A few questions are bouncing around upstairs, hope someone can shed some light.
Picture;
http://test.rebreatherworld.com/gallery/files/6/2/5/RBunit011.JPG
Shows the two extendair carts in place, how is it sealed to prevent bypass? Most Micropore units have a sealing ring required, there isn’t one here.
[/quote]
You have a lot of good questions. If I can answer them in turn:

The use edge seals, like a oil can on a car. For the prototypes we simply glued draft excluder around the Micropore cylinders.

The main problem with this was the things were hard to get out.

The Micropore seals waste 11% of the usuable area. We have a variant of this which gets it down to about 2% which we are using on Proto IV.

There was a lot of debate about how to handle the notch on the Micropore canister. It acts like another pore. Ignoring it means about 0.2% of the gas glow bypasses the scrubber. Proto II and III dealt with it, but now we are going to simply use a keying notch to partially block it so bypass is about 0.05% of flow.

This is Proto II, used for flow, WOB and condensation testing (hence aluminium for worst case), so the O2 sensor was replaced by nylon rod the same diameter. You see the real sensors in Proto III.

The bag was used to test two different flow paths, so either has a plug in one port for the serial scrubber config shown here or has the scrubber with the second port (parallel config). The plug saved us doing two bags of each type. We have now settled on a parallel scrubber and abandoned the serial scrubber, after a lot of testing and trying out different variants.
It is an ADV and auto shut off valve. See:
http://www.deeplife.co.uk/files/auto_shut_off.pdf

We have now separated the components physically, as the cord was hard to reach. Proto IV has the shutoff valve on the scrubber, the ADV is one which is extremely resistant to freeflow and on the inhale bag, the exhale vent is moved from the mouthpiece to an exhale bag, as even if the exhale hose fails the mushroom still means all is breathable. However, the fully integrated version works.

We used ally for Proto II because the purpose was to do flow testing, WOB testing and condensation worst case evaluation. Ally has much worse condensation than Delrin or Kynar, and was used to develop the flood recovery system seen on Proto III: Proto III was made from Delrin.

In terms of costs, we have our own workshops but Delrin costs about twice that of machining ally.
Closely guarded secret at the moment.
This is a galvanic sensor from an established, though unusual, sensor manufacturer with who we work.

We are trying to remove Galvanic sensors entirely using a Sol-Gel sensor for both O2 and CO2. This would be disposable, as well as having a very long life.

For the first versions, it will take the same Galvanic sensor as on the Evo or sensors a third party supplies that are rated with a 5 year life. We have not convinced ourselves that this is really 5 year - it is 5 year without abuse, but we are hard people to convince that with salt, muck, moisture it is 5 years without actually fitting a load and waiting. They certainly last 3 years.
Shutting the gas off forces the user onto open circuit.

That is, when the shutoff valve closes, it closes the loop. The ADV supplies gas, which is vented to clear the loop. The user still breaths but can see they are on open circuit, so it does not endanger the user.

The situations where "You are a loony!" message comes up is if the unit is dived without checks or after failing checks.

The "Too deep!" message comes up if the depth limit for the unit is exceeded. The unit can be configured for an particular depth limit, including in the expedition version, unlimited. The expo version has WOB assistance using a fan and various other features.
It warns when gas is under 50 bar and further warnings every 10 bar.

Remember, the valve pushes the unit into semi-closed unless the loop is completely unrecoverable (that is, exhaling clears the loop, and if the loop is then breathable the valve opens).
To override it, just pull the override lever.
Until the loop is breathable, the user has to have pressure on the manual override. When it is breathable it resets automatically. The design is shown in the presentation mentioned above.
It is brushless.
Yes, but the size of the fan is an embarrasement just now. I will do some pictures and post them. The main issue is how to get enough pressure to be useful and keep the fan and its power consumption low.
Burst lungs occur if someone forgets the first rule of diving: breath in, breath out.

There is overpressure on the exhale bag which will be venting continuously, and the breathing loop is shut (unit is on O.C.). It is quite hard to leave the surface in this configuration, unless they are weighted really heavily in which case the system has simply put them on O.C. until 10m.
13 has killed people. Some of these may be manifestations of failures 1 or 2, but training can program a user to always switch the unit on if there is clear evidence that users are dying because they forgot to switch it on.

Remember, the user has done a pile of checks, may feel very seasick, jumps overboard and fins hard in a current before descending. Did he switch the unit off after doing the checks? Looking at the displays on a rocking boat made him more seasick.

In the design of aircraft control, reactor control and other critical areas, if something that can be done to prevent a user error turning into a critical area. Why kill the guy just because he felt really seasick? Or his mate switched off the unit for him to "save the batteries"?