CO2 sensor design

PacketSniffer

__________________


Far better it is to dare mighty things, to win glorious triumphs, even though checked by failure, than to take rank with those poor spirits who neither enjoy much nor suffer much, because they live in the gray twilight that knows not victory or defeat.

AD_ward9

May I shed some light on the issues of why CO2 sensors on a rebreather are available from just two companies for now (HSM Engineering, and DL).

1. CO2 sensors using infra-red need a constant emission spectrum. Pressure and helium both change the spectrum very considerably by cooling the emitter due to the increased thermal capacity of the gas. Sealing the emitter works for a while, until helium gets in. There are better solutions, which if someone collars me I can reveal, but a bit too detailed for here.

2. Liquid water (condensation or the product of washing the rebreather) in the measurement channel kills the IR signal. Unlike an O2 sensor, there is no liquid on the other side of a hydrophobic membrane to prevent a fairly low pressure of wash water from migrating through.

3. High Planck radiation background from the highly humid gas leaving the scrubber.

4. Momentary false alarms a scrubber exhibits as its capacity reduces (mean is OK, short term signal can be quite high).

5. Power consumption: the DL CO2 sensor takes 170mA when on (from 3V). Switching it on for a second every 10 gives 17mA: still a fair bit. This means a beefy battery.

6. Need for helium compensation, which takes more power still: typically 250mA from 3V.

7. Problems of calibration. Background levels vary seasonally and depending on where you are. We have found levels over the range of 200ppm to 1000ppm: the world mean is reportedly 380ppm just now. Calibrating in a poorly ventilated building is obviously not a good idea, but we got 1000ppm in a harbour in Orkney Isle one day. A 5:1 calibration range means a 500% error in the result.

8. Fluoroscent sol-gels are very sensitive to contamination. If anyone thinks they have one that works, send us one and we will send back a nice (confidential) report free of charge as we would love to have one work, and we know exactly what makes them fail - checking that is quick and easy.

9. Size. IR CO2 sensors for use in an office have reflectors. Try that in a rebreather, and you will find the light path changes. The path needs to be direct. The shortest direct path we have got working reliably is 35mm (which is what we use). Now add on the sensor and emitter, board etc, and it quickly grows to 100mm. Where in your loop can you simply plug in something so big?

10. Cost. How many rebreather divers would buy one of these, if they cost 1K Euro. I bet under 1000. What do you think the development cost is, and the manufacturing cost? Answer is a negative return on investment: i.e. the developer subsidises the diving of complete strangers. All technical things above are solvable, but the two companies with CO2 sensors built them for a larger market than sports diving.

Alex

NB: the Laser IR or sol-gel O2 sensors are even worse. Companies that claim these are suitable for rebreathers are lying, plain and simple.

Scubascooby

I don't know exactly what sort of straight path is needed but both the KISS and Inspiration have lengths of straight pipe but they are on the exhale side. This is probably the least of the problems.

__________________
"dove" is NOT the past tense of "dive"

Better off OUT of the EU !

www.hugsac.org.uk

swampdiver

Thank you for that excellent summary Alex.

We have looked at the issue of elevated ambient CO2 levels here in Toronto when during the winter with the loss of plant photosynthesis to convert CO2 to oxygen and also with the increased CO2 production from the burning of natural gas to heat buildings the ambient CO2 levels often reach 600 ppm on windless mornings.

Calibrating indoors as you point out could also be problematic due to elevated CO2 concentrations from human respiration and poor building ventilation.

Here are some CO2 measurements made from around Paris, France. Outside of the main city the ambient level is closer to 380 ppm, around parking lots concentrations rise to close to a 1000 ppm, and in one classroom the CO2 concentration approaches 5000 ppm.

The table is reproduced from this reference.
CO2 Science

Attached Images

Paris Data.jpg (105.5 KB, 292 views)

deepwrecksc

Yes this is a problem for nearly all optical spectroscopies where measurements are made by direct absorbance/fluorescence intensity. Any fluctuation in source intensity (from emitter aging or from interferences) will change the response factor. One simple and usually effective way to account for this is to use a ratiometric method by which multispectral data are internally referenced. Most modern analytical sensors based on absorbance/fluorescence are built this way now.

Yep - these and more (including what you say below) are all good reasons why I think IR spectroscopy is not the best technology to use in rebreathers. It is a great method for making measurements under ideal conditions in laboratories or with big portable instruments, but like mass spectrometry - it doesn't lend itself well to development of cost-effective sensors.

Sol-gels aren't themselves fluorescent, they are merely the support for the fluorescent luminophore sensors. There's nothing magical about them - simply an organosilane polymer (or fully inorganic silica matrix) formed into a microporous structure. You can also use organic polymers for this application in some cases. There are many possibilities here and unfortunately there are only a few groups who have regularly published on fluorescence-quenching CO2 sensors. Take a look at the references in the papers I posted. If I can find some time later this summer I will assemble a few DLR fluorescence chemistries for CO2 and send them off to you. Do you have a phase fluorometer to use for testing? Ocean Optics markets a relatively inexpensive one built by TauTheta (good optics company). I can imagine that a CO2 sensing element similar to the Ocean Optics Redeye patch could be made inexpensive enough to simply discard after every dive. I could certainly make them for a few dollars each once we sorted the chemistry properly.

The cost is an interesting question. All comes down to which technology ultimately is used. The most cost-effective way would be to make use of all of the R&D that has ALREADY gone into analytical sensors for CO2 in academic labs. One thing many people (and companies) don't realize is that just because a company doesn't make a product already does not mean that the academics haven't worked it out! Sometimes it can be a challenge to get access to the published literature, but often all it takes is a trip to your local University library.

Lee

deepwrecksc

Thanks for the link - interesting reading!

AD_ward9

The problem is if you take a typical IR emitter (silicon micromachined emitters for example), and put it in helium under pressure, there is no emitted spectrum at the 4260nm absorption peak. This kills it dead. There are some solutions to this: I write them up when I have time over the next couple of weeks and post them here.

The sol-gels work by shining an ultra-violet LED at them, then switching the LED off and looking at the time taken for the light (of a longer wavelength) to subside. Due to the short time for decay, this is implemented using a phase difference approach. The LEDs are not subject to variation due to pressure or helium, but the sol-gels themselves seem to be very subject to contamination and some do not like helium.

The problem is not the chemistry (the science), but the engineering. It is the contamination problem I mentioned. Yes we do. I would be very happy to test some of the sensors you send.
I will do the write up I promised and post it. The problem again is not the lack of science but the engineering issues that I listed.

Alex

AD_ward9

Thanks. That is a really useful reference. I had not seen the 5000ppm classroom before, but certainly have been in some lecture theaters that felt like that!

Alex

Faceless

IMHO a possible approach for CO2 architecture could be a dual-channel IR with some algorithm based on op-amps with a minimal temperature drift & mathematical model to compensate output variations due to changes within gas mix, temperature, & ambient pressure.
As an additional aid we could also calculate CO2 more precisely if we'd know the loop volume & amount of gas being injected into the loop, then with given ambient pressure & known gas (He+O2) mix we could calculate CO2 with PPO2 change over time
Also, there are bad side effects on helium ingression into CO2 sensor, so it should be totally sealed with some highly transparent glass, resistant to max possible pressure. Also sensor should be well-placed within loop at coldest place to minimize drift for example a place where cold water is
condensing.

__________________
Peoples risk perception is almost always focused on the most dramatic rather than the most probable. (c) Gordon Smith

RZEP17

How dificult is making a rebreather workable He and N2 sensors?
If designing CO2 sensor is so dificult, what about trying to determine everything else in the mix and then making simple math?
rgrds
Tomek