Are rebreathers getting safer over time?

[victorzamora]

I had no idea victor dove a CCR...

I don't....and I know it shows. The reason I got involved initially was (and I'm trying to keep it this way) strictly from a controls perspective. I have a lot of experience with controls algorithms, equations, and hardware/software interface. My college career was spent in the field of UAVs where controls were the most exciting part to me.

The mixing of gas is not the problem. The problem is that if you ascend too fast you can't keep up with the drop in ppO2 due to the expansion of the loop. You could inject O2 nearly continuously and STILL not keep up with the expansion and resulting drop in ppO2. Mixing faster is not the problem, the runaway ascent is the problem. Slow down (~30ft/min) and the "problem" is solved.

Mixing faster isn't the problem on ascent, certainly, and slowing your ascent down is the easy fix. My mention of mixing taking forever isn't just the mixing, but getting the mixed gas to the sensor for it to make its reading and then make a decision. It makes the entire system less robust in that it takes forever (relatively) to get feedback. On my one rebreather dive, I had a few issues with this. One was slowing my breathing way down....I got a little flustered, my breathing rate went up, and I slowed it down. This caused the solenoid to fire a few times and jack my PO2 up above where we set it. We were shallow, so PO2 was set low and nothing dangerous happened....it's just an example of the delay causing problems. Another one was at the surface, if the PO2 was below the setpoint, it would inject O2. Then it'd wait a few seconds, and fire off again. And again and again and again. It triggered the OPV on the counterlungs before I could figure out what was wrong. Again, nothing dangerous but certainly an example of the controller being confused due to how long it takes to get a reading of the results of an input.

There are "fully automated rebreathers on the market now - and they are not "safer". Several people have perished on the Poseidon already and its a recreational fully automated CCR.

I'm actually interested in these stories, I haven't heard anything about those. Are the causes known or is everything speculation like most deaths?

If you are ascending so fast that the PO2 drops to dangerous levels you have much bigger problems than low PO2

Agreed. I was simply stating things from a controls problem. I guess that becomes human control, but it's beyond my experience. I know that I had a rough time controlling my ascent on my trial dive. Someone was nice (naive?) enough to give me 45 minutes on their Optima (Instructor trainer in a confined water environment, it wasn't just some random moron with a breather. We assembled it, he briefed me, we tried it out)....so I have a mild taste but certainly understand that that was inexperience. My point is, from a strictly control-algorithm perspective...there are things that a purely-Automatic rebreather can never do for you. There are things that require human input, and human decisions, and the ability to preempt things.

---------- Post added May 27th, 2015 at 08:30 AM ----------

'Infinitely higher' is a poor choice of words. I bet that integrating depth and the change of depth could really resolve the problem here. Gasoline cars integrate MAP and TPS sensors to determine load pretty instantaneously. Why? The 02 sensors can't be relied upon to keep up with hard accelerations or deceleration. Sounds a lot like what we have here.

For those not familiar with automotive components, MAP= Manifold Air Pressure and TPS= Throttle Positioin Sensor.

By "infinitely higher" I didn't mean how many inputs you had. I meant how much more frequently you could take samples. And of course "infinitely higher" is mathematically inaccurate, but when you're talking about millisecond readings versus 5+ second delays....the resolution difference is tremendous. Your car's cruise control can make decisions based off of .001 seconds. That's 5000 samples per one "decision cycle" of a rebreather (assuming 5 seconds between O2 solenoid firing and the appropriately mixed gas getting to the sensor).

 

[tbone1004]

I am not a CCR diver, yet I struggle to understand why you say these dives can't be done on OC? Do you not carry bailout for these dives? I know OC gas for a 5 hour dive is massive for the full run, but don't you still need the gas on OC to complete a safe dive if bailout switch is required?

yes and no. If he was conducting dives in open water he would only need the requisite gas for ascent. This is surprisingly little when you think about it. 130m takes about 15 minutes to surface straight up with about 50cf of gas. For CCR that would likely be 4al80's, one with bottom mix to get up to about 130ft where you'd switch to ean32, then another of 50/50, then potentially 2 of O2. I haven't run the deco numbers, but that should be enough for full decompression. With OC you would need all of that, plus at bare minimum 8cf/minute of bottom gas. So a set of double 130's would only get you about 20 minutes of bottom time at best, it's a monstrous dive. Now, bottom mix for that deep? VERY expensive, with CCR you can actually afford to conduct that dive and if you don't have to bail out, the dive should only cost about $100 or so in gas vs easily $1k.

 

[The Chairman]

By "infinitely higher" I didn't mean how many inputs you had. I meant how much more frequently you could take samples. And of course "infinitely higher" is mathematically inaccurate, but when you're talking about millisecond readings versus 5+ second delays....the resolution difference is tremendous. Your car's cruise control can make decisions based off of .001 seconds. That's 5000 samples per one "decision cycle" of a rebreather (assuming 5 seconds between O2 solenoid firing and the appropriately mixed gas getting to the sensor).

But that's just it. They can only achieve that millisecond response by having many more inputs. If any of those fail, the car goes into 'limp in' mode. If all you had was an oxygen sensor, the car would drive horribly and the reaction time would be in seconds, not milliseconds. Let's list the multitude of sensors on a modern day car:

Vehicle Speed Sensor
Throttle Position Sensor
Manifold Absolute Pressure Sensor
Ambient Pressure Senor
Up to three O2 sensors
Temp Sensor (engine)
Temp Sensor (ambient)
EGR Sensor
Knock Sensor
Mass Air Flow Sensor
Boost Pressure Sensor

I'm sure I've missed a few and no one car would have all of these. For instance, you would never find a MAF and a MAP on the same vehicle. All of these sensors are used by my speed control in order to decide how much throttle is needed to maintain a precise speed.

 

[descent]

[Automobiles] can only achieve that millisecond response by having many more inputs. ...

Good point, more sensors give tighter control.

I have always wondered why modern rebreather electronics don't use one or more pulse oximeters.

 

[The Chairman]

I have always wondered why modern rebreather electronics don't use one or more pulse oximeters.

Cost and effectiveness.

 

[victorzamora]

But that's just it. They can only achieve that millisecond response by having many more inputs. If any of those fail, the car goes into 'limp in' mode. If all you had was an oxygen sensor, the car would drive horribly and the reaction time would be in seconds, not milliseconds. Let's list the multitude of sensors on a modern day car:

Vehicle Speed Sensor
Throttle Position Sensor
Manifold Absolute Pressure Sensor
Ambient Pressure Senor
Up to three O2 sensors
Temp Sensor (engine)
Temp Sensor (ambient)
EGR Sensor
Knock Sensor
Mass Air Flow Sensor
Boost Pressure Sensor

I'm sure I've missed a few and no one car would have all of these. For instance, you would never find a MAF and a MAP on the same vehicle. All of these sensors are used by my speed control in order to decide how much throttle is needed to maintain a precise speed.

Good point, more sensors give tighter control.

Cost and effectiveness.

These are untrue statements. Pete: Your list of sensors is simply not a list of sensors required for Cruise Control. They're sensors for a HUGE variety of other things, and other outputs. They're providing inputs for MANY more outputs than simple cruise control. More sensors does NOT mean tighter control, at all. The only sensor you need for a good cruise control in a car is your current speed. Everything else is superfluous to cruise control, which is what we were discussing. The cruise control's decision to add speed is purely one made by the speedometer (and CC "set speed" and some programming). All of the other sensors are for the car to decide HOW to achieve that goal....downshifting to accelerate, downshifting for engine braking, more throttle, less throttle, brakes, etc. The car deciding it needs to slow down is a function of the speedo. Period. Regardless of the health of any of the other sensors, I can tell you that your speedo is reading in very tight increments and the cruise control portion of the car's brain is reading those at a very high frequency. The cruise control in your car really is just saying "go faster" and "go slower" (Delta-V vectors) with the help of a PID controller and speedo inputs.

More O2 sensors would do nothing for the scenario I painted. It would still take a ton of time (relative to computers making decisions) for the O2 solenoid to fire, O2 to mix properly, and then go through the scrubber and finally end up, fully mixed, at the O2 sensors. 1 sensor or 1000 sensors, that doesn't change. You CAN get more complicated, but I see no need. Added complications would measure total loop volume, volumetric flow rates at different points along the unit's loop, metered O2 and Dil injections, metered off-gassing, and even possibly measure CO2 absorption by the scrubber to determine metabolic rate, scrubber life, required O2 settings, etc.

Regardless, I think my controls talk is getting a little off-topic from the thread's main purpose.

 

[KevinNM]

Cost and effectiveness.

I remember a class where a instructor talked about trusting a pulse ox, and pointed out you could get 87% if you put it on a cabbage patch kid doll. "Now 87% isn't great, but hey, the doll isn't breathing either". So no, you can't trust it alone. It's useful, but it's both a trailing indicator at best and easy to confuse or dislodge.

Now everyone in EMS thinks that end-tidal CO monitoring really is a game changer, but that has proven really hard to put into rebreathers.

 

[descent]

More sensors does NOT mean tighter control, at all.

Ah, cool. Programmer cage match. :grrr: :grin:


First, a little context. One of the questions that kr2y5 (the OP) had was about why it is so difficult to be deterministic about how long the the solenoid should fire.

Injecting one gas into another seems like it should produce a seamless, homogeneous mixture immediately. In fact, the two can remain separated for an irritatingly long time. You have seen something similar if you have ever done an open water dive that crossed a thermocline, or a cave dive in an area with a halocline.

In a rebreather loop full of warm, moist sort-of-nitrox, the injected bolus of pure oxygen keeps a separate identity for a while until it is smeared around by a process called diffusion. If you have ever dripped ink into water, you have seen diffusion in action. With cold water or dense inks, diffusion takes a while. Without agitation, it might not happen at all. (I point this out for other readers; I know that Victor was born with a commercial gases account at Air Liquide)

The injected oxygen is cooler. It comes from a cool cylinder, is expanded a hundredfold by the reg, and then passes through an uninsulated rubber hose for a while. On top of that, it's dry as a bone. The bolus keeps its chilly little blob behavior for a short time, until movement around the loop tears it apart and warms up the pieces.

If a diver is working hard and breathing a high pp02 already, inhaling a fat, cool slug of pure O2 is high on the list of things to avoid.

Now to your point, and the meat of our disagreement.

Assume for a second that some mad inventor has perfected a kind of Miracle-Pipe™, with internal surfaces that can (perhaps) measure the refractive index of gases as they pass by. The pipe is lined with thousands of sensory facets. It has so many that we don't really care if several dozen are dead or insane at any given time.

A rich sensor bed like I describe would be able to form crude images of the bolus as it travels through the loop, would it not?

It would also be able to show any lingering concentrations drifting past, even when the solenoid has not fired for a while, perhaps bumped loose from some cold corner of the scrubber container.

In this case, is it better to have one sensor, or a small multitude? Which gives you a better idea of how often to fire the solenoid?

 

[The Chairman]

These are untrue statements. Pete: Your list of sensors is simply not a list of sensors required for Cruise Control.

O Rly? Superfluous? Just unplug all them thar sensors and see how your cruise control works. I didn't even add in all the sensors used for ride control, traction control and ABS. Most cars disable cruise if there are any problems with those. Of course, what do I know? I was an ASE Master Certified Auto and Truck Technician for thirty years. You're referencing cruise control made over 25 years ago where all you needed was a vacuum source and a cable on your accelerator and before cruise control was integrated into the ECM. My Sprinter is a diesel, which means the speed of the engine is governed by increasing the fuel getting to the engine. It's even more complicated than you can imagine. Rebreathers are amazingly simple compared to today's vehicles. They're not even as sophisticated as an 1981 Plymouth Horizon. That car was scary in it's own right.

Like cars, rebreathers are evolving. Evolution is best when it's slow. I bet we'll see a lot of innovation in how future machines blend and measure O2. We might even see improved measuring of CO2 as well.

 

[victorzamora]

Ah, cool. Programmer cage match. :grrr: :grin:

I'm a pretty big guy, but I have a bad back. This might be interesting!!! :grrr: :grin:

First, a little context. One of the questions that kr2y5 (the OP) had was about why it is so difficult to be deterministic about how long the the solenoid should fire.

Injecting one gas into another seems like it should produce a seamless, homogeneous mixture immediately. In fact, the two can remain separated for an irritatingly long time. You have seen something similar if you have ever done an open water dive that crossed a thermocline, or a cave dive in an area with a halocline.

In a rebreather loop full of warm, moist sort-of-nitrox, the injected bolus of pure oxygen keeps a separate identity for a while until it is smeared around by a process called diffusion. If you have ever dripped ink into water, you have seen diffusion in action. (I point this out for other readers; I know that Victor was born with a commercial gases account at Air Liquide)

With cold water or dense inks, diffusion takes a while. Without agitation, it might not happen at all.

In a rebreather, the injected oxygen comes from a cool cylinder, is expanded a hundredfold by the reg, and then passes through an uninsulated rubber hose for a while. The bolus keeps its chilly little blob behavior for a short time, until the action of respiration through the loop tears it apart and warms it up.

If a diver is working hard and breathing a high pp02 already, inhaling a fat, cool slug of pure O2 is high on the list of things to avoid.

Now to your point, and the meat of our disagreement.

Good, I really do agree to all of the above!! Well, except for my "Air Liquide" account. I don't have one, sadly, and my knowledge of fluids came through work, school, and research.

Assume for a second that some mad inventor has perfected a kind of Miracle-Pipe™, with internal surfaces that can (perhaps) measure the refractive index of gases as they pass by. The pipe is lined with thousands of sensory facets. It has so many that we don't really care if several dozen are dead or insane at any given time.

A rich sensor bed like I describe would be able to form crude images of the bolus as it travels through the loop, would it not?

It would also be able to show any lingering concentrations drifting past, even when the solenoid has not fired for a while, perhaps bumped loose from some cold corner of the scrubber container.

In this case, is it better to have one sensor, or a small multitude? Which gives you a better idea of how often to fire the solenoid?

Those sensors would do a better job, but that "Miracle-Pipe™" would be a very different type of sensor. If your solenoid were right before your sensor array (Miracle-Pipe™) then maybe it could do good enough of a job at integrating the data to properly determine the total unmixed concentration. If you placed the Miracle-Pipe™ at the same place relative to the solenoid (after scrubber, O2 injection before hand) it would take the same amount of time. However, a Miracle-Pipe™ as you described it really is a different sensor...and it would definitely reduce the time if positioned appropriately. But that's one sensor-array versus 75 R0 13 sensors. Either way, the "resolution" I was talking about was the frequency with which a decision could be made. Your Miracle-Pipe™ sounds great, and people are working on light-refraction as a sensor type. That'd be GREAT! Theoretically infinite life spans!

O Rly? Superfluous? Just unplug all them thar sensors and see how your cruise control works. I didn't even add in all the sensors used for ride control, traction control and ABS. Most cars disable cruise if there are any problems with those. Of course, what do I know? I was an ASE Master Certified Auto and Truck Technician for thirty years. You're referencing cruise control made over 25 years ago where all you needed was a vacuum source and a cable on your accelerator and before cruise control was integrated into the ECM. My Sprinter is a diesel, which means the speed of the engine is governed by increasing the fuel getting to the engine. It's even more complicated than you can imagine. Rebreathers are amazingly simple compared to today's vehicles. They're not even as sophisticated as an 1981 Plymouth Horizon. That car was scary in it's own right.

Like cars, rebreathers are evolving. Evolution is best when it's slow. I bet we'll see a lot of innovation in how future machines blend and measure O2. We might even see improved measuring of CO2 as well.

I'm sorry, Pete, you're twisting my words to win an argument I'm not engaged in. I'm not saying a car has more sensors than a rebreather, I'm saying that the other sensors are for other decisions. Your ABS sensors play no role in whether your car needs to speed up or slow down. Neither does your A/C setting, the song on the radio, or what color your shirt is. What matters is whether or not you're too slow or too fast. Your car's speed is relevant, and your required speed is relevant....not your exhaust temps.

I agree that cars need those other sensors to run properly, efficiently, and safely. I agree that cars will shut down and get grumpy if the right sensors (really, almost any one) gets pissy and stops responding (or gets unplugged). I'm sure that sensors going away would trigger some sort of "limp mode" in the car and electronically disable cruise control. However, that doesn't mean your car can't figure out it needs to speed up without MAPS....it can, it just can't do anything about it. My point is that when your car does react to the cruise control deciding it needs to implement a change, the reaction back is much quicker than on current CCR units. Your speedometer doesn't have to wait 10 seconds to see if applying brakes did anything. There's nearly instant feedback....you don't have that on a CCR.

The only sensor you need for a good cruise control in a car is your current speed. Everything else is superfluous to cruise control, which is what we were discussing.