Nitrox stick: Measuring after compressor

[estresao]

HI everyone,

I have already made my nitrox stick and I've been using it for a while and once I got used to everything my mixes are pretty accurate..... until nowadays.... In my country (Spain) a famous saying goes "if it works... don't touch it"...... but I did... I touched it.

Well, this was in the aim to get more accurate measurements to reduce mental adjustments. I've had constructed an O2 analyser using an ESP32 processor collecting values from two cells: in the future I want to mix trimix, but so far I'm measuring twice the nitrox to compare and fine tune.

I'm using a static mixer which diameter is 40mm, and I have to reduce de diameter to 20mm which is the inlet of my Bauer PE-100, which can do 100 litres per minute. To measure I'm using an used rebreather cell and a brand new Vandagraph O2 Sensor R-17VAN (which is the same my Vandagraph Tek-Ox uses so I'm expecting having very similar results)

I have not found yet a place where my cells (mainly my Vandagraph O2 Sensor R-17VAN) reads accurate values... sometimes I've got similar, but not good enough and I end up making mental calculus. My guessing is that gas flow is too much for the cell, which seems to "work with a flow rate between 100 millitres per minute to 2
litres per minute" (SIC from Vandagrash site)

I've tried in different ways:
1) mounting a 40mm PVC T after the mixer and locating the cell in the side branch.... including these variants:
1.a) Located in a PVC lid using a Flow Divertor (9711000)
1.b) same than before but with no flow divertor
1.c) attached one more PVC T and located the cell and flow divertor in the 2nd T side brand in vertical
2) Mounted 20mm PVC T after the reduction from 40mm mixer and right before the hose to the compressor.

My next tries are following options:
1) Setting a bypass from the 40mm section to the 20mm section to get some Venturi effect and locate in that bypass the O2 cell and a flow reducer
2) Same than 1 but connecting to the bypass to a Vandagraph Quick-Ox Sampler (Reference 9730210) which is the same my analyser uses


Those who made similar sticks, are you guys getting accurate values?

 

[estresao]

After running some test I still cannot get proper measurements from the cells located in my stick. I managed to add a bypass with venturi effect to control the flow crossing the O2 cell.... but even controlling that never got proper values.

I managed to use both, my home made analysed based on an ESP32 microcontroller and directly using my pocket analyser i both case using exactly the same O2 cell... while my pocket analyser provided lower values compared to when I measure directly in the tanks, the ESP32 based was even worse, so I'll get a commercial analyser to add to my stick.

I've got two questions for those who are using commercial analysers
- Are you getting proper values? (I know there would be around 0,5% because the humidity difference)
- Where and how have you installed it?
- Are you using a flow divertor? Like this one: Flow Divertor (9711000)

 

[clownfishsydney]

I could never get accurate results from the stick, so I just did some sampling of the final result in tanks and worked out a graph showing % Nitrox from various oxygen regulator psi. I can get to half a percent every time.

 

[Wookie]

What pressure are you feeding into your compressor? When I was in the nitrox making business, I sampled lp air, hp compressor suction and HP compressor discharge. The lp air was to be able to calibrate the analyzer.

We did not measure hp compressor suction when the hp compressor was running, as it was at 0 pressure or thereabouts and it’s impossible to obtain an accurate reading especially with pressure fluctuations. But if the compressor was running, take it from the compressor discharge.

 

[estresao]

What pressure are you feeding into your compressor? When I was in the nitrox making business, I sampled lp air, hp compressor suction and HP compressor discharge. The lp air was to be able to calibrate the analyzer.

We did not measure hp compressor suction when the hp compressor was running, as it was at 0 pressure or thereabouts and it’s impossible to obtain an accurate reading especially with pressure fluctuations. But if the compressor was running, take it from the compressor discharge.

My compressor is a small Bauer PE100-TW which do 100 L/m. The inlet of compressor's diameter is 20mm and the static mixer and stick inlet is 40mm.

I've tried both, setting the cell at the 40mm section and at the 20mm section and at some point was able to get some accurate reference value, I mean, not exactly the % in the tank but a reasonable value which required a little offset. After some fills you get used to that and know what value you want in the stick analyser to have the expected value in the tanks.

I know there will be always some differences due the humidity difference between my ambient air and the filtered (and dried) gas produced.

I'd like to have continuous gas sampling downstream, but my compressor only have one outlet. When filling double tanks I can sample in the other post of the manifold, but when filling single tanks I just cannot.

 

[estresao]

Still thinking....

Because the flow inside the stick is obtained by suction from the compressor, if I'm not wrong the O2 cell, wherever is located is going to be negative pressure, which is not good.....

But.... what if I take gas samples using, for example, a fish tank pump to feed a bypass where the cell is located and send back the gas to the stick?

 

[TravisD]

If you have much/any vacuum in your stick - or rather at the intake of the compressor, that's the problem that needs to be solved. It's not good for the compressor. But yeah, O2 sensors read PP02, not F02, so pressure matters, which is why you calibrate with the compressor running before adding the O2 into the stream.

 

[rjack321]

Your bypass (picture?) is probably causing more problems than solving anything. The cell can just go into the gas flow immediately downstream of the stick. The less of a "T" there is the better.

 

[tmassey]

I have a single piece of 2 inch PVC filled with aquarium bio balls. I inject the O2 at the top, then probably 4 inches of empty space, then 18 inches of bio balls, then 4 inches of empty space, then my O2 sensor. My O2 sensor uses the screw-on sampling tip: that is what is inserted into the flow.

I then have a second section just like the first section below it for helium injection.

I start my compressor, let it run for ~10 seconds, then start the calibration of my two O2 sensors. The oxygen sensors are hooked up to MiniOx 3000 analyzers. Letting the compressor run for a little bit before starting the calibration allows me to calibrate in the same airflow it will experience throughout the compression session.

My O2 flow goes through a 10 inch Dwyer flow meter. The amount of O2 I have to supply to the compressor is very consistent. I dial-in the flow at about 32.4%. Throughout the session I have to make small changes to the flow, but that’s mainly to compensate for changes in the source bottle, which causes the flow to change, probably because I have a low-end single-stage oxygen welding regulator. My adjustments are almost always to return the flow back to the same original amount, bringing me back to the same original oxygen content.

I also have post-compression analysis. I have plumbed in a 300 bar DIN female connector, to which I have hooked an old first stage. I then use an LP BC hose to connect to a low-end shop air regulator. I plumb that into a 3 L flow meter, where I set the flow to 1 L per minute. I then flow that past a third oxygen sensor, again with the oxygen flow thingy screwed onto the end of it. I also flow it through a carbon monoxide sensor at that point, ensuring that every liter of gas coming out of my compressor is carbon monoxide free.

The reading coming out of the compressor matches the number coming into my compressor, just roughly 0.4% lower (because of humidity).

I think that the plus-shaped flow thingies that you screw into the end of the oxygen sensor are essential. Without them, I too get inconsistent results. With those inserted perpendicular to the flow, I get very consistent results, pretty much no matter what. (probably why they exist and are used on medical flow devices! )

I also have found that actually doing the calibration under the conditions that the compressor will be running makes a noticeable difference as well. Doing the calibration in static air and then turning on the compressor absolutely changes the O2 percentage I see, sometimes by as much as roughly one percent.

As finicky as I’m being, it seems like I am really striving for pinpoint accuracy. I’m really not. As long as my destination is anything between 30% and 32%, I’d be perfectly happy. None of my dives on 32% are ever going to get close enough to the MOD to be relevant: if it is, I’m adding helium, anyway. But I do value consistency simply for consistency sake. I found the details matter enough that it’s worth paying attention to them.

Besides, sometimes I am mixing trimix where I really do care about precision, and I find it’s worth it to perfect my technique as best I can. But even with EAN32, I find that unless I’m being lazy I’m usually within a half a percent of what I’m targeting. And that’s with no electronic control or feedback: just Mark 1 eyeball looking up at the analyzer every few minutes or so.

Of course, I have found that the variability between different O2 sensors, different calibrations, different conditions, etc. mean I’m going to end up with variability of at least half a percent anyway.

Anyway, hopefully some of that might be useful for you to compare to your set up and see what differs and what you might be able to adjust.

 

[estresao]

Hey @rjack321, this was my first venturi bypass... later on I've added a PVC T after the flow meter to add the cell sensor.
Forget
it, it doesn't work because the pressure in the bypass is negative and the cell goes bananas. I managed to have a lot of flow in this bypass and were able to control it to set it in between 0.5 and 2 L/m specified by Vandagraph for this cell, but values were very far from accurate, the only way to get it a little closer was letting the flow go much higher... but it was very inaccurate.