performance at any time, use your computer to select available gas that will produce PPO2 close to or equal to the target PPO2. Note the expected PPO2 and flush the loop. If you hit the expected target, your cells are linear up to that PPO2.

If cells read lower than expected, select the next leaner mix and try again. Determine the PPO2 that cells can read while staying linear and use that PPO2 to finish the dive on the loop, if you want to. Exercise common sense, if your cells are linear only to 0.5, staying on the loop is probably not the best idea.

If cells read higher than expected, there is something off. Select the best emergency procedure for the conditions and turn the dive.

 

[LFMarm]

Thanks everyone for their valuable suggestions. Here are my takeaways:

• Before diving:
  • Full linear dry test on sensors to see if they are showing signs of current limitation or non-linearity and need replacement, regardless of their age.
  • Daily calibration with pure O2 and comparison with readings in air. If deviation from linearity is larger than 5%, consider changing sensors (or check that calibration in pure O2 is done properly).
• During dive:
  • Periodically manually add O2 and see that all sensors track to ~1.6 (exact amount of O2 added is not important, here we are just checking that all cells move in sync).
  • In case a problem is detected at the previous step, do a dil flush and compare with DilPO2 calculated by Shearwater for that diluent at that depth and identify which sensors are reading incorrectly.
  • O2 flush at 6m/20' to check that cells gets up to ~1.5 (as long as they are reading the same value, not an issue that it's below the theoretical 1.6 because most likely the loop is not completely flushed with O2 and O2 is safe to breathe at this depth).

 

[LFMarm]

Recent anecdote on this topic. A couple of weeks ago in the first minutes of a dive I was descending and noticed that my controller was not agreeing with my monitor (about 0.3 pO2 difference). I immediately checked if the 3 readings for controller were similar between each others and same for the monitor; they were, which told me that the cells were likely working well but calibration of either the controller or the monitor was off (i.e., conducted at a time when the loop was not full of O2). To rule out which of the 2 was correct, I did a dil flush and unfortunately found out that my controller was off and reading higher pO2 than actual (which means that my actual setpoint was lower than intended). If it was a long deep dive, I would have swapped underwater the connection between controller and monitor so that my correctly calibrated computer was the one driving the solenoid. This was an easier dive so decided not to swap and rather set the controller to 1.5 (that translated in actual 1.2) and then manually add O2 to stay closer to my desired 1.3. Off course the deco calculation on my controller were off by several minutes as it was under the assumption that my pO2 was higher for the whole dive. I just ignored the ceilings on my controller (which annoyingly was constantly vibrating and showing all sorts of scary messages in red ) and based all my deco on the monitor. I don't think I will ever forget again to make sure that the readings between monitor and controller are the same before jumping in the water.

 

[Heat Miser]

This document has been debated on this forum before.

However all cells are non-linear at some level ... Take all the cockamaimy schemes discussed on this post and run them on this set of cells and tell me how your stategy goes.

source: Understanding Oxygen Cells

 

[Mobulai]

This document has been debated on this forum before.

However all cells are non-linear at some level ... Take all the cockamaimy schemes discussed on this post and run them on this set of cells and tell me how your stategy goes.

View attachment 907795

source: Understanding Oxygen Cells

I’ve been observing lately a recent trend of recycled arguments and discussions, some threads have been utter trainwrecks (deep air stuff for example) and some have actually been great “meta-studies” and filtered out info quite nicely like this one (I don’t have to shuffle through various threads to get good coverage of the concepts — as it it were the pub)
Nothing is new under the sun

Now that the mini-rant is out of the way; here’s another: nothing is ever linear, the whole point of “linearization” in modeling is to extract a useful range for the context of the application
Linearity in this context has to be paired with the current limiting, as demonstrated by the plot above: if my cells limit at 1.5 that could be disastrous (if I actually have 1.9+)
That’s why I really like the idea of doing o2 flushes deeper than 6m to “sanity check” — if only briefly (and/or off loop)

Thank you all for such an informative and non distracting thread btw (very useful to a noob ccr diver like me) — it’ll prove useful as well for SEO searches on google and provide good reference to others in a pinch