It has 2, 3 is only useful if voting logic is implemented. There are some fundamental reasons why 2 can be better than 3, especially in a calibrated and monitored scenario.
Not true, an extra cell is 33% more data with which to make a decision, regardless of whether or not the electronics utilize It to make addition decisions. The issue with two cells is that the only calibration and monitoring is against a single other point of failure. It’s the same argument Andy Davis uses with the stock Pelagian, and falls apart because of the same reasoning. With 3 cells you can easily determine which of any 3 cells has an issue, or even if multiple cells are faulty. With only two, even monitored and calibrated, you’re counting on the other being fully functional as well.
Think about a frozen pitot tube affecting airspeed and an autopilot system, and that’s got substantially more intelligence backing it up. Not the best analogy, but if you only have one other comparative measurement to which you can reference, it’s easy for the calibrated and monitored system to fail. Adding a third point of reference allows you to check either of the other systems, as well as the system as a whole. The problem with the rEvo’s using a bazillion cells is you start introducing a fairly substantial number of failure points since you need an odd number to rule out any “tie.”
The idea that the system can do the math for you with two cells is fine until it runs into a situation where it cannot make a reasonable determination. With a third cell it would allow for a more failsafe reference.
Ultimately it comes down to trusting the machine. A calculator is better than the brain until you give it an illegal operation. And in the same way I’d fly with a G1000 glass cockpit, I wouldn’t rely on my little eTrex to fly IFR. There’s a substantial difference in the machines, and I don’t think Poseidon is the “autopilot” of the rebreather world yet.
