Sign up for our monthly newsletter so you never miss the latest from InDepth!
By Walter Starck
Header image: W. Starck diving with first production run of the Electrolung on a drop-off in San Miguel, Cozumel, September 1968. The 180º Fisheye image used an optical dome port also developed by Starck in 1964. Photos courtesy of W.Starck
This article was first published in serial form on the Rebreather Email List in July 1998, at the suggestion of Dr. Peter Heseltine, a researcher in diving and hyperbaric medicine.
Development of the Electrolung came about through my chance meeting with John Kanwisher aboard Ed Link’s diving research vessel in the Bahamas in early 1968. Ed was trying out his new diver lock-out submarine Deep Diver and had invited along several researchers with relevant interests. I was there to do some deep biological collecting, and John was there to do heart rate/respiration measurements on divers using some new acoustical telemetry equipment he had developed.
Lock-out dives from Deep Diver were done using hose-fed OC Kirby Morgan helmets. Gas for this purpose and to pressurize the lock-out chamber was supplied from a large, high-pressure sphere carried by the sub. The large amount of gas required for a single dive severely limited the number of dives which could be made and involved substantial cost and logistic considerations. The need for more efficient utilization of gas was clearly apparent.
It turned out that John and I had both been considering the feasibility of a mixed gas CCR using electronic sensors to control PPO2. We both knew in general terms what was needed, but John wasn’t a diver or a machinist, and I didn’t know that much about electronics. However, I had been diving for 15 years and had built a wide range of underwater equipment. John, in addition to being a physiologist, had invented the first polarographic oxygen sensor and held a dual appointment at Woods Hole Oceanographic Institute and MIT where he lectured on electronic instrument design.
When we returned to our homes John started putting together the sensors and control circuit, and I started getting together the hardware and machining all of the necessary bits. Six weeks later we both had our respective parts together. John sent me the board and sensors; I installed them, and it worked. The overall configuration and design was basically as described, but there were, of course, numerous details to clean up. The electronics for example were wire connected on a breadboard, and the solenoid valve I had hand made and actuated with a solenoid scavenged from a battery operated coo-coo clock.
Although the prototype was put together quite quickly, it was far from a “first thing which comes to mind” effort. Quite a few years’ experience and thought had led up to it, so that when actual construction began we both knew pretty clearly what needed to be done and how to do it. Later, at Beckman, I had the opportunity of working with a whole group of specialists on improving the same device. The outcome was some tidying up of details but no fundamental improvement.
The first production units with printed circuit boards and commercial pneumatic control valves were produced in August 1968 at my company, Oceanic Equipment Company in Miami. The price was $1995 and it was the first commercially available mixed gas electronically controlled rebreather. Production and sales were continued by Oceanic Equipment for two years and was then continued with further refinement of details by Beckman Instruments in Los Angeles.
Buyers included commercial diving companies, the National Aeronautics and Space Administration (NASA), for use in their weightless simulation facility), the central Intelligence Agency (CIA) for unknown purposes, the Israeli Defense Forces, and the US Navy. At Beckman they heard that the Navy had used the Electrolung at 1000-foot depth/306 m in a lockout dive from a submarine in the Arctic, but no other details. Then, last year declassified information revealed that the Navy had developed a nuclear sub with deep lockout capabilities at that time and had used it to tap into the Russian military communication network on a submarine cable running from Vladivostok across to the Kamchatka Peninsula in the Arctic. It appears that this must have been where the Electrolung was used; but what role it may have played is unknown.
At Beckman, the biggest problem was to prevent the creation of problems which didn’t previously exist but could be introduced through changes made by specialists who were unaware of consequences outside of the narrow area of their expertise. The experience gave me a real appreciation of both the power and the limitations of specialist expertise and the importance of systems analysis in coordinating and integrating the input of specialists.
Polarographic Oxygen Sensors
The Electrolung used three polarographic oxygen sensors. The sensors were robust hand-made ones we constructed ourselves. They had a central platinum cathode about 1/4″ in diameter surrounded by a concentric silver anode about 3/8″ Diam. In between was an annular groove for the KOH electrolyte. A .001″ teflon membrane held in place by a thick silicone rubber boot retained the electrolyte. Although the sensors would run for weeks before desiccation of the electrolyte became limiting, our SOP was to make them up fresh and calibrate them for each day’s diving. At the end of the day’s diving, the membranes would be removed and the sensors washed with distilled water. Making up and washing off the sensors only took a few minutes and assured that we always had fresh sensors.
Sensors of this type don’t wear out, so they are hardwired into the circuit. Unlike galvanic sensors, they don’t use oxygen but rather just respond to its presence. They work equally well submerged, so the effect of any condensation is negligible. A drop of water fully covering the end of the sensor would only slow the response time. In practice, we never had any condensation in the sensor area ,as this came immediately after the canister, so the gas was at its warmest and driest point in the circuit. Thick plastic walls probably also helped in avoiding condensation on cold surfaces.
The chief advantages of the sensors were that they were always fresh, and condensation wasn’t a worry. The disadvantage is that in making them up with fresh electrolyte we could screw up by contaminating the sensor via sloppy technique. Any significant change in calibration after a fresh makeup would be an indicator, and determining why should be mandatory before proceeding further. Still, there are the black box mentalities who will simply crank the trim pots until they get the reading they want and then assume all is well.
There were two trim pots for calibrating each sensor. One for zero. The other for gain. Zero was checked each time before the sensors were made up. Gain was calibrated initially with air, and then the unit was put together, and a check with pure O2 was done. The permeability of teflon to O2 varies with temperature. The sensors were of potted epoxy construction with the electrodes embedded in the epoxy. A thermistor in contact with the underside of the cathode was also embedded. This thermistor had a similar response curve to the teflon and compensated for the temperature effect keeping output linear over the desired range.
Our chosen set point was 0.5 atm partial pressure of oxygen (PPO2).
The bottom line was that with proper care the sensors were very reliable. Enough so, that they could be hard wired in, and I know of no case where one ever had to be replaced.
The Electronics: Control, Readouts, Alarm
Unlike galvanic sensors, polarographic electrodes don’t generate electricity. Rather, the conductivity of the cell varies in the presence of oxygen. A bias potential from an external source is applied between anode and cathode, and the resulting flow of current is a function of the molecular concentration of oxygen present. The current involved is very small, so an Op Amp is used with each sensor to boost power to a level useful for control and monitoring. Hermetically sealed trim pots which incorporate an O-ring seal around the adjustment screw provide for zero and gain adjustment of each Op Amp, thus enabling calibration.
The amplified signal is read out to a wrist display consisting of a stack of three edgewise panel meters. We used 100 microamp meters in conjunction with high resistance to prevent a possible short in this circuit from affecting the solenoid control. Mil Spec, so-called “shock resistant” meters were used. These resist minor bumps but still they won’t stand up if you drop the display on a steel deck or concrete. In practice it wasn’t a significant problem, but occasionally a meter did require replacement. This was quick and easy to do.
The big advantage of this type of analog display is that you can tell at a glance everything you need to know. In use, all you need to verify is that all of the readouts are in line with one another and at or a bit above the set point which was exactly mid-scale. This kind of meter is also precise enough for calibration purposes. If I were doing it today, I would look at bar type LCD or LED readouts for monitoring and perhaps a separate switchable numeric display for calibration. I would also seriously consider a miniature heads-up display in the mask instead of one on the wrist. I don’t like numeric displays for monitoring, as they entail reading and mentally comparing numbers which requires much more attention than just noticing if position and alignment are where they should be. Possibly some of the commercial RBs have already done all this.
The amplified signals from all three sensors were fed into a fourth Op Amp which in effect averaged them and used the resulting value to control the solenoid set point via a switching transistor. We used a fixed set point of 0.5 atm PPO2, but it would be simple to add a trim pot to provide an adjustable set point. Clipping circuits limited the input to the control Op Amp from each sensor to values corresponding to 0.25 and 0.75 atm PPO2. If any one sensor began to read drastically different from the others, its effect on automatic solenoid control was thus limited. Clipping came after the meter display, thus they would continue to read true output even if the input to the control Op Amp was clipped. Clipping also activated an audible alarm. If the alarm sounded, a glance at the meters would tell you what the situation was. If only one was off, the other two would continue to exercise control. If all were high, low, or different from one another, you could use manual control while aborting the dive.
The Op Amps require a + and a – voltage power supply. This was supplied by a pair of 9V Manganese Alkaline transistor radio batteries. Bias to the sensors was provided from the same source via a voltage dividing resistor circuit. A second pair of the same batteries provided switchable backup power. A third pair used in parallel provided separate power for the solenoid. The solenoid did not have backup, as this is non-critical because manual control of O2 is easily affected. The snap terminals used for this type of battery were securely attached to a bulkhead. A screw-adjusted base plate held the batteries firmly in place, and against the terminals, avoiding any possibility of a loose battery connection.
All the electronics were incorporated on a single circuit board about 4”x5″/10-13 cm. This was mounted on one side of a longitudinal bulkhead in the electronics housing with the batteries and audible alarm on the other. This longitudinal bulkhead was itself mounted on a transverse bulkhead which separated the electronics compartment from a plenum above the absorbent canister. The solenoid and sensors were mounted on the opposite side of this transverse bulkhead, thus everything electrical other than the wrist display was immediately adjacent to each other.
In the units I made, all of the electronic components were on a printed circuit board. After assembly, the boards were coated with a spray-on waterproofing compound as is widely used for marine electronics. At Beckman, the components were assembled into 4 micro-welded, epoxy potted modules which plugged into gold plated sockets on the circuit board. In theory, this is a better way to go, but in practice, it didn’t make any noticeable difference.
With respect to the reliability of electronics in this kind of application, recently someone posed the question, “When was the last time your TV failed?” to which Robert made the wonderful reply, “The last time I took the bastard underwater.” Both comments reflect important points. Electronics in themselves can be extremely reliable. In terms of MTBF, they are far more reliable than most mechanical devices. Enough so that they can be trusted for things like passenger aircraft control systems where thousands of systems are in everyday use and a single failure means the loss of hundreds of lives. But Robert is right too. If you flood them with water, they fail.
The problem then is really a mechanical one. Can electronics be reliably enclosed so as to prevent flooding in underwater use? If it were solely a matter of constructing a watertight pressure-proof housing for the electronics, that alone wouldn’t be too hard. Unfortunately there is also the matter of connections for sensors, displays, a solenoid, and a switch, plus keeping all these external devices themselves dry. The possibilities for leaks begin to multiply. With a great deal of care in construction and use, high reliability is achievable, but I think there is a much easier way to reliably keep out the water.
The key to the solution is pressure. Keeping things watertight under 100-200 psi is difficult. Doing it under 0.5-1 psi is easy. In the Electrolung everything was at ambient pressure. The electronics compartment was vented via a small canister of silica gel with the rest of the system. A standpipe for the vent orifice prevented any accumulated moisture in the canister plenum from being pushed into the electronics compartment. In anything but a head-down position, the electronics were above the counterlung; thus, any leak would normally result in gas escaping rather than water coming in. In practice, with the kinds of seals involved and the very low pressure differentials, leakage anywhere in the electronics section was never a problem.
Humidity and condensation were also not problematic. Plastic construction probably helped in avoiding the latter, and the waterproof coating seemed to be quite sufficient for the former, as is well attested by a wide array of complex devices and vast usage experience in the marine electronics industry.
The only practical way to get your TV underwater with this type of system is to flood the entire system. This is inherently no more likely nor any more or less disastrous than it would be with any other rebreather, regardless of type.
Two, 9 cf at 2100 psi (1.75 ltr @ 145 bar) steel gas cylinders were used for O2 and inert gas. These were lightweight models FAA certified for use in aircraft. Initially we used chrome plating to protect them; later we went to teflon coating. Beckman liked the more military look, and there was some concern over possible hydrogen embrittlement from the chroming process.
Standard, old style, “K” valves were used as cylinder valves. On the inert side, a SCUBA regulator-type yoke was used to mount a high pressure 1/8″ NPT needle valve operated by rotary action of a T-shaped handle. Inert gas was valved in manually directly from the tank as needed using this valve. In use, it had a very smooth precise feel. Inert gas was valved into the plenum at the bottom of the absorbent canister so that some mixing would take place before it got to the sensors.
On the O2 side, a piston-type first stage of a U.S. Diver’s single hose regulator was used to reduce tank pressure to about 60 psi. This is somewhat lower than such first stages normally delivered and was achieved by using a weaker piston spring. The normal hose to the second stage was used to connect the O2 supply to the solenoid valve. The octopus port of the first stage was used to attach an O2 bypass valve. This was a spring action, lever activated low pressure valve and it was protected by an enclosure which required opening a spring closed cover to get at the valve. The manual bypass valved O2 directly into the sensor compartment, so the result was immediately readable.
I will digress briefly on O2. In addition to the physiological risks recently discussed in some detail on the list, there is also the danger of fire and explosion. Valves, regulators, fittings, and any other equipment used for O2 have to be thoroughly degreased of any petroleum based lubricants. If lubrication is required, as for example with the o-ring seal of a regulator piston, non-combustible silicone based lubricants must be used. Be aware that even a fingerprint oily with suntan lotion can start an explosive fire with O2. Once an O2 fire starts, all sorts of things you might not ordinarily think of as combustible burn ferociously. I have heard stories of chamber fires in which everything inside, including the occupants, was reduced to ash.
My partner Kanwisher was on one of the advisory panels to National Aeronautics and Space Administration (NASA) in connection with the Apollo program. Although he recommended using a mixed gas atmosphere in the Apollo capsule, he was overridden by the engineers who felt that monitoring the PPO2 was too difficult. John knew better, as he had been doing it for several years in conjunction with his work on respiration, but the engineers prevailed. The result was the fire that killed three astronauts.
The solenoid valve we used was a miniature 12-volt one made for pneumatic control. We equipped it with a miniature screw-adjusted needle valve outlet. When the setpoint is reached and the solenoid is triggered, it takes perhaps three or four seconds for the sensors to respond and rise enough to cut it off again. The solenoid needle valve was adjusted so that the O2 injected raised the PPO2 to a peak pulse of about 0.75 atm and would usually trigger a couple of beeps from the audible alarm. Within a couple of breaths, mixing brought the level back to perhaps 0.65, after which it dropped more slowly, as it was consumed by metabolism until the setpoint was reached again after about a minute or so. That would be for moderate activity such as easy swimming. At complete rest, it would of course take longer to drop back to the set point and less time if you were actively swimming.
If the needle valve was adjusted to a lower flow rate, solenoid activation would be more frequent and of longer duration, placing an unnecessary drain on the solenoid batteries. If much higher flow was adjusted for the O2 spikes would be too high, and the alarm would be sounding much of the time. I think there are now smaller, more power-efficient solenoid valves available.
The solenoid and manual bypass valves were of the downstream type so that if high pressure leakage from the regulator occurred, it would release when it reached the level where it overcame the spring tension which normally closed the valve. This is important to prevent either valve lockup or blowing out the supply hose in the event of a high-3wqaz¸pressure leak. In the event of O2 leakage from either valve, the cylinder valve could be used to cut it off.
The gas cylinders were mounted on either side of the central larger cylinder containing the absorbent canister and electronics section. This assembly was worn as a backpack with the valves at the bottom at hip level. Inert gas had to be added several times on descent and at other times if you lost any from nasal exhalation. Manual O2 was normally only used in decompression. The inert gas valve was therefore on the diver’s left side leaving the right hand free for more complex tasks. Swapping sides for southpaws would have been easy, but I don’t recall anyone ever raising the question. It was no big thing either way.
The Breathing Circuit
The Electrolung mouthpiece was originally constructed of PVC. The main body of the mouthpiece assembly was made in three parts glued with PVC solvent cement. At Beckman, this part was manufactured as a teflon-coated aluminum investment casting which made a nicer looking part. A rotating drum-type valve operated by a small lever permitted closure of the mouthpiece if you wanted to take it out of the mouth underwater. Two check valves directed gas flow to the proper hoses for inhale and exhale cycles.
Three breathing hoses were attached to the mouthpiece. One went straight down to the counterlung which was worn on the chest. The other two were routed to either side over the shoulders to the inlet and outlet of the absorbent canister/electronics section which was worn on the back. Initially, hoses from the old style twin hose scuba regulators were used. At Beckman we found a manufacturer in L.A. who made high quality fiber reinforced hoses for the O2 breathing systems in military aircraft. The price was only a little more than the scuba hose, and they could be made to order in small quantities in various lengths. Scuba hoses tended to start to leak after a year or two. The others are still usable after a quarter century in tropical conditions.
Hose clamps may seem mundane but are worth considering. A hose coming off with scuba is an inconvenience. With an RB, it is a life-threatening disaster. Initially, we used the spring steel ratchet-type clamps used on scuba regulators, but I didn’t fully trust them, and they rusted. We tried chrome plating them and found they would then break unexpectedly due to hydrogen embrittlement. In the end, we went to good quality heavy duty SS worm screw type clamps. Some SS clamps have ordinary mild steel worm screws which rust, electrolyze, and break after a while in marine use. Lightweight SS clamps can also break for no apparent reason due to stress fatigue. SS is prone to this. SS fitting used in sailboat rigging are notorious for letting go at the most inopportune times.
A lot of people seem to now use the nylon ratchet type ties for clamping hoses, but I wouldn’t trust these either for critical applications. Nylon is subject to cold flow under stress, and after a while they become looser. Using them as a backup next to a good SS clamp however, might not be a bad idea.
The counterlung was a bag made of two pieces of plastic material electronically welded together over a 1/4″ surface around the entire perimeter about 1/2″ in from the edge which was then sewn together and wrapped in a heavy edging tape. Originally, I used clear vinyl for the material. This worked well enough, but later I found a lighter, more flexible, translucent fiber reinforced plastic material which was better.
Being clear or translucent offers two advantages. It lets light in and so reduces the growth of microorganisms. It also makes it easy to see if any water has accumulated and get rid of it before it gets to a level where it might be drawn into the absorbent canister. Places which make awnings, boat covers, etc. or those who make plastic zipper bags, folders, and the like can easily and inexpensively do this kind of work. All you need is a paper pattern of what you want. One-offs for prototyping are not expensive and with a $100 or so for a die for the welding, they can pop out small quantity runs dirt cheap. Usually they also have samples and catologues of all sorts of material to choose from. Teflon coated nylon is now available and might be very good for this application. I mention all this because some homebuilders may be interested.
The counterlung had a single hose attached about 1/3 of the way up from the bottom on the front (away from the body) side. A drain plug was at the bottom near the lower left corner. At Beckman we added an overpressure relief valve near the top. The fittings all used a flange and threaded collar-type attachment similar to a kitchen sink drain. The flange incorporated a groove and o-ring in its face which ensured a firm grip and seal with the counterlung material. The fittings were machined from PVC, except for the small drain fitting, which was SS or chromed brass with a 1/4″ plug on a short lanyard. It was basically the same as a control gland used in underwater camera housings.
The overpressure relief valve released at somewhere around 0.75 psi. Its practical use was only the prevention of possible counterlung rupture if gas was accidentally valved in with the mouthpiece shut-off valve closed. It was introduced at the suggestion of experienced OC divers who, not being used to getting rid of excess gas via the nose during ascent, tried to exhale against a full counterlung and couldn’t.
The counterlung was attached by grommets at each corner which mated with twist studs mounted on the shoulder straps at the top and on short adjustable straps paralleling the backpack waist strap on each side at the bottom. The counterlung volume we used was about 4 L.
There has been some discussion on the list recently re: the relative merits of chest mounted (resistance on exhale) vs. back mounted (resistance on inhale) counterlungs. In an earlier post, which has been quoted in the recent discussion, I opted for chest mounting as preferable because the mechanics of breathing musculature is such that the power available for exhalation is greater than that for inhalation. The counter argument is that resistance on exhale reduces the volume of exhaled gas leading to CO2 retention.
First, we need to keep in mind all this is somewhat hypothetical, and in the real world both configurations have been used successfully. With sustained high level exertion where any advantage might be important (and in the continuing absence of any proper comparative testing) I would opt for the chest mount for two reasons: It’s less tiring to put a bit of extra effort into exhaling than it is into inhaling, and the bottom line at the extreme for ventilation lies with how much gas you can move in and out in a given time. Given equal resistance in either direction, the more powerful exhalation cycle will move the greater amount of gas. In the end, over a few breaths, expiration and inspiration must be equal. Restriction of either sets the limit, so if there has to be a restriction I would rather it be on the side which can best handle it.
We sold a number of Electrolungs to commando-type users. One of their prime concerns was breathing resistance in sustained hard swimming. After trying it, all gave it their thumbs up in this respect. In out-of-water chamber tests at 1000 fsw/306 msw pressure, breathing resistance during exercise was encountered. This was in the breathing circuit itself and could of course be relieved somewhat by bigger hoses, larger absorbent bed cross section, etc., but as the market for that capability was effectively nil, it was never pursued.
One memorable experience with the commando types took place in the Bahamas. A British Royal Marine Commando attached to the Canadian Navy flew down to join me on my vessel and try out the Electrolung. He was a big, bullet-headed guy, built like a fridge with a head. After a couple days’ instruction diving, he wanted to do a long, hard swim with it, and as there was no one else to do it with him, I ended up going along. He went for a couple of miles virtually, flat out. Luckily the water was crystal clear, so I managed to at least keep him in sight. When we came to the surface, the bastard wasn’t even winded. He was satisfied it could do the job and just wanted a smoke. It was flat calm, and the water was only about 30 feet deep, so I had a skiff following us with his smokes and we could ride back.
Although I still don’t know for absolute certainty whether chest or back mount is optimal, I do know that chest mount is good enough. What I really do like about it is that it is easy to see if there is any water in it and easy to pull the plug, squeeze the bag, and expel it.
We did have a couple of experienced open circuit (OC) divers, new to the Electrolung, let water leak in around their mouths until it was gurgling away with each breath. They continued until they had largely flooded the absorbent canister and eventually got a mouthful of absorbent cocktail. They were quite irate about all this and swore it was the fault of the Electrolung. This kind of thing is a recurrent problem with rebreathers (RBs). Experienced OC divers have habits which don’t go with RBs. They also tend to think of themselves as expert divers rather than as novice RB users. As a result, they often don’t really listen and don’t take advice well, and they tend to blame the device if anything is not right, rather than realizing that they have to learn to use it right.
The actual breathing circuit for the Electrolung was: Exhale directly to counterlung via bottom mouthpiece hose. Inhale draws gas from counterlung back out the same hose into the left mouthpiece hose then to the bottom of the canister via a central tube inside it. At the bottom, the gas emerges into a plenum which distributes it over the inlet surface of the absorbent column. After passing through the absorbent, it emerges at the top into the chamber where the sensors and solenoid are located. From here, it continues via the right mouthpiece hose into the mouthpiece itself.
The Absorbent Canister/Electronics Housing
In the Electrolung the CO2 absorbent and electronics were housed in a transparent acrylic (Plexiglas) cylinder which, together with the gas bottles, was worn as a backpack. This cylinder was 24″ long x 4″ I.D. with 1/4″ thick walls. It was divided into two pieces. An 18″ section contained the absorbent bed with a 4″ space at the top which accommodated the sensors and O2 solenoid. Above this, a further 6″ section separated by a bulkhead contained the electronics. A small knob on the electronics section operated a double throw, center off switch controlling power to the electronics via either set of batteries. This was reachable behind the head if development of your biceps didn’t prevent it. Actually, geeks do make the best rebreather divers anyway, but out of respect for the temporary cease fire, I will refrain from further comment along that line.
The ends were sealed by 1/2″ thick PVC O-ring sealed plug-type closures. A thick PVC double plug-type bulkhead joined and partitioned off the electronics section from the absorbent section. The entire assembly was held rigidly together by a central 1/4 SS tie rod running from top to bottom. It had a large external knob at the bottom for tightening and loosening which was affected by screwing into a metal socket on the top closure. O-ring seals were used to seal the tie rod penetration of the bottom closure, the metal tie rod socket on the top closure and where the tie rod passed through the electronics bulkhead. This last was only to prevent capillary action from possibly drawing any water from the absorbent section into the electronics section. The electronics section atmosphere was vented to the absorbent section through a small canister of silica gel via a separate bulkhead penetration and a small standpipe as mentioned earlier in the description of the electronics.
The O-ring seals for the end closures and the join at the electronics bulkhead were all radial type seals which automatically affect a proper seal when they are plugged into the cylinder. Sealing is effectively independent of how tightly or loosely things are clamped together. With this type of seal and operating at ambient pressure, the possibility of leakage around a seal is vanishingly small.
A brief aside for homebuilders: Although O-rings are marvellously effective seals and are universally used in all types of underwater equipment, it is remarkable how often manufacturers use them improperly. O-ring suppliers have various free pamphlets and data sheets on proper application of O-rings which includes data on the correct shape and tolerances for the grooves which accommodate them. It is well worthwhile to avail yourself of this information.
Common errors in O-ring usage often seen in marine equipment are: Grooves too deep, resulting in inadequate sealing pressure. Too shallow, resulting in too much compression of the seal, leading eventually to fine, radial cracking of the O-ring itself and consequent leaking. Too narrow, which interferes with proper compression and sealing at low compression and distortion toward a square cross section under full compression, this leading again to radial cracking of the seal. Finally, and most ignorant of all, is the use of rounded U-shaped grooves which defeats the whole principle and advantages of the circular cross section and turns it effectively into a flat gasket.
The absorbent canister portion of the cylinder was a 12″ section toward the bottom defined by two 1/4″ thick acrylic or PVC internal bulkheads perforated with an array of holes. Plastic screen was used to keep the absorbent from falling through the holes. The top bulkhead was fixed in place. The bottom one was free to move but held in place against the absorbent column by a large spring. This served to keep the absorbent compacted without channelling despite any minor settling of the granules after filling the canister.
We used Baralyme as an absorbent. This is a National Cylinder Gas trade name for Barium Hydroxide. It was widely used in hospitals and came in hermetically sealed one quart cartons of the type used for milk. The Electrolung canister held two cartons which would be sufficient for six hours of moderate activity. We changed them after four hours. Baralyme came with a color indicator—pink when fresh, blue when expired. It is less caustic than soda lime and worked well for us.
As described earlier, gas was drawn from the counterlung to the space at the bottom of the canister down a central 1″ I.D. tube leading from the inlet hose attachment down to the bottom end of the canister section. From there it passed back up through the absorbent into the sensor/solenoid chamber and on via the inhale hose to the mouthpiece.
The gas supply cylinders were mounted on either side of the absorbent canister/electronics cylinder using spacer blocks conforming to the curvature of the respective cylinders. The three cylinders were secured rigidly in place by two large SS hose clamps. One was adequate for the purpose, so there was backup in the event of one breaking. The spacer blocks also served as attachment points for the harness, which was a U.S. Divers, one of the types widely used before B.C.’s took over this function. There was a wide vinyl strap for each shoulder plus a waist strap. The overall configuration of the Electrolung backpack was similar in many respects to that of the small triple tank OC rigs favored by the French at that time. It rode well on the back and was quite comfortable. All up, the weight of the Electrolung was about 30 pounds/14 kg.
Some people expressed concern about the use of acrylic, fearing the possibility of breakage. This is one of those things which is more apparent than real. In this case, it is protected on one face by the wearer’s body, on either side by steel gas cylinders, and at top and bottom by thick PVC ends. The only real exposure to any possible impact was the curved surface of a 4 1/2′ O.D. ¼”- thick cylinder, which would be extremely hard to break. We did look at using polycarbonate (Lexan) which is literally bullet proof but found it crazed and crumbled into small pieces when exposed to hydroxides.
It would of course be easy to make the whole thing out of PVC, but I feel the advantage in being able to see condensation, water, and the condition of your absorbent more than outweighs the non-problem of smashing heavily into things while going backwards. Beckman offered a fiberglass fairing for those who might be concerned with this, and of course some then bought it because they liked the way it looked and others did so because they felt that if it was offered they probably should get the complete setup.
A final aside for homebuilders: The Electrolung was really a homebuilt which became a commercial product. It was built entirely with a drill press, lathe, and jigsaw, plus a bench grinder for shaping and sharpening lathe tools as the only power tools. For anyone attempting to build any kind of underwater equipment, a metal lathe is really a must. You can easily make all kinds of cylindrical housings, O-ring sealed fittings, ports and closures, and any kind of threaded fitting you might need with one.
A small lathe with a five or six inch swing over the bed will enable you to make housings and ports up to 10-12″ in diameter. Good quality, Chinese-made lathes suitable for this kind of work are now available for about US$1500. For another few hundred dollars, you can also get a milling attachment as well, which is a useful addition. Teaching yourself how to use it is not hard. Good textbooks which cover this kind of machine tool work are readily available and easy to follow.
Here is the Electrolung patent. It contains much more detail including various drawings.
Reflections and Speculations
Although development of the Electrolung was interesting, even exciting, in itself it was just an interesting incident in a bigger, far more interesting, and significant picture. Like most historical events, I suppose, what was happening to the participants at the time didn’t appear to be so remarkable as it later does with the broader perspective of hindsight. The larger perspective on what is taking place right now tends to be somewhat obscured by the ordinary events of living. Except for rare instances, whatever we are doing, however interesting and exciting it may be, tends to still feel like life, not like history in the making.
In retrospect however, I have come to realize that from the mid 1950s through the mid 70s, something really remarkable was taking place in diving. During that period, diving grew from the obsession of a small group of generally impecunious young people mostly in Florida, California, France and Italy to a global industry catering to well-to-do hobbyists. Remote tropical islands all over the world began to sprout airports and dive operations and diving became strongly oriented to travel to exotic locations. Though all this was in itself remarkable, something truly unique was at the real heart of what was happening.
For the first time in history, humans could freely enter, explore and personally experience the oldest, richest, most beautiful and exotic communities in nature, tropical coral reefs. Coral reefs are truly remarkable places. Nowhere else can one experience such an abundance and diversity of life. Nowhere else is it so colorful, exotic and so easily experienced at close range.
Diving on a reef is like a trip in a time machine to a time before humans existed and nature ruled in primeval pristine abundance. Fossils of many reef creatures from as much as 60 million years ago are essentially the same as those on reefs now. In fact, some Pacific reefs have existed as reefs for that period of time.
For a biologist, being among the first to dive on reefs was a most extraordinary experience. In a way, it was a bit like landing on another planet. On nearly every dive, we were going where no human had ever been before. The discovery of phenomena of life, as well as strange and beautiful creatures whose existence we never even suspected was an everyday occurrence. At the time this kind of experience was so commonplace, tropic seas so vast and remote, and so few people were doing it, that it began to seem as if this was just the way things were, and this kind of experience would continue on indefinitely.
Already, however, this era has become history. Although there are vast amounts still to be discovered about the details and inner workings of reefs, still undiscovered species are getting harder and harder to find and remote locations are becoming less and less remote. The experience of being among the first to explore the richest realm of nature has come and gone, not to be repeated.
On reefs, one niche still remains. Actually it is a really big one. The zone below 200 ft/61 m, down to the deepest limits of what you might call part of the reef community at about 600 ft/184m, is still largely unexplored. Although it is not as rich in life as the shallower regions, it is still incredibly rich in life and is an area about which we know very little.
As far as I am aware (in 1998), the only person on the planet regularly exploring this zone is icythologist Richard Pyle. What he is doing is really exceptional, and he is doing it essentially on his own. While discussion on the Rebreather List is largely restricted to the technology and physiology as an end in itself, what Rich is doing goes well beyond this. As well as making more deep free dives than anyone ever has before, he is coming back with knowledge and specimens from every dive. What he is doing is a permanent contribution to knowledge which will stand long after any of today’s diving records are broken and forgotten. I have never met him personally, and am commenting only out of recognition of something exceptional.
Over the past 25 or 30 years advances in diving technology have been almost entirely small and incremental. The only real exception I can think of is the development of dive computers. It appears we are up against the realities of human physiology. With every increase in depth and bottom time, the cost, complexity, effort, and risk increases exponentially while the return of useful achievement remains more or less linearly related to bottom time.
The future, it seems, lies in other directions—especially robotics. Here the advances have been impressive, and future development promises to become even more so. Already we are at a point where more and more functions which previously required a diver can be effectively carried out by ROV’s. It is not hard to foresee that in a few years most of what we do at great effort and risk by diving can and will be done by nerds at consoles. In fact, right now in the Gulf of Mexico, Shell and BP are drilling in 5000-6000 fsw/1532-1839m and all work at the wellhead is done by ROV’S.
If you find this kind of scenario uncomfortable, don’t let it worry you. Long term prediction, no matter how well reasoned and seemingly inescapable, has a way of almost always being wrong. So much so that I have often wondered if beneath the facade of Newtonian certainty of our universe, somewhere in the iffy probabilistic realm of quantum mechanics, there is not a relationship which dictates that the very act of prediction sets in motion forces which generate a different outcome. So, if you don’t agree with my prediction, the good news is that I may well have avoided it by predicting it.
Fortunately, the real outcome is usually even more interesting than any of the predictions.
How did Starck and his colleagues manage their Electrolung decompression? Stay tuned for Starck’s follow-up story: “Confronting the Unknowns of Decompressing on The First Electronic Rebreather” in a coming issue of InDepth.
Alert Diver (2017): Oxygen Sensing in Rebreather Diving by Michael Menduno
Popular Mechanics (July 1965): Half A Mile Down With Scuba by C.P. Gilmore
Reprinted with permission from the Historical Diving Society USA, The Journal of Diving History Volume 23 #85 4Q 2015.
To become a member and access our treasure trove of dive history, visit us at: www.HDS.org. We are also on Facebook: Historical Diving Society USA
Download aquaCORPS Journal #7 C2 (DEC93): aquaCORPS’ rebreather issue See “Electrolung,” by Walter Stark p. 6 & 8. Proudly sponsored by Divesoft (Original magazine with two Divesoft inserts).
Walter Starck is one of the pioneers in the scientific investigation of coral reefs. He grew up in the Florida Keys and received a PhD in marine science from the University of Miami in 1964. Since 1978, his home has been in north Queensland, Australia. Throughout his career in marine biology, participating in expeditions around the world, Dr. Starck has been extensively involved with development of the technology required to facilitate his activities. In several instances patented inventions and commercial products have resulted. In addition to the optical dome port and the Electrolung other noteworthy achievements in this area have been: The Bang stick, a hermetically sealed underwater firearm for hunting and defense. Underwater housings for numerous cameras and instruments. Underwater lighting systems. A multipurpose commercial waterproof electrical connector. Design of the unique research vessel El Torito, a 9 meter high-speed diving launch, a 24 passenger eco-tourism vessel, and the Oceanic 8000 Longboat. The longboat was a long narrow high efficiency powerboat inspired by the efficiency of the log canoes of the Solomon Islands. He has also built and flown an amphibious aircraft of advanced canard wing design using high technology composite materials. Recently (Aug 2017) he was senior author on an extensive update on the Alligator Reef study that brought the total species list for that locality up to 618 species.
Dr. Starck has authored over 100 articles and books, which include numerous technical and peer reviewed scientific studies as well as many articles in leading popular publications. His photography has been widely published in conjunction with his writing, and he has produced nearly 20 films and videos. Throughout his extensive career, he has managed to inspire not only admiration, but also the ire of some detractors who have taken umbrage at his efforts to inject what he believes to be “a rational perspective on human ecology into the eco-mania that has become epidemic in our struggling Western economies.” His criticisms of the “poor science and blatantly false claims widely used to support various environmental agendas” have earned him some criticism.
What Happened to Solid State Oxygen Sensors?
The news in 2016 that Poseidon Diving Systems would be incorporating a solid state oxygen sensor in their rebreathers sent a buzz through the rebreather community. Galvanic sensors, along with their legacy-1960s “voting logic” algorithms to boost reliability, had long been considered the weakest link in closed circuit rebreathers. Many heralded Poseidon’s subsequent 2017 roll-out as the dawn of a new era in rebreather safety. Five years later, Poseidon remains one of two companies (the other strictly military) to have adopted optical sensors. Technology reporter and tech diver Ashley Stewart examines some of the reasons why.
by Ashley Stewart.
Header image: Karst Underwater Research (KUR) rebreather divers at Weeki Wachee. Photo by Kirill Egorov
For years, it’s been said there’s a revolution coming for the closed-circuit rebreather— a new, more reliable, safer replacement for the traditional electro-galvanic oxygen sensor, widely considered the weakest component of rebreathers. In March 2017, that revolution looked to be just over the horizon. Poseidon Diving Systems began shipping an offboard solid state sensor to supplement the MKVI’s and SE7EN’s galvanic sensors and offered to license the technology to other manufacturers. Though Poseidon subsequently incorporated the solid state sensor into its SE7EN rebreathers, nearly five years have passed, and not much else has changed.
Poseidon remains the only manufacturer using solid state sensors in recreational rebreathers. No other companies have licensed Poseidon’s technology. Major tech diving manufacturers—including JJ-CCR and Divesoft—say they don’t believe the technology in general is ready for use in rebreathers. Some manufacturers worry that the sensors won’t function accurately in humid environments over a wide range of pressures, and they claim that addressing these challenges will be costly. Meanwhile, divers who tested Poseidon’s sensors offered mixed reviews, and even the inventor who sold the sensor validation technology patent to Poseidon believes they should be used along with traditional sensors. (Poseidon gives divers the option of combining the sensors).
Oxygen sensors are the enabling technology that made mixed gas rebreathers possible, replacing rebreathers that could only be used with pure oxygen. In 1968, marine scientist Walter Starck introduced the first commercial CCR, called the Electrolung, which used polarographic sensors. The next year, BioMarine Industries launched its CCR-1000, the predecessor of the US Navy’s Mk-15/16. The unit was the first mixed gas rebreather to use galvanic sensors, which do not require a power supply.
In addition to removing a diver’s exhaled carbon dioxide, a rebreather must measure and maintain a safe and efficient level of oxygen, as measured by the partial pressure of oxygen, or PO2, via oxygen sensors.
Measuring PO2 correctly is critical, and failures can be fatal. Too little oxygen can cause hypoxia and loss of consciousness, and too much can result in central nervous system toxicity and convulsions. Since sport divers began using CCRs over twenty years ago, both conditions have caused numerous drowning fatalities.
With the exception of Poseidon and military manufacturer Avon Underwater Systems, modern close circuit rebreathers have more or less used the same type of sensor since the 1960s. Rebreathers typically use three galvanic sensors, averaging the readings of the two closest sensors and ignoring the third in a protocol called “voting logic,” originally created by Starck in response to the sensors’ noted unreliability.
Even with this voting logic, however, the sensors can be unreliable (See “PO2 Sensor Redundancy” in Additional Resources below). The galvanic sensors are cheap and time-tested, but they need to be recalibrated before every dive and expire after about a year. The new sensors—called “solid state sensors” or optical sensors—are expected to be more precise, reliable, and durable, though significantly more costly.
Galvanic sensors are essentially wet-cell batteries that generate a millivolt current proportional to the PO2 in the loop. Conversely, Poseidon’s solid state sensor uses luminescent quenching, wherein a red LED light excites the underside of a special polymer surface, which is covered with a hydrophobic membrane and exposed to the gas in the breathing loop. A digital color meter then measures the responding change in fluorescence, which is dependent on oxygen pressure, and an algorithm calculates the PO2.
Experts more or less agree that the right solid state sensor could make rebreathers safer, but the market is split on whether the technology is ready for use in rebreathers and just how much better they’d have to be to justify the cost.
Field Test Results
Poseidon advertises its sensor as “factory-calibrated and absolute, delivering unsurpassed operating life, shelf life, and calibration stability.” Richard Pyle, a senior curator of ichthyology at Hawaii’s Bishop Museum who works with Poseidon-affiliated Stone Aerospace, has tested Poseidon’s sensors for years, initially as a passive offboard check against Poseidon’s traditional galvanic sensors. Later, in November 2019, he said he began testing Poseidon’s prototype with the solid state sensor as the primary sensor in the unit. “From my perspective as a rebreather diver, this is the most significant game-changing way to know what you are breathing,” Pyle said. “We will never go back to the old oxygen sensors.”
Pyle said he’s yet to fully analyze the data he’s collected to compare the performance of the solid state sensors against the galvanic sensors, but that he’s had zero failures with the solid state sensors in the time he would have expected to have 50 to 100 failures with the galvanic sensors.
Likewise, Brian Greene, a Bishop Museum researcher who has tested the Poseidon sensors with Pyle, estimated that he’s made hundreds of dives with the solid state sensors without failure. But, not everyone has had this experience.
Sonia Rowley, an assistant researcher at the Department of Earth Sciences in University of Hawai’i at Mānoa, told InDepth that she experienced a variety of repeated failures when testing Poseidon’s system alongside Pyle beginning in 2016 and 2017, and Rowley dictated to InDepth specific dive logs detailing many of the failures. She wrote about her experience in the book “Close Calls.”
Poseidon CEO Jonas Brandt said the company has tested the sensors since 2017 at different depths and temperatures, and that it has only seen one possible failure.
Arne Sieber is a sensor technology researcher who said he developed the O2 sensor validation technology used in the Poseidon rebreather and sold the patent to Poseidon. Sieber is now researching uses for the solid state sensor including in the medical market. He told InDepth he believes the best way to incorporate the solid state sensors into rebreathers would not be to substitute one for the other, but to combine sensor types and design a rebreather that incorporates both.
Traditional galvanic sensors have advantages over the solid state sensors, Sieber said—they’re cheap, simply designed, low-voltage, and time-tested. Also, while solid state sensors are very accurate at measuring low PO2, they become less sensitive at about 1.6 bar, and are more prone to incorrect readings of higher PO2 levels than galvanic sensors. As for whether the sensors can function in humid environments, Sieber said the sensors can work well in liquids, such as when used for blood analysis (though the sensors are used to measure much lower partial pressures of oxygen) and for measuring oxygen content in the sea. Liquid can delay the amount of time it takes a sensor to read a partial pressure, but it does not falsify the results, Sieber said. Of Poseidon’s system, Sieber said, “It’s a good start. It’s very important that someone starts. Someone always has to be the first one.”
Brandt said divers have the option of combining sensors in the company’s SE7EN rebreather, using either two galvanic sensors, two solid state sensors, or one of each, and said it could be argued that using one of each sensor is the most reliable.
Meanwhile, a catalyst may be coming to encourage the development and adoption of solid state sensors in Europe, Sieber said. European Union rules restrict the use of hazardous substances in electrical and electronic equipment, but galvanic sensors (which have an anode made of lead) have been granted an exemption in medical products because there is not a suitable alternative. The exemption is set to expire.
Poseidon’s solid state sensor sells to end users for as much as around $1,500USD, and its SE7EN rebreather units use a maximum of two onboard sensors. [Note: Poseidon sells the sensor for 6800SEK plus VAT from its website, which equates to 944 USD, some outlets in the states sell them for much higher]. Galvanic sensors, meanwhile, cost around $100USD, last one year and divers use three at a time. And, that’s just the cost of the sensors themselves: Manufacturers have to make significant investments in, and upgrades to, electronics systems to accommodate solid state sensors.
As for how long the sensors actually last, even the manufacturers don’t yet know. Poseidon has some from 2014, and they still work but have to be factory calibrated every two years. Galvanic sensors need to be replaced annually, while solid state sensors are expected to last much longer.
Brandt chalks the debate about its sensors up to competitiveness in the market. “I don’t think anyone likes that somebody cracked the nut,” Brandt told InDepth. Poseidon is ready to share the technology with other dive companies and manufacturers, Brandt said, but there have been no deals to date. “We wanted to raise the bar in technology and safety with the rebreathers, and to be honest, we haven’t said to anyone in this business that this technology is exclusive or proprietary.”
When the company first debuted its sensor, Brandt reported that companies like Hollis and Shearwater Research expressed interest in licensing the technology, but nothing has come so far of those discussions. Brandt did say one manufacturer reached out right before the pandemic. He declined to say which, but shared that it was a European company. Hollis brand manager Nick Hollis said his team recalls a conversation with Poseidon, but that it was back in 2014 or even earlier.
Shearwater director of sales and marketing Gabriel Pineda said the company is still interested in solid state sensors, but they see an issue with the price. “If you make the economic case of traditional galvanic sensors versus solid state or optical sensors, you have to dive a lot, and it takes a long time for these to make economic sense for a diver.”
Of course, Shearwater is not a CCR manufacturer, but the company is interested in seeing whether the sensors would be viable for use with its electronic control system that is used by a majority of rebreathers on the market. Shearwater currently has no immediate plans to license the technology from any manufacturer but Pineda said the interest remains.
Meanwhile, Poseidon’s solid state sensor CCR is still making headway, Brandt said. The current biggest buyer of the Poseidon units is the military (Brandt said three European Union countries’ forces are actively using the sensors). The sales have continued throughout the pandemic, and, over the past six months, Poseidon has started an upgrading program, allowing divers to add the new sensors to their old units. Poseidon is looking into a program Brandt compares to Apple Care, where customers can pay a fee for maintenance throughout the life of the sensor.
Meanwhile, Avon Underwater Systems is using three solid state sensors in its MCM100 military rebreather. Kevin Gurr, a rebreather designer and engineer who sold his company, VR Technology Ltd., to Avon, said the company uses the sensors “because of the increased safety and the decreased user burden as far as daily calibration.”
Gurr, who designed and produced the Ouroboris and Sentinel closed circuit rebreathers at his prior company VR Technologies Ltd., believes it’s the cost that has discouraged other manufacturers. “It shouldn’t be about cost at the end of the day,” Gurr said. “The digital interface is so much safer.”
Martin Parker, managing director of rebreather manufacturer AP Diving, said his company follows solid state sensor development but has yet to come across a sensor that meets its accuracy requirements. One such sensor using luminescence quenching can achieve good accuracy through a replacement disk the user must apply to the sensor surface after each use.
“Having been in the diving business for 50 years, we don’t believe it is on any diver’s wish list to have to re-apply every diving day a new component, as simple as that is to do,” Parker told InDepth. “With no easy external measure of accuracy prior to the dive, it is easy to foresee that many divers would ‘push their luck’ and use the discs for multiple days, then when they get away with it, they would encourage other divers to do the same… with the inherent risk of DCS or O2 toxicity.”
Parker said that he’s aware of two additional sensors under development, but neither has shown a working product yet. He declined to identify any of the manufacturers, citing commercial sensitivity. “Hopefully, we will get these to evaluate in the next 12 months,” Parker said.
Divesoft co-founder Aleš Procháska said he believes Poseidon’s approach to the sensor could “lead to success.” Speaking generally about solid state sensors rather than about Poseidon’s specifically, Procháska said his company isn’t yet utilizing solid state sensors because he believes the sensors are unable to function in humid environments with extreme water condensation and not applicable over a wide range of pressures. To be able to use one of these sensors in a rebreather, Divesoft wants it to be durable in high humidity, consume less energy, and have a good price-to-lifetime ratio.
“It is possible to build a CCR with the currently available O2 solid state sensor but not without sacrificing important properties of the breathing apparatus,” he said, such as size and energy. “Overall, the reasons why no one currently sells this technology on the market seems to be quite simple. It’s extremely difficult to come up with a suitable and functional principle that would lead to a cheap, small, and low energy consuming solid-state sensor. Despite this, I do believe that it’s only a matter of time until someone solves this one.” Asked via email about the status of DiveSoft’s own work on the technology, Procháska replied, “Well, as I said earlier, it’s just a matter of time,” adding the text, “Aleš smiles.”
Halcyon COO Mark Messersmith said that divers are slow to embrace new technologies in general, and the current sensors just aren’t deficient enough to merit widespread adoption or the investment from manufacturers. “It’s not unlike many other technologies,” Messersmith told InDepth. “People are often slow to embrace a new technology if the existing technology is functional. The existing tech needs to be vastly deficient, and existing oxygen sensors are still largely functional.”
The bottom line: Solid state sensors might very well be safer, but there isn’t enough incentive for the market to make them a reality.
David Thompson, designer of the JJ-CCR, told InDepth they don’t use the sensors because he doesn’t believe the technology is ready yet and research in that area is extremely expensive and difficult for what he believes is essentially a small market. “Analog cells have a long history, and in the right hands are very reliable, easily available, and have a long history of working in a rebreather environment which is very hostile,” Thompson said, adding that high humidity and temperature in a rebreather is a challenge for any sensor. “I am sure it will be in the future, but that future won’t be here yet.”
Rebreather Forum 3 Proceedings: PO2 Sensor Redundancy by Nigel A. Jones p. 193-202
Alert Diver: Oxygen Sensing in Rebreather Diving by Michael Menduno
Wikipedia: Electro-galvanic oxygen sensor
Ashley Stewart is a Seattle-based technology journalist and GUE Tech 1 diver. Reach her via email: firstname.lastname@example.org, Twitter: @ashannstew, or send a secure message via Signal: +1-425-344-8242.
Thank You to Our Sponsors
Top Tech Stories of 2021
We’d like to kick off the new year with a selection of 11 hand-curated stories from our deepening well of...