Measuring O2 in enriched air

[jgttrey]

How did I miss this goat rope? Better late than never I guess.

Technically, galvanic voltage sensors are extremely sensitive to temperature. They address this with a small chip that compensates for the temperature. Without compensation, the sensor output would vary wildly. In most instances, the chip is part of the O2 sensor itself but a few designs rely on an external circuit for temperature compensation and the disposable sensor unit has no built in compensation circuitry.

The compensation circuit takes temperature mostly out of the equation except for one scenario.

Many O2 sensors have the compensation circuit at the base, opposite end from the membrane. If you blast cold gas on a warm sensor, the temperature compensation will not be accurate because of the differential between the membrane side and the side where the temperature sensor is located. The compensation is less accurate the bigger the difference between the temperature that the circuit is seeing vs the temperature where the galvanic reaction is occurring.

So, bottom lining it, you don't want to blast a whole bunch of cold gas on to a hot sensor. Another reason to keep the flow limited and why I like units like the Divesoft where I have not only a flow limiter but a hose that let's the gas temperature warm up to something closer to the ambient temperature of the O2 sensor before it gets there.

 

[jgttrey]

p.s. If you really are interested in this stuff, this is a good reference and will tell you more about O2 sensors and best practices for O2 measurement than you'll ever want to know.

https://www.amazon.com/Oxygen-Measu...nt+for+divers&qid=1567114490&s=gateway&sr=8-1

 

[tarponchik]

LOL! I rarely have any idea what you are arguing about, other than to be contrary.
Re (1): of course there is a temp effect, as there is a pressure effect, and a humidity effect....what do you think the relative importance of those factors is?
Re (2): basic thermodynamics, why are you even asking?

LOL, heh? So you agree with what I said, after all. Yet at the same time you enthusiastically claimed that I was wrong when I said those things you agree with. I am at a loss.

 

[tursiops]

How did I miss this goat rope? Better late than never I guess.

Technically, galvanic voltage sensors are extremely sensitive to temperature. They address this with a small chip that compensates for the temperature. Without compensation, the sensor output would vary wildly. In most instances, the chip is part of the O2 sensor itself but a few designs rely on an external circuit for temperature compensation and the disposable sensor unit has no built in compensation circuitry.

The compensation circuit takes temperature mostly out of the equation except for one scenario.

Many O2 sensors have the compensation circuit at the base, opposite end from the membrane. If you blast cold gas on a warm sensor, the temperature compensation will not be accurate because of the differential between the membrane side and the side where the temperature sensor is located. The compensation is less accurate the bigger the difference between the temperature that the circuit is seeing vs the temperature where the galvanic reaction is occurring.

So, bottom lining it, you don't want to blast a whole bunch of cold gas on to a hot sensor. Another reason to keep the flow limited and why I like units like the Divesoft where I have not only a flow limiter but a hose that let's the gas temperature warm up to something closer to the ambient temperature of the O2 sensor before it gets there.

Agreed. I like to analyze from my LPI hose, using a flow limiter, and calibrating to an air tank before hand to take the humidity out of the equation. The old Analox chart is a useful reference:

 

[jgttrey]

Of course, none of this makes a great deal of difference from recreational EANx. A bit of difference here or there is not going to make any never mind.

But, it can matter for really rich mixes. The chart is a good illustration of why I like to calibrate off 100% O2. If you calibrate off air and you're off by a little, either because of technique, failing to adjust for humidity, having a reference gas that is a little off, etc., that discrepancy will be multiplied when you analyze richer mixes.

Let's say that between inherent cell accuracy limitations +/- 1% and humidity, etc., you are 2% off (19 vs 21, or 23 vs 21) when you calibrate. Since the unit assumes linearity of the relationship between voltage and O2, that error will be compounded as the mix gets richer. No big deal on recreational nitrox, where what you analyze as EAN32 might really be EAN35 or EAN29. But, if you're testing EAN80, you might show EAN72, or EAN 88 (roughly).

On the other hand, if you calibrate off pure O2 and are off by that same 2%, that's the outside range of error. Any leaner mix you analyze will be off by less than that, in theory.

It also gives you a good idea of cell linearity and can help identify sooner when the cell is going bad.

 

[tarponchik]

Of course, none of this makes a great deal of difference from recreational EANx. A bit of difference here or there is not going to make any never mind.

But, it can matter for really rich mixes. The chart is a good illustration of why I like to calibrate off 100% O2. If you calibrate off air and you're off by a little, either because of technique, failing to adjust for humidity, having a reference gas that is a little off, etc., that discrepancy will be multiplied when you analyze richer mixes.

Let's say that between inherent cell accuracy limitations +/- 1% and humidity, etc., you are 2% off (19 vs 21, or 23 vs 21) when you calibrate. Since the unit assumes linearity of the relationship between voltage and O2, that error will be compounded as the mix gets richer. No big deal on recreational nitrox, where what you analyze as EAN32 might really be EAN35 or EAN29. But, if you're testing EAN80, you might show EAN72, or EAN 88 (roughly).

On the other hand, if you calibrate off pure O2 and are off by that same 2%, that's the outside range of error. Any leaner mix you analyze will be off by less than that, in theory.

It also gives you a good idea of cell linearity and can help identify sooner when the cell is going bad.

There is another reason to calibrate with 100% O2: sensor aging begins with non-linearity at high O2 percentage. Thus, for your EAN80 you'll get the reading of 75%, etc.

 

[Storker]

Can you guys take this to a PM thread?

 

[ScubaWithTurk]

...provided the flow rate is low enough that you don't get an increased pressure at the sensor.

An O2 sensor is an electrochemical cell. Its voltage output is proportional to the PO2 close to it, not the FO2. So if the flow is so high that the cell experiences a pressure significantly above ambient, it will give a too high reading.

This is exactly why I teach my students to slowly open the cylinder first to make sure you have a light flow and THEN put the analyzer against the valve. They are NEVER to put it against the open orifice and then open the cylinder.

I also teach to calibrate it on a cylinder of air prior to analyzing their Nitrox.

 

[kelemvor]

The only details I can give you is that the density of O2 is some 10% higher than that of air. (Gases - Densities).

But I can give you an illustration. Ever ordered, seen or heard of layered drinks? You make them by first adding the heaviest liqueur, then the others in successively lower densities. Just be careful to avoid mixing while pouring. Pour slowly, preferably over a spoon. Now, one of the more famous layered drinks, the B-52 is made from Kahlua (density: 1.13), Irish Cream (1.06) and Grand Marnier (1.03). Note that Kahlua is only 7% heavier than Irish Cream, and Irish Cream is only 3% heavier than Grand Marnier.

Since Mother Nature abhors differences, if you let that B-52 stand long enough, it'll end up as a dull mixture due to what we call diffusion. Kahlua moves into the Irish Cream and subsequently into the Grand Marnier layer, Irish Cream moves into the Kahlua and Grand Marnier layers, and Grand Marnier moves into the Irish Cream and subsequently into the Kahlua layer. And you're left with a shot glass of a mixture which lacks merit both in flavor (which it always lacked) and in appearance.

Same thing with a PP EAN mix if you top up with air slowly enough that you don't get much turbulence and mixing in the tank. If you top with air pretty fast, you can compare that with splashing the Irish Cream into the glass atop the Kahlua (it'll mix pretty well, and you won't get that cool layer effect), but if you top slowly - which you probably would want to do, since that reduces the temperature increase when you fill - you're basically doing the same thing as if you were pouring the Irish Cream slowly into the glass on top of a spoon, and you'll get a heavy O2 layer at the bottom and a lighter air layer at the top. Let it stand, and Mother Nature AKA diffusion will even out those differences, but you can speed up that process. Lay the tank on its side, and the cross-section of the diffusion front will be a lot larger, leading to shorter time for equilibrium. Throw the tank into your car's boot and drive a bit (preferably on bad roads; we don't have much else), and the tank will be shaken, which improves mixing. Roll the tank back and forth on the floor, and you'll mix its contents.

When I'm filling my tanks from our club's inline mixer, I fill, cool and top off with air. I never analyze for marking the tanks before I've driven home with my tanks, usually the day after or later. It's good practice if you believe in the physics, and it makes life easier for a lazy bastard like me.

This ^^^.

Some shops really slam the air in after the O2, which mixes the contents well. You can analyze mostly right away. However, some people believe that slower fill rates are safer (like 600 PSI/min max for air), and that generally does not stir the pot--er, tank--vigorously enough to get an accurate reading right away. My fill whips have flow restrictors, so it takes some time for the contents to settle unless the tanks are rolled, shaken, taken for a drive around the block, or whatever.

Thanks! I think I understand, and also understand why I don't have to bother waiting or rolling tanks at my LDS. The tanks are indeed filled pretty quickly.

 

[broncobowsher]

View attachment 537487

You confused me for a moment. I was wondering why Randall was putting an old comic up on Friday? Then refocused to realize I was not on my Friday morning XKCD viewing yet.