REBREATHERS FOR LEISURE as opposed to military divers have been in the public arena for nearly 20 years.
In this time we have seen some development of the kit - and the deaths of some respected and experienced users of it.
Many different closed-circuit rebreathers are now manufactured around the world, and many look seductively engineered - but how safe are they to use Are some manufacturers using their customers as crash-test dummies
At the time of writing, only the APD Inspiration and Evolution, Poseidons Discovery Mk6, and VR Technologys Ouroboros and Sentinel have CE approval for the complete units, yet there are many more makes and models currently in use.
APD has for years dominated the international market, with thousands of units sold. Poseidon is hoping to create a new mass-market with its intuitive Discovery, which purports to be simple enough for diving beginners to use, and VR Technology has until now targeted hardcore technical divers, although its Sentinel is aimed squarely at the Inspiration market.
Kevin Gurr of VR Technology, technical diving guru and the man who launched a thousand close-cropped haircuts for divers, recently described the problem thus: Anyone can construct a rebreather, but can they build a life-support system
Taking this to mean that the ultimate aim is to engineer out the risk of human error when diving on a rebreather, I wanted to find out exactly what the Sentinel was able to offer, compared to other CCR units available to UK divers.
But you cant expect a manufacturer to hand a CCR over for review to someone who has not had the equipment-specific training. So who better to conduct the course and guide me through the machines intricacies than its originator

The first thing to understand is what gives rise to fatal rebreather incidents. The answer may be one or more of the following: oxygen solenoid and automatic diluent addition valve (ADV) failures; excessive oxygen volume in gas lines causing oxygen spikes; rubber component failures; electronics affected by moisture; water ingress to one display affecting the function of all electronics; failure of the electronics or its electricity supply; mushroom-valve failures in the mouthpiece resulting in bypass; and temporary floods that make a unit unusable.
How do divers contribute to incidents
They might start the dive with their unit, or their oxygen or diluent gas, turned off. Divers have died surface-swimming while breathing hypoxic diluents.
Divers fail to follow pre-dive procedures, too. They exceed canister-duration limits, and have suffered decompression illness through inability to maintain a near-constant partial pressure of oxygen (PO2).
Poor packing or incorrect scrubber assembly lead to carbon dioxide bypassing the scrubber, and subsequent poisoning.
I once suffered CO2 poisoning because I assembled the scrubber in a prototype rebreather incorrectly. I was so confused that I noticed neither the increasing rapidity of my breathing nor the onset of a headache before jumping into the water.
I can understand how many CCR divers have died from unexpected heart attacks, and I put the blame firmly at the door of CO2 poisoning.

Accident investigation and failure-mode analysis might be vital design elements, but they are seldom conducted outside the military sphere. Kevin Gurr believes that unmanned testing of rebreathers is vital.
Two of the most vital aspects of rebreather design involve scrubber-canister performance and breathing performance, or work of breathing (WOB).
Scrubber-canister duration varies widely with depth, design and gases used. High-rate test data shows that efficiency drops at an alarming rate as depth increases.
Kevin showed me the results of tests on 10 current rebreathers. Only four scrubbers were more than 60% efficient at 40m. Im pleased to say they were all British.
Contrary to popular belief, scrubber duration is affected by the mass of gas passing through, and that depends on depth. A canisters duration is gauged at the depth at which it is tested, and many CCR divers today habitually exceed the manufacturers specifications.
What Dave Shaw was thinking when he took his Mk15.5 rebreather to a depth of 270m in Boesmansgat Cave is a mystery. He paid the ultimate price for his folly.
WOB is affected by hydrostatics brought about by the relative positions of counterlungs and the divers lungs. The rebreathers resistive effect must be engineered to be as low as possible.
I know of only three manufacturers that have carried out ANSTI tests for WOB, and few rebreathers pass the CE limit in this respect. Other manufacturers seem content to test their products in their baths!

The VR Technology Sentinel is a bit of a beast, because it is so heavily engineered. Each unit approaches 37kg ready to dive, excluding any side-slung bail-out. I added 6kg of lead to that weight. In its heavy-duty transport box it weighs so much that I had to organise special terms with the very helpful Monarch Airways to get our two units out to Sharm el Sheikh.

The canister is packed with 2.2kg of scrubber material, closely following a ritual inscribed on its side to avoid mistakes. The canister is secured in place by a spring mechanism and a fixed conical rubber lip makes the seal, so there are no loose O-rings or spacers to leave out accidentally.
A properly packed canister cannot let poisonous CO2 pass.
Scrubber activity can be monitored on a temperature stick. The wrist-mounted computer counts scrubber time, too. The scrubber is good for three hours at 40m in water at 4 °C (1.6 litres co2/min).
The scrubbers base has an adjustable over-pressure valve (OPV) to take care of expanding gases in the rebreather during, for example, an ascent on off-board open-circuit bail-out.

Oxygen is automatically injected to maintain a required PO2 set-point, or a set-point set to vary point by point according to depth. Diluent is added on descent by an automatic diluent valve (ADV), but both diluent and oxygen can be added manually using two gas-block controls.
I found these quite stiff to operate, but the oxygen addition was controlled enough never to be overdone.
Oxygen is injected ahead of the scrubber so that it is well mixed and dried before reaching the three analysing cells. The cells stay dry, so rarely need recalibration. They provide read-outs to three independent destinations - the computerised wrist unit, the secondary display (with readings directly from the cells), and the head-up display (HUD) that imparts information via four differently coloured LEDs.

Once the scrubber is loaded and the diluent and O2 cylinders analysed for content and connected, the computer takes you through a long sequence of system checks, including a five-minute pre-breathe before diving. This is important, because it reveals the scrubber activating, though it seems to take forever.
The Sentinel can detect whether someone is breathing from it. Should you jump in without doing any checks or even switching the unit on, it will still take care of you, although it will chastise you via its first priority warning, unless anything more sinister occurs.
Set up the unit up to an hour before diving and the pre-dive checks take less time, but I still got caught out by delays in entering the water at the swim platform, or longer-than-expected RIB rides to the dive site, and had to suffer the machine accusing me of missing the pre-dive checks that came up as soon as I submerged.
The main display gives you the aggregated set-point for the three cells. These are so positioned within the scrubber that they needed to be calibrated only at the start of the week, and were good for at least the 10 long dives we did.
The display also shows scrubber duration by both time and by using a heat sensor to gauge activity within the absorbent; takes care of deco requirements and displays any warnings in order of priority; and tracks mandatory deco and CNS oxygen loading.
In the water, information displayed was instantly understandable.

This reads directly from the cells, either in millivolts or in PO2 readings, instantly confirming what is actually going on. If you suspect that one or more cells is reading incorrectly, its simple to navigate round the main display to deactivate as necessary. I got into the habit of comparing the PO2 on both displays every few minutes.
Both the secondary and head-up display allow you to fly the Sentinel manually, using the O2 injection button of the gas block should the main system fail.

The HUD is not only mounted in your line of sight but is also duplicated behind you to keep a buddy informed. A steady green light is good. Slow flashing green means that the PO2 is low, and fast flashing green that it is creeping up above set-point.
Green and blue lights tell you to check the main display because there is major activity on the solenoid, and flashing white indicates that you are into deco. Once you reach the deco ceiling the white light becomes steady, and it goes out when deco is completed.
A red light (which I saw only during setting up) indicates that its time to bail out to open circuit, and it comes on with an unmistakably noisy vibrator. I called it the angry frog.
It is feasible to dive by monitoring the HUD alone, and this constitutes a segment of the training. The display is quite large and can obscure the view from time to time. There were moments during my underwater photography sessions when I hinged it out of the way.

The mouthpiece has a built-in open-circuit bail-out valve (BOV) that operates when it is closed. However, as the cylinders are of only 2 litres, its advisable to carry more OC bail-out in the form of an additional sling-tank.
The BOV is merely used for a couple of sanity breaths while swapping to the off-board supply.
To close the mouthpiece and open the BOV, you rotate a knob at the front through 90°.
I found the knob very stiff, and would have liked a harder mouthpiece such as a Jax or Seacure that I could have braced in my mouth while operating it one-handed. I felt I was going to rip the soft Poseidon mouthpiece on my unit.
The hoses are very heavy-duty, so no extra ballast weights are needed. It is essential to check that the mushroom valves are working properly, and I checked mine before every dive.

Oxygen partial pressure can be set to be automatic or manual (semi-automatic). In automatic mode, it will gradually increase from 0.4 PO2 at the surface to the chosen set-point (normally 1.2 with a Sentinel) by the time you get to around 20m.
It will maintain this until you get back close to the surface, when it switches automatically to 0.4 PO2. Choose Manual and it will maintain your selected set-point.
During training we used a lower set-point to get into deco and see the HUD white LED working. Later, during a dive through a deeper part of Thomas Canyon in the Straits of Tiran, we temporarily dropped the set-point to avoid PO2 warnings.

This single bag high within the top of the unit is back-mounted and thus protected, leaving the chest uncluttered compared to, say, an Inspiration.
If you lie on your back, however, breathing may be more difficult, and underwater photographers may find this restrictive.
It is always best to run a CCR at minimum loop content, but my long lungs defied the
logic of designer Kevin, and I needed the full range of the 4.5-litre bag to be comfortable and breathe deeply.
During ascents, one has to vent the lung through the nose. This is normally OK, but there were times when I was photographing Kevin in the shallows that I lost control for a moment, and had to do a quick duck-dive to counteract expanding gas that was making me look like a frogfish that had swallowed something bigger than itself.
The over-pressure valve obviously couldnt cope at the point at which it was set. A dump valve on the counterlung would be comforting at such moments.

The Sentinel uses very fine-bore piping for the oxygen supply, so should the solenoid fail there isnt enough O2 in the pipework at any one time to cause a spike. One training exercise was to turn off the O2 cylinder and empty all the residual oxygen in the system into the loop. Only a slight rise in PO2 was evident.
The buttons on the gas block allow the user to add O2 and diluent at will. This is of practical use only to solve a problem, but is done during training. I found the buttons very hard to push, and as there is no sudden audible rush of gas,
I was often unsure whether I was pushing hard enough. This resistance is intended to prevent accidental pressing of the buttons.
Sliders on each gas block allow you to isolate the ADV or the O2 injection solenoid. There were times when I preferred to add diluent manually on descent, but again these sliders are hard to get hold of and not that easy to operate, following the same design logic.

Kevin gave me the full IANTD Sentinel-specific Level 1 course, without short-cuts. This assumes that you have no previous CCR experience.
Its sometimes hard to teach an old dog new tricks, and there were times under water when I couldnt remember what the current pretend problem was meant to be.
Similarly, I had such trouble navigating round the menus of the wrist-mounted main control, which is all about long and short pushes on either or both control buttons, that I resorted to writing a list of prompts on a slate.
It also took me a while to get used to the slight delay before the screen reacted - everything happens slowly on the Sentinel. Habitual VR3/VRX computer-users will have no such problems.
The first thing you do after submerging is to check for bubbles that might indicate a leak. Diluent flushes are an important skill to learn, and even DSMB deployment requires a slightly altered technique, as you need to switch to OC mode first.
Drills cover actions to take should the PO2 drop or rise suddenly, and maintenance of PO2 assuming a total failure of all the electronics and through simulated deco stops during an ascent.
Intentionally removing the mouthpiece under water to induce a partial flood can be rectified by draining the hoses down into the scrubber.
The water will eventually make its way down into the base of the scrubber bucket, and can be expelled by doing a Louis Armstrong impression via the over-pressure valve. The heat of the ongoing exothermic reaction vaporises and dries the water left in the scrubber material.
You also learn to use the OPV for an ascent in semi-closed mode, breathing the diluent injected manually as a simulated total failure of the oxygen supply.
We carried 10-litre cylinders of air as off-board OC bail-out, and at the start of every dive would switch to this and back under water until it became second nature. We practised handing off our bail-out cylinders to each other, too.

It is on record that I have declined to dive with another (non-CE-marked) make of rebreather in which I had no confidence. The Sentinel certainly appears to be safe to use, although it is important to be able to vent expanding gas from the counterlung quickly through the nose when you have to. My overall impression is that if you follow the instructions and pre-dive sequence without getting inventive, the only place youre likely to run into any difficulty with a Sentinel is at the airline check-in counter.
I was happy with my own performance, both in the theory and practice that amounted to
800 minutes in the water, but to be fully certified I would need to purchase a unit within six months of passing the course.
The Sentinel I used costs £5700 and Level One (with 500 minutes of in-water) training
costs around £650. For details visit

The award-winning liveaboard Typhoon, used as a dive platform for this test, carries oxygen supplies and a booster pump so that rebreather O2 cylinders can be filled to full working pressure. Two or 3-litre cylinders and scrubber material can be supplied if you book ahead,

OPEN-CIRCUIT SCUBA-DIVERS inhale their breathing gas and then exhale it out into the water.
Closed-circuit rebreather (CCR) divers continue to breathe the same gas in a circuit, but the O2 lost to metabolism is replenished as needed, and the CO2 produced as a waste product is chemically removed.
It is the partial pressure of oxygen (PO2) that is important, and not the volume.
The PO2 increases with increasing depth, so you can see that very little actual oxygen gas is needed.
Because so little O2 is used, gas supplies can be minimal for long, deep dives. A suitable diluent gas is selected to suit increasing depths.
The length of dives is governed by continuing ability to remove CO2.
Carbon dioxide breathed at anything more than, say, 5% can be debilitating or downright poisonous, so the efficacy of the scrubber and the rebreathers design to stop CO2 being re-circulated are vital in safeguarding life.
Because it is the partial pressure of oxygen that is important, as the CCR diver descends less O2 is needed. The diver must manage the mix of oxygen and diluent gas appropriate for the depth, whether manually or automatically.
If the PO2 drops below a certain level, unconsciousness and drowning will occur, but if it rises above a certain level, oxygen poisoning can be just as disastrous.
The rebreather design must take care of CO2 removal and maintenance of the appropriate PO2. A rebreather diver can always breathe, but breathing the wrong gas can be fatal.