nitrox analyser

[ultravinnie]

comments are of course always welcome.

the MOD / TOD idea is a good one. i'll program the bottom button to show a second reading (which is also presettable in terms of PPO ) if you just push briefly on it ( the menu needs a 2 second push to activate. so if you just tap it it will show an alternate setting.

And i got a question. the setup is now for PPO 1.2 to 1.6. Anybody want to run at something else ? ( 1.4 and 1.6 are most commonly used. anything above 1.6 is special , and anything below is not really 'nitrox' )

The calibration issue. and the necessity of 100%

The sensor curve is a so called offset-gain path.
that means the the output is defined as follows

Fo2 = offset + (vout x gain )
Vout is the voltage produced by the sensor
Fo2 is the fraction of o2 in the mix

There are two ways to find out the offset : 'starve' the sensor : that means close the intake to the sensor and let it 'die'. This can take a while and you have to be sure that there are no leaks. Due to the small difference between the null points the accuracy is reduced. ( ideally you should measure 0% and 100% )

The other way is applying 100% : take a reading , apply a different known gas that is as far away in terms of Fo2 as possible ( improves the resolution ) Air is readily available.

You then get two equations:

100% = offset + ( Vout1 x gain)
21% = offset + (Vout2 x gain)

solve the above equations to Gain and offset and you got the exact slope of the line nailed down. A line is defined as something connecting two X-Y points. you can not define a line using only one point since there are an infinite number of lines that cross any point.

that was my main frustration with ALL the analysers out there. Not 2 of them seem to agree witch each other and none take into account the sensor offset.

A one point calibration can not take into account the sensor offset. ( which is in the range of 7 to 8 % Fo2 ! )
So a reading on for instance 34 % can be off by as much as 2 to 3 %.

anyway your comment is valid. I will play around with a different model.
I have a bunch of the sensors. I will baseline them all to 100% and see how far they are apart from each other. , make a 'generic' baseline for 100% out of that and store that as 'default' in the algorithm. Then run them on 21% and see how the spread is in terms of accuracy. If this works out i can modify the algorithm as follows

Unit leaves the factory with a 'generic' baseline. User can do a short-cal to compensate for sensor aging. ( can be done on power-up for instance ) I'll make this an option in the menu : ShortCal . Simply hook it up to a tank of air and run the shortcal. This gets the baseline for 21% ( technically 20.9% and that is also what i am using in the math onboard )

The longcal will still be in there ( that one requires access to 100% ) and that one can override the 'default' baseline.

Any more idea's are welcome. I got plenty of memory left in the CPU to stuff additional things in there.

 

[homo maris]

Here are a few suggestions.

For the calibration (I have no idea if that would work)
As an alternative to 100% O2 why not use CO2 (easier to obtain) or any inert gas like nitrogen, argon or helium or gas like methane (i.e. gas without O2 and easily available). For that matter, could water be used to starve the sensor. That way you are sure there is no leak.You could give the option to calibrate with 100% O2 or 0% O2.

Air is 20.95% O2 under dry cool conditions but that percentage decreases with temperature and humidity. This becomes important in tropical areas if not using cool dry air. Have you considered calibration involving moisture and temperature or is the sensor selected impervious to that. This point is moot if using dry cool compressed air to calibrate instead of ambient air.

It would be nice to have a visual alarm if the gas is hypoxic (e.g. hypoxic trimix). Say show the %O2 and a warning.

As for ATA, there could be two settings:
0.5 to 1.6 and, with some sort of a security override, 1.7 to 2.0

While you are at it, why not display EAD/END as well for the ATA selected?
Also display maximum O2 exposure time limit as per NOAA tables if there is some memory left.

A battery level readout or number of hours of operation would be nice.

Are you averaging the sensor reading or displaying the absolute value?

Of course it is a lot to ask, but what would it take to add an He analyzer?


Thanks again and I can't wait for my yellow one,
JL

 

[ultravinnie]

water usage is impossbile. the sensors are sensitive to humidity. Therefore you NEED to calibrate from a scubacylinder. That air is moisture free ( hopefully , or else :find a new fill station )

calibrating with 0% has 2 problems
first of all it decreases the accuracy. ( the higher the delta between the two calibration gases the higher the accuracy. ideally you would use 100% o2 and 100% N )

I dont agree with two settings. i dive frequently at 1.4 and 1.6 ( tuning the MOD )
but i can easily make it tunable between 0.5 an 2.0 in steps of 0.1

Why do you care about the END or EAD ? if you dive nitrox your computer calculates based on the PPo2 and all you care about is the MOD associated with it ( to avoid ox toxing and busting your ox clock).
Keep in mind that i am NOT familiar with trimix diving. i may be going off the deep end here ( so to speak). I am interested in the technicalities of trimix but i have no personal interest in diving anything deeper then 140 feet, or diving doubles for 2 hour bottom times anyway so why bother with trimix. That does not mean i am not willing to develop the analyser. all i'm saying is that i personally am not interested in such diving. My gear is minimalistic: A scubapro travelwing with barely 25 pounds of lift, a HP steel 100cuFt full with nitrox to my liking (3400 to 3500 psi ) and my little yellow divebuddy ( my HD videocamea ). thats plenty of diving. i hold out two dives with one tank , am the first in and the last one out ( typically by more then half an hour )

Battery gauge is already in : it shows on powerup for 2 seconds as a bargraph

Sensor input is NOT averaged but uses a variable slope algorithm ( digital filter comparable to a Kalman filter : look it up on google)

I have been looking at He analysis but the sensor poses a problem for the moment
Katharometers are very precise instruments ( and expensive ), hard to read and difficult to drive .Somehow i don't feel that that is what is used in the 900$ stuff out there. Real katharometers start at like 5K $ ....

does anyone have a picture of what is in the commercial helium analysers.
especically the sensor element interests me.

a true Katharometer element should have a bunch of wire. 2 for the heating element and 3 (sometimes 4 or 5 for the thermo-pick off. one (or two) comes from the reference chamber and 2 or 3 come from the sampling chamber.

Ideally the heating element should be a thin film element that is exposed to both the sampling an dreference chamber ( one side in the sampler , the other in the reference )

Katharometers work as follows : a known current is run through the heater element. A temperature sensor in physical contact witht the element gives a reading of the absolute heat beeing generated.

at a given distance ( absolute doesnt matter as long as it is the same distance in the sampling and reference chamber )
there are 2 thermo pickups. one in the reference chamber and one in the sampling chamber.

the heater is switched on for a fixed amount of time
and the two temeprature readings are done.
Heat is conducted by the gas. since helium conducts heat about 5.6 % times better then normal air. you can do the math between the reference channel and the sampling channel

Actually , a katharometer-like element is present in the exhaust of modern cars. it is used to figure out the combustion by sensing the amount of remaining oxygen. ( Bosh makes these things. A heating element and three thermal pick ups.
one far away from the measurement chamber and one close to the heater and the last on the heater itself

the one far away measures the incoming gas temperature.
a control loop tunes the heater to be a certain temperature higher than the incoming gas temperature. the pickup on the heater creates that feedback signal.
this is required as the gas is colder when the engine is just started then when running ffull power + the gas cools down in the exhaust while driving as well.
the heater needs to be at a known delta temperature above the gas temperature

Tables are not an option. that eats too much memory. besides, i have a professional aversion for tables. It is pre-chewed information. i want to know where the numbers come from. I need formulas. That is how I do the MOD for PPo also : use the gas laws , not table lookups. I'll see what i can find.
that is a good idea. it tells you how long you can stay on a certain depth with your Fo2 and PPO setting. Let me see if i can find the formulas.

 

[homo maris]

Thanks again for the explanations.
Having it tunable between 0.5 and 2.0 would be great.
I read about the Kalman filter (not sure I understand the details but not that I need to either).
The reason for EAD was that some folks dive with tables or an air-only computer. As you pointed out, not a big deal if you have a computer.

The reason why I brought up Trimix is that it can be mixed with a low % of O2. Good as a bottom gas for deep dives, deadly in shallow waters. So even if you don't plan to dive Trimix you could end up with an hypoxic Trimix fill by mistake. If the analyzer shows an alert for gas with less than 21% O2 it would help keep folks safe (I know people are supposed to have a brain too).

For He, they do not use an He sensor. The sensor they use is is non-specific and determine the relative difference in thermal conductivity of the test gas to the reference cell. Because the gas mix is known, they assume that the difference is the result of the addition of helium to the gas mixture but it could be CO2 or Argon.

Is the casing water resistant?
Thought about explosion protection for 100% O2?
Can you cover the LCD so it does not need to be put in another case?


Thanks,
JL

 

[ultravinnie]

lcd is covered by a lexan window
casing is NOT water resistant ( it is splash proof , but if you dunk it it will die )
there is no way to make an off the shelf waterproof casing. and injection molding is a very costly process.
Explosion protection for o2. Don't smoke when dealing with that stuff.
no serieously , the flow throught the system is extremely low ( in the order of maybe 1/10 liter per minute the only place where there is 02 is the sampling cavity and that is made out of Delrin which is an inert material )

 

[homo maris]

ah if you can't smoke with it then it must be DIR
This is great news.
Perhaps you could start with a batch of 10 and see if folks sign in?

JL

 

[ultravinnie]

DIR as in Diving Is Ridiculous ? ( Pun intended )

before i get flamed here - I HAVE NOTHING AGAINST DIR. Anyone is entitled to his own beliefs. we live in a free country - See my avatar.

If i need a good dive buddy i'll clip a rubber chicken off my BC. It'll always be there in arms reach , never panic and outlast me + it has excellent gas consumption and buoyancy control.
Right now my camcorder in its funky yellow housing is my best buddy. it has never let me down.

 

[spyder]

Interesting project... I built an El Cheapo and it has served me well but I would be interested in a kit or unit.

 

[carbon]

Hi ultravinnie,

You mentioned some sort of compensation for the effects of pressure. Do you actually measure the sensor chamber somehow?

I have been thinking of something very similar to your design because of the shortcomings of many of the available commercial units. I was thinking of instrumenting the sensor chamber with a pressure sensor, humidity sensor, and a temperature sensor, along with the oxygen sensor.

Carbon

 

[RV]

I'd like one too..

I wonder what happen'd to ultravinnie?