Scrubber Sensor

PaulTG2

Hello,

I am begining to understand how people come to make their own rebreather. The concepts and options on how to "improve" rebreather design is facinating. It's a good thing I don't have enough time on my hands to actually start building on. Like my airplanes, I'd rather fly them than spend all the time building them.

Why put all the sensors essentially in one place to tell you essentially the same thing. Granted it makes figuring out if one it bad super-simple, but it seems to me if you place individual sensors around the loop you can use software programming to tell if any are far enough off "reasonable" to isolate that sensor while using the enhanced dataset to tell you much more about what is happening.

In addition, while we don't currently measure anything but O2, changes in the O2 can indirectly measure the CO2 concentrations.

Assume this O2 sensor configuration:

* one at the mouth input (issoalted from inhalation bag)

* one at the mouth output (issolated from the exhalation bag)

* One at the entry to scrubber (issolated from exhalation bag

* one at the exit to the scrubber (issolated from inhalation bag)

Issolation is just done through one-way mushroom valves to prevent injections and additions from changing the readings significantly.

Using these sensor points you can use averaging and software logic to issolate any sensors that have gone bad.

You can also use the data for the computer to better understand what's actually happening in the rebreather.

Using the mouthpice in and out you can tell what the diver is using. The only meaningful difference in O2 readings should be make up in expired CO2.

For the in and out at the scrubber the only difference should be the removal of CO2 which will tell you about the scrubber health.

I imagine you can tell alot more if you give some thoughts to having the four different readings.

So, why shouldn't this work and be better than putting all the sensors together.

One downside I can see is that there are more holes and wires to various parts of the loop -- increasing some complexity and opportunities for leaks.

Just keeping thinking.

--Paul

AD_ward9

Putting sensors around the loop would give confusing readings. You also mention use of a magic wand "Software programming to tell ..". In engineering, all magic wands work against you.

The way a safety engineer solves this problem is to start with a proper study of how sensors fail. We did a 5 year study.

There are only 2 fault modes, out of 14 fault modes that O2 sensors have, that causes them to read high. Both of those two modes are very rare, and both can be avoided completely by using the right sensors.

There are some modes that make sensors change slowly, with tendancy to low. All other modes simply cause the sensor to read low. By "low" or "high" we refer to a reading which is lower or higher than that expected after calibration. For example, a Sensor A reads 10mv in air, and Sensor B reads 8mV. In a PPO2 of 0.42, the first sensor reads 19mV so is reading low, but Sensor B reads 16mV which is reading neutral.

This means that if you have calibrated the sensors correctly, then if you have n sensors, the one that gives the correct reading, is the one that reads highest. The only time one has to do another check is if the PPO2 is reported to be falling, such as on an ascent, check that the sensor you are using with the high reading, reacts to O2 injection as fast as the others.

Once you can pick out which sensor is working out of n, adding sensors in one place, can give you whatever reliability you want.

Cheers
Alex

Faceless

In other words, what we truly need it a kind of finite state machine, which will incorporate not yet another plain stupid voting logic among O2 sensors, but rather a predictable one, taking into consideration O2 sensors fault model and current dive condition (descent/ascent/hovering).
Interesting idea ...

rdmmdr

the o2 sensor response and accuracy is insufficient to do this with our current sensors. and the detector filters in the Ir sensors are not made for the pressure. i tried the reactive sensors but it was unstable under pressure, i suspect due to the membrane. the current best we seem to have is the temp stick design. i have been thinking about trying a modified detector bench for the ir type. where by you could take a standard detector and encapsulate it inside of a tube at the end with holes drilled crosswise for air flow. the filter would have to be suspended in the epoxy with some type of light shield around the edge. and all of this would have to be cleared for a high o2 environment.

it would be a fun project but have too many others going on right now.

rick

caveseeker7

One other item to consider is WOB. The one way way valves are among
the major design problems contributing to breathing resistance, the more
you add the worse it'll get.

__________________
Cheers
Stefan



"It is still a good day if you are on the green side of the grass!

Su amigo Roberto!"

Sponsor Lou in Race For Life!

Benthic

Alex,

I'd be interested in reading the findings of the study you mention. Is it available here on Rebreather World? If not, can you point me in the right direction?

Thanks!

Brian

__________________
Well lit rooms will not repel determined beer ninjas.

AD_ward9

The purpose of the work was to find good O2 sensor vendors, identify the failure modes and come up with a scheme to overcome them. The report spends a lot of time going over failure modes which one vendor seems to have far more of than the others (Teledyne, due to design as well as production issues). Exposing all the dirty laundry of one sensor vendor in too much detail is probably not good for the industry, so we will keep the report itself confidential for now.

Our clients do have copies of the report, as do each of the sensor companies whose sensors we used for the trials.

We do still have some work to do on sensor approval. We are trying to accelerate the final acceptance of Analytical Industries PSR-11-33s for the Open Revolution products: they are by far the best available, based on the information we have at this time.

The following link has some extracts and pictures from the report:
http://www.rebreatherworld.com/megal...ghlight=Insovt

Here is an extract of the things that can be helpful on the failure modes. This lists 11 failure modes overtly, but the environmental failures include 4 modes as there are 4 things to fail, so there are 14 in total.

"
1.1Failure Modes Common to all Galvanic O2 Sensors

• Unused sensors in storage will fail eventually due to the water carrying the KOH electrolyte evaporating. This results in a zero output from the sensor. This fault is predictable and is eliminated by the manufacturer determining the rate of evaporation and amount of electrolyte in reserve, and stating a shelf life. For example Insovt and AI both state a 60 month shelf life for their sensors. This is specific to the packaging: if the package is opened then the stated shelf life no longer applies, but the sensor service life then comes into effect.

The maximum life of a sensor is the maximum of the service life and the operating life. For example, if a sensor with a 5 year shelf life and 2 year operating life is stored for 4 years before being opened and used, then it must be discarded after one year.
The stated service life must be down rated for pressure and temperature. A 2:1 downrating is appropriate: this figure is based on discussions with several manufacturers.

• Used sensors fail permanently due to a combination of factors:
  • Exhaustion of the anode surface: this occurs because the reaction consumes the lead anode in the presence of O2. The result is first an increase in the response time of the sensor, then in a voltage limit from the sensor. That is the voltage output from the sensor is linear to a particular PPO2 level, then flattens off and becomes fixed, not increasing with increasing PPO2. This fault is tagged the “Ceiling fault”.
  • A temperature drift, of up to 2.5% per degree Celsius if the temperature compensation circuit fails. This generally results in the output voltage falling, that is the sensor reading low, but a sudden temperature drop can cause a cell to read high, depending on the fault.
• Sensors can fail temporarily if water is allowed to fall or condense onto the face of the sensor. This results in a dramatic increase in the response time from the sensor and the sensor reading a much lower PPO2 than is the reality. This can be avoided by good design: the sensors should be mounted such that water falls off the face and cannot collect. During descent divers are often head down, but during descent PPO2 rises in the rebreather so this phase is not critical: if the sensor has a delay before showing rising PPO2 during descent it does not affect the diver’s safety dramatically. During all other phases of the dive, the diver is either horizontal or head up, so orientation of the sensors toward the scrubber is ideal.

1.2Failure Modes Arising from Design or Manufacturing Defects

The trial identified the following failure modes and rates that are specific to particular sensor designs and construction:

• Low shelf life. The reason for this is inappropriate packaging (the sensor bag should not be impervious to gas), and less than optimal design of the sensor allowing the water in the KOH solution to evaporate. To assess this a control group of sensors in each batch was stored in an office environment, then half of the batch opened half way through the manufacturer’s stated shelf life, then if those sensors still operate, the remainder are opened at the end of the stated shelf life and tested.
• Drift. Good cells exhibit very little drift. For example, the Insovt cells tested here did not exhibit any measurable drift over a period of 5 years. In contrast, all Teledyne cells drifted every month, until they failed. Discussions with cell manufacturers exposed the reason for the drift to be organic contamination. The KOH solution is very aggressive and if contaminated by any organics, the result is usually a reduction in the cathode area. This results in a gradual reduction in the cell output. The cathode continues to be damaged by the contamination and the cell will fail early. Sources of contamination include soldering to the cathode (a process unique to the Teledyne sensors), use of epoxy resin to seal the wires into the cell chamber (another weakness unique to the Teledyne sensor), detritus introduced during assembly. AI and Insovt go to great lengths to eliminate this failure mode: down to << items removed to preserve confidentiality>>.
• Helium bubbles in the electrolyte causing fluctuations in output level. These fluctuations tend to cause the cell to read low. The cause can be either the front cathode being bonded to the sensor membrane, or being allowed to move, or the lack of a buffer before the membrane, causing the electrolyte to press on the membrane and causing it to dome.
• Pressure migration of helium into the sensor causing early rupture of the rear plastic film designed to contain moisture. This results in the sensor failing with a lower output than expected due to drying of the electrolyte, damage to the circuit board or temperature compensation circuit. The plastic film is behind the sensor, and fills with gas during the dive. During the ascent the gas expands and normally diffuses back through the electrolyte and the hydrophobic membranes. If the ascent is too fast, the rear membrane can rupture, and the electrolyte dries out. This is a design defect caused by inadequate strength of the membrane and inadequate off-gassing pathways via the front membrane. The result is usually the cell reads low, but can cause an increase in the cell output if the pressure pushes the cathode towards the anode, or visa versa.
• Blockage of a pressure relief port. This caused a reduction in the output of the sensor in the sensor examined. The possibility of this fault is a design defect: it cannot occur in the AI or Insovt designs. The result is the cell reads low.
• It is claimed on Internet forums that if the sensors are stored in a high O2 environment without a load attached, excess charge accumulates resulting in the PPO2 reading higher than it is. On detailed investigation, this failure turned out to be dry joints: the bare micro fuel cells do not exhibit this behaviour – the fault is caused by faults in the temperate compensation circuit, such as dry joints or faulty components. The result is the cell reads low.
• Environmental damage, particularly corrosion of contacts and the circuit board in an operational rebreather environment. This results in an open circuit or a zero output from the sensor. The Teledyne sensors are particularly prone to this due to the construction depending on the rear membrane: if that membrane leaks, then Potassium Hydroxide is deposited onto circuit board, causing rapid degradation of the board, the components on the board, and failure of the cell. The other cells tested had either a solid wall behind the cell or an improved membrane to prevent this occurring.

"
Hope that is helpful.
Cheers,
Alex

Johnny Christensen

Hi Alex

You have made many references to the Analytical Industries PSR-11-33s being the better sensor. I have done some searching on the internet, trying to find that for my Inspiration but have only found the AI PSR 11-39-MD. Is this the same sensor that you have tested, only in a 3 pin molex version ?

I don't care about price really, I just want the superiour cell.

regards
Johnny

__________________
www.johnnychristensen.com
Wreckphotography from the cold north

SINUS

Cropped out from Aii1's site. It does'nt fix your inspo.

Attached Images

PSR_11_33.jpg (66.4 KB, 203 views)

__________________
"Silence Is Always Beautiful"

AD_ward9

You have cropped only a few from the 11-33 range. Expand the scope and you will find what you are looking for.

AI recommend their PSR-11-39-MDs as the R22 replacement. You should find the AI PSR-11-39s an improvement over Teledyne R22s.
Cheers
Alex