I dive with two computers, each displays different tank pressures ?

[gr8jab]

They could also be non-linearly wrong and their wrongs cross at some point...

Absolutely! Assuming a linear relationship between the sensor voltage and pressure, there are actually two numbers that need calibrated. Remember y = mx+b? The slope, m (change in voltage vs pressure), and the intercept, b (offset), must both be set.

But, remember that you have two sensors (two mis-matched computers). If they are always exactly 200psi different, as you suggest, then only offsets are to blame. However, if they read 200psi different at full tank, and 100psi different at mostly empty, the slopes are to blame (and maybe the offset too).

NOTE: It seems that most pressure transducers are non-linear, and some mathematical best-fit algorithm attempts to flatten them. Since this math is likely hidden in the 'black box' software, we can pretend the final output yields a linear relationship between actual pressure and measured pressure.

So, as a geek engineer, I would do an experiment. Call it a gauge study, or an R&R (repeatability and reproducibility study). Get a good analog gauge, that you maybe hope is calibrated, from somewhere. Fill a tank and take readings with computer1, computer2, and analog. Let some air of the tank and repeat multiple times until the tank is empty. If you know excel, you can chart (use x-y) computer1 and computer2 vs analog.

I'm rather decent at data analysis, you could send me the data if you want me to play with it.

 

[RonR]

Electronic pressure sensors are still analog sensors, the voltage needs to be converted to get digital output. In general electronic sensors will give more accuracy than mechanical SPG’s, assuming all the hardware and A to D conversion supporting that level of accuracy are in place. With small SCUBA sized dials on a typical SPG, it’s pretty hard to resolve with much precision anyway. Large dial calibrated instruments are another matter, and can be very accurate.

But the OP was, as I understand, talking about two digital instruments- The Cobalt being directly connected, the other mediated by through water RF transmission.

Presumably they both read zero when the tank is empty, if so the relationship is not just offset. Testing as gr8jab suggests will show where the differences lie- it’s not necessary to drain a tank to test this, though. Just pressurize the regulator, turn off the valve, and purge slightly to lower the pressure. Just a few data points will show the relationship.

-Ron

 

[KDAD]

As gr8jab and RonR mention, a multi-point comparison against another gauge that is reasonably accurate (certified would be best) is the way to go. Starting at 3000 psi and dropping every 500 psi yields seven points including a zero. That is more than enough but more is good if it does not add too much time. Maybe do 3000/2100/1200/300.

While a small difference may not matter, what if one gauge is off 300 psi across the range. That could be significant. At 3000 psi that is 10% but as you get down to 1200 psi that difference is 25%. If you are using AI, or any gauge for that matter, to track your consumption and calculate a SAC for future dive planning, well I can see where inaccurate measurements can become an issue.

As a side note, I highly suspect that my SPG that I primarily use is not linear and have been meaning to do a similar test. It will only be a simple comparison against another analog SPG that I own. I guess I can check the gauge at the fill station first to determine what it's stated pressure is for the tank that I will use and try to get some insight into its accuracy and then do the comparison as stated above.

 

[gr8jab]

Alternative (and slightly easier) experiment, if you can't get 3rd analog gauge.

Get a full tank with computer1 and computer2 both setup, valve closed. Carefully crack the valve on the tank open and take a reading. Open the valve a little more, take another reading, repeat.

You could also go backwards. Fully pressurize the system, then close the valve. Take a reading. Bleed out a little air with the regulator purge, take a reading, repeat.

Again, I would be happy to graph the data for you.

 

[gr8jab]

Just pressurize the regulator, turn off the valve, and purge slightly to lower the pressure. Just a few data points will show the relationship.
-Ron

Looks like we both had the same 'bleed' idea, and crossed posted. I didn't read your post before my 2nd post, and would have acknowledged you in it if I had.

Some of the analog 'test' gauges I've seen (and was considering when I wrote the first post) connect directly to the tank, not the HP regulator port. Bleeding off pressure can't be done with those. However, if you can add an actual SPG to an HP port, bleeding would work, and you could get the data for all three (computer 1, 2 and gauge).

 

[dmaziuk]

NOTE: It seems that most pressure transducers are non-linear, and some mathematical best-fit algorithm attempts to flatten them.

Yes, the clever mathematical trick known as different spacing between the marks on the dial.

 

[gr8jab]

Yes, the clever mathematical trick known as different spacing between the marks on the dial.

Ha. Good one! I'll bet you are a slide-rule or whiz-wheel fan! Gotta love those. I used to have one to calculate glide ratios at various speeds to optimize flight times when racing sailplanes. Old school style. As a side note, I wasn't very good at racing.

One of my colleagues (now retired) was a math guru, and loved to point out simple and elegant solutions that worked for him. He would poke fun at me as I crammed more crazy brute-force calculations into my firmware. Usually his solution was better.

But...

My post said 'transducer', which converts variations of a physical quantity into an electrical signal. I guess you could convert it back to mechanical dial indicator, and then use differential spacing. But isn't the point to enable additional calculations and data logging? Gotta linearize (is that a word) in the digital realm for that.

 

[fnfalman]

I've never had air integrated computers giving me reading differentials in the hundreds of PSI.

To date I've used four air integrated computers, of which three are wireless. They're all within 25-PSI of each other.

1. Sherwood Wisdom II non-wireless air integrated.
2. Aeris Elite T3 wireless air integrated
3. Hollis DG05 wireless air integrated
4. Aris A300 CS.

Granted that Numbers 2, 3 and 4 all used the same transmitters, so I don't know if that has anything to do with very little pressure reading differentials.

Anyway, I always plan on returning to the surface with 500-PSI or more in my HP tanks.

 

[dmaziuk]

Ha. Good one! I'll bet you are a slide-rule or whiz-wheel fan!

Actually, no: they did teach us to use slide rules in school, but that was just a hang-up from the bad old days. We had them new-fangled lictronyc calculators. (They also had a couple of those: http://www.johnwolff.id.au/calculators/pinwheel/FelixBlack-749-IMG_2308-5.jpg)

But...


My post said 'transducer', which converts variations of a physical quantity into an electrical signal.

Yeah I know. I won't let that spoil my story though.

 

[stuartv]

I would only use an AI comp as a guide
I would still keep an analogue spg since they are accurate

If they "are accurate" why do the really good ones have a way for the user to calibrate them? If the really good ones can get "off" and need to be recalibrated every so often, why do you think the ones we use on our reg sets wouldn't also get "off" over time?