So this is the place where you all went...

[telemonster]

I'm making electronics because I want to. This project is also my last course work before getting my Master's degree electrical engineering.

One idea behind the design is using only 1-2 O2 cells per controller/monitor unit. Data between units is shared via CAN bus. If the CAN bus fails you can check individual cells from different units.
Example: Handset computer unit have 1 cell. Led HUD unit have 1 cell. Secondary handset got 1 cell. Total 3 cells but all units can see 3 cells as long as CAN is working.

Share the data, not the cells™

Not trying to poke holes in design, just thinking.... Always find this stuff neat.

Is there a downside to feeding the solenoid values to both controllers, except the failure mode of a flooded controller throwing off the readings and cabling?

Controller 1 fires the solenoid? Immediately goes into alert mode if a keep alive isn't RX'ed from the controller 2 or HUD controller?

Is the control functions and sampling the o2 cells handled by a MCU with a full stack operating system or lighter codebase? Real time? Management plane separate from the control?

Years ago on RBW there was an interesting topic about software compilers that ... do a better job of error handling for critical applications. Like filling unused memory with return instructions that will land things somewhere safe and predictable and what not. But I don't know how accessible those types of compilers are, and for what hardware platforms they exist.

 

[TONY CHANEY]

Love more and more each dive my “The Heat!” Not an eCCR and with 4 O2 cells: two to the standard hand set and two to my Dive Soft computer. The DS computer can support up to three cells but space is limited as far as wires go. The DS computer is great.