Shearwater Feature Request – “Vote Out” a cell

(NB: This post is targeted at rebreather divers)

The purpose of this post is to detail the situation when you would want to manually “vote out” a cell on a Shearwater computer and alternate solutions considered.

Background

The way I have my rEvo configured is that three oxygen cells are connected to the main Petrel controller and two backup cells are connected to the NERD.

The Shearwater computers have an automatic “voting logic” algorithm to determine which oxygen percentage to use for decompression calculations.

If you have three functioning cells, it basically averages them. If you have three cells and one of them is “acting up” then it figures that the two that are close together are correct and it “votes out” the bad cell and averages the two remaining cells.

If you have two cells, things get a little dicey when a cell fails. This has happened to me a couple times. Because the computer doesn’t know which cell is reading correctly, it uses the lowest PO2 cell for calculating decompression and highest PO2 cell for calculating CNS oxygen loading. This is the “right” thing to do in this case as it is the most conservative.

However — it results in a greatly increased decompression obligation. Just to give you an idea, I did three dives on 26 October while we were out diving airplane targets. I did two relatively short dives to 170 fsw and one dive to 120 fsw. By the time of the third dive’s decompression phase, I had two hours of decompression according to my NERD.

Below are graphs of the Shearwater Petrel and the NERD with the bad cell for each of the three dives:

Dive Number	Petrel	NERD
1
2
3

Below is a bigger version of the NERD profile during the 3rd dive. Note that the O2 mV is basically “stuck” for one of the cells abnormally low and you can see some some stats in terms of Time To Surface, etc.

You can see that I definitely violated the ceiling and ignored the deco profile for the NERD. The effects of violating the earlier dives compounds to the point where the NERD computer is basically useless at this point.

Potential Solutions

So, how can this problem be solved? There are numerous ways that I will detail below along with pros / cons and why I want Shearwater to implement cell voting.

Connect a Third Cell to the NERD – Fellow SoCal diver Drew Wilson does this. Then, if a cell goes out, it will use the normal three-cell voting logic and continue to “correctly” calculate the decompression obligation. Below is a picture of Drew’s cell tray showing six cells. The downside to this is that it introduces complexity and adds the cost of maintaining yet another cell.

Split a Petrel Cell – In this case, you would add a splitter and connect one of the cells to both the Petrel and the NERD. This would result in the NERD essentially monitoring three cells and one cell failure would not cause a problem. The downside to this solution is small but you are connecting the two monitoring systems which I like to be completely independent — but it would definitely be a way to solve it.

Use Internal Setpoint – In this case, when the problem happened, you would turn off monitoring of O2 cells and “hard code” your set point into the NERD which would then calculate decompression profiles assuming you maintained the set point. The downside in this case is that you are “wasting” the fact that the NERD still has one good cell. In addition, I use an “off board” Teric for this already.

Manually Vote Out Bad Cell – This is my preferred solution. It would give the user the ability to manually “vote out” a bad cell at any point before or during a dive. How do you know which one is bad? There are at least a couple ways to do this. The first would be to compare the cells in the NERD to the cells in the Petrel. It will be obvious very quickly which cell is bad. You could also do a Diluent flush and check which cells read correctly for the Diluent mix at the depth. There really aren’t a lot of downsides to this solution that I am aware of. Also note that I know of at least one other dive computer manufacturer that has this capability.

Shearwater — PLEASE give us the ability to manually “vote out” a cell!