How could the CCR-controller validate O2-delivery?

York

This is a spin-of from http://www.rebreatherworld.com/rebre...tml#post145487

The discussion was to add safety features to the controller(s) of the eCCR, and the point here is the measurement and feedback on oxygen-delivery.

I see two major categories of O2-problems, with sub-categories:

1. No oxygen delivered although it should
a) O2-delivery-system not pressurised (tank closed or empty)
b) hoses obstructed/dirty
c) solenoid jammed shut
d) solenoid electrically broken/disconnected

2. Oxygen delivered although it should not
a) solenoid jammed open
b) solenoid leaking
c) leak in the delivery system inside the loop

Regarding 1. I would think that the ultimate measurement system would be a electrical flow-meter on the loop side of the solenoid. I have been unable to find such a meter at a decent price, small size and small power consumption. A simple mechnical switch (operated when gas flows) might be an alternative, but would probably add more failure points.
1.a could (and already is done so in the Boris) be measured with pressure sensors on the first stage
1.b, 1.c and 1.d: I had thought about a current-measurement system. When the solenoid is opened, it initially draws a peak of current to open the solenoid, with less current required to hold it open. A jammed-shut solenoid should require more current throughout. I am not sure about the solenoid operating without pressure, I would think it depends on wether the solenoid works upstream or downstream. A broken wire or disconnected solenoid would result in no current at all.In all cases, the current-profile should be different, which a controller could measure and detect.

Re 2.a and 2.b: Same as 1, a flow measurement system would be optimal I suppose.
2.a:the current-profile of the solenoid WHEN it is operated should be different if jammed open - could this be detected in between firings? Possibly a very short fire sequence (too short to actually open the solenoid) in regular intervalls might aid in detecting a jammed-open solenoid, but that would mean high power consumption.

2.c - tough luck? design the loop so the solenoid is outside the loop?

Overall, I would think that the optimal eCCR controller would measure gauge-pressure on the first stage, and monitor the current-profile of the solenoid to detect anomalies.


Thoughts?

Joerg

David Pye

Some good ideas I think.

I'm curious to know though why change in PO2 can't be used to indicate successful delivery of O2. I know sensors have a significant lag time, but if the controller injected a whack of O2, depth hasn't changed, and PO2 has actually decreased (rather than just staying the same), then surely after say 30 seconds, then there's a fault, and you could flash lights at the diver?

Ben Field

IP sensor? Should read tank on/off, slight drop when solenoid open or freeflowing, also monitor if IP is correct?

__________________
www.hugsac.org.uk

AD_ward9

Trivially.

The O.R. units monitor tank pressure and check they go down by the right rate. It works out what is the right rate by how much they go down by when injecting gas during the pos pressure check. Tank sensors are integral: there is no O.R. unit without digital tank contents sensors. These are RF for SCUBA unit and wired for the SSUBA unit.

There is noise from temperature effects/cooling, means the cylinder can down faster than the gas is consumed and we are trying to tighten up that margin. The pressure is never allowed to go up, and never stops going down, otherwise it comes up with an alarm saying the O2 is off.

This is additional to the PPO2 monitoring, and separate from the PPO2 cell monitoring.

Alex

gtzavelas

That’s an interesting concept!
Given the fact that the O2 consumption is really small (for a 3 Lt tank we could be talking about 0,3 bat / minute – that’s around 0,15% of full scale ) can you really detect the pressure drop with enough accuracy?
What’s the accuracy of the pressure sensor?
Do you compensate for the temperature?

Does it work in practice? very interestuing to see...

Beanie

What happens when you have a small leak of similar size to solenoid operation?

__________________
Beanie
Gallery Admin & Library Assistant.
www.outlawdivers.org.uk
www.beandiving.co.uk
www.beanengineering.co.uk


0871 242 7808

PCDiver

What about pressure increases, going from cold water into warmer water? (habitat, above thermocline, ...) In Croatia we had 28°C at the surface and 13° at the bottom. I see the potential for the pressure going up due to increasing temperatures.

richc

You save battery power because the solenoid doesn't need to fire so often....

AD_ward9

Cheap digital sensors are 12 bit resolution and monotonic within 1 digit (11 available bits). That is 1/2048. This is much less than the thermal noise (see below). On the O.R. units we use a 24 bit Sigma Delta ADC (Sigma Delta is always perfectly monotonic, and also linear to 1 bit), with about 3uV of noise, so have over 18 bits of data from the electronics.

The accuracy limit is the thermal effects. These are much bigger, even just within the sensor itself (thermal compensation of the sensor is not perfect), then the tank volume changes with temperature too: we look at just the rate of change of temperature in monitoring O2 consumption, not actual temperature.

There is a 2litre O2 tank on the O.R. units. This should drop by 0.4 bar per minute minimum and 15 bar per minute maximum. The rate of change of temperature then puts a margin on this.

Alex

David Pye

I'm curious to know your thoughts on this aspect of things, Alex:

If you load things up with electronics, the more there is, the more there is to fail. If something fails, let's imagine, the pressure transducer, and generates an alarm due to an incorrect reading, is the user going to be able to choose to ignore the alarm?

If this is the case, then what is there to stop the user ignoring the alarm if there is a genuine fault condition?

David