Voting logic doesn't make sense to me. It seems like most rebreathers that use 3 cells (or more, some can take a ridiculous amount of cells) and employ "voting logic" if they use a solenoid.
From what I understand, voting logic finds the two cells that are most alike and then averages them, ignoring the cell that is must unlike the other two.
This seems beneficial only if you make the assumption that any error (both minor accuracy errors AND major errors) that occurs in one cell is unlikely to occur in another cell at the same time. I don't think this is a very valid assumption, and in fact, it seems like problems within the rebreather that would cause one cell to fail or report erroneous or innaccurate values are actually very likely to effect the other cells in the rebreather which are usually nearby, subjected to the same environment, have similar ages, histories, and were probably manufactured at the same time. You get some dampness in your loop that makes one cell report a tenth of an ATM lower, it is likely at least one other cell will also be effected.
In that case, it seems that voting logic actually does more harm that good if more than one cell is effected, and a flat out average of 3 cells would actually report a MORE accurate value when two are innaccurate and one still works, wheras voting logic would discard the reading from the only working cell.
In addition, voting logic always assumes that one cell is broken. It doesn't assume that NO cells are broken. It doesn't assume that two are broken. It won't assume that all the cells are broken. It always assumes that exactly one cell is broken.
It seems like every manufacturer's approach is to avoid engineering a solution to the problem and just cramming in more and more cells with more computers to do more voting logic and maybe a bunch of other cells that are separate from the voting logic so that the diver can see a whole assload of different PPO2 numbers during a dive and try to figure out which one is the most accurate one. I mean, I guess it is OK to assume that if you have a hundred cells in your rebreather, at least one of them is going to work, right? But how do you know which one is the working one when they all say different things? Should you even have to worry about that? How is this even remotely considered an efficient solution to the problem of cell accuracy?
Why not just use two cells and some clever routing of the gas additions to the loop and use them to check and calibrate each other?
One cell can provide readings to the diver and drive the main O2 addition solenoid, while the other can get validated periodically during the dive by the electronics by using two other solenoids to blow tiny gusts of O2 and known composition Diluent across it at a known ambient pressure to verify voltage linearity and calibrate it. Then just check the primary cell that is driving the electronics and see if it matches the backup when this validation of the secondary cell isn't being performed. By doing this constantly during the dive, basically if the cells ever say two significantly different things, or the voltage linearity of the secondary cell is not confirmed, then something is CLEARLY wrong and the user goes to open circuit. Why? Either the primary cell isn't showing what it should because it doesn't match the cell that we are constantly testing and confirming that it works, or the secondary cell isn't passing the tests that we are performing so we can't validate the primary cell and be sure it works. Otherwise, you can rest pretty much assured that the system is not going to tell you a PPO2 value that is misleading.
So, is there something I am missing? How did voting logic become so popular among eCCR users and manufacturers?
From what I understand, voting logic finds the two cells that are most alike and then averages them, ignoring the cell that is must unlike the other two.
This seems beneficial only if you make the assumption that any error (both minor accuracy errors AND major errors) that occurs in one cell is unlikely to occur in another cell at the same time. I don't think this is a very valid assumption, and in fact, it seems like problems within the rebreather that would cause one cell to fail or report erroneous or innaccurate values are actually very likely to effect the other cells in the rebreather which are usually nearby, subjected to the same environment, have similar ages, histories, and were probably manufactured at the same time. You get some dampness in your loop that makes one cell report a tenth of an ATM lower, it is likely at least one other cell will also be effected.
In that case, it seems that voting logic actually does more harm that good if more than one cell is effected, and a flat out average of 3 cells would actually report a MORE accurate value when two are innaccurate and one still works, wheras voting logic would discard the reading from the only working cell.
In addition, voting logic always assumes that one cell is broken. It doesn't assume that NO cells are broken. It doesn't assume that two are broken. It won't assume that all the cells are broken. It always assumes that exactly one cell is broken.
It seems like every manufacturer's approach is to avoid engineering a solution to the problem and just cramming in more and more cells with more computers to do more voting logic and maybe a bunch of other cells that are separate from the voting logic so that the diver can see a whole assload of different PPO2 numbers during a dive and try to figure out which one is the most accurate one. I mean, I guess it is OK to assume that if you have a hundred cells in your rebreather, at least one of them is going to work, right? But how do you know which one is the working one when they all say different things? Should you even have to worry about that? How is this even remotely considered an efficient solution to the problem of cell accuracy?
Why not just use two cells and some clever routing of the gas additions to the loop and use them to check and calibrate each other?
One cell can provide readings to the diver and drive the main O2 addition solenoid, while the other can get validated periodically during the dive by the electronics by using two other solenoids to blow tiny gusts of O2 and known composition Diluent across it at a known ambient pressure to verify voltage linearity and calibrate it. Then just check the primary cell that is driving the electronics and see if it matches the backup when this validation of the secondary cell isn't being performed. By doing this constantly during the dive, basically if the cells ever say two significantly different things, or the voltage linearity of the secondary cell is not confirmed, then something is CLEARLY wrong and the user goes to open circuit. Why? Either the primary cell isn't showing what it should because it doesn't match the cell that we are constantly testing and confirming that it works, or the secondary cell isn't passing the tests that we are performing so we can't validate the primary cell and be sure it works. Otherwise, you can rest pretty much assured that the system is not going to tell you a PPO2 value that is misleading.
So, is there something I am missing? How did voting logic become so popular among eCCR users and manufacturers?