Hello JM,
This gets into an engineering problem, and as you know, engineering is about living comfortably with the compromises made.
If you have cells that are inconsistent from cell to cell, you can raise the cut off voltage, but you loose capacity and performance. The next step is to "match" the cells before they are made into a pack, however, this is time consuming. I think the best bet is to purchase cells from a manufacturer that will insure you that the cells you will be getting meet your expectations, however, unless your are ordering several million cells, this is almost impossible to do.
In competition RC use, cells are matched to 1% or closer.
In less demanding use, you can match to 2 - 5%.
When I am putting together a pack, the first consideration is how many cells are going to be used. The more cells in the pack, the tighter the matching to get reliable long term performance, but with more cells there is increased variability between them. I generally match on 2% for smaller packs and 5% on larger packs in less demanding service.
For example, I picked up 20 cells to check out. One cell shorted out and one had abnormally low capacity during the break in of them, so I tossed both of these cells. Out of 18 cells, if I averaged them, then took a high value of 102% and a low value of 98%, I ended up with 8 cells that fell in that range. Out of the remaining 10 cells, I find that I have 3 (out of 4) that are matched at a lower capacity, and the remaining 6 are matched at a higher capacity.
Out of 20 cells, I ended up with 2 dead, a well matched 8 cell pack, a slightly lower capacity 3 cell pack, and a slightly higher capacity 6 cell pack, and had one left over that still worked but did not fall within any of the other ranges.
Using a very well matched pack and taking proper care of it, I can "adjust" my low voltage cut off to a lower value and still get good performance from the pack. However, if I let the pack "waste" away, I run a chance of reverse charging a cell if the pack becomes unbalanced.
Panasonic suggests a 1 volt per cell cut off for packs up to 7 cells. For a 10 cell pack they suggest using a cut off voltage of 1.2 volts per cell.
Powerstream suggests a 1 volt per cell cut off for a 10 cell pack, if the cells are matched within 3% (absolute). This means that if your lowest cell comes in at 1700 mAh, the highest cell in the pack should not be over 1751 mAh.
When I use the 3% absolute formula, I start with the lower capacity cell and work up from there for a tighter match, but you can also start with the highest capacity and work down.
My usual method is to run an average, toss the high and the low out, then look at 2% above the average and 2% below the average. This often comes very close to the 3% absolute figure.
This is where the compromises come in...
Can you afford to take the time to match your cells? For general matching purposes you can get by with a 2 hour charge and a 2 hour discharge. Let' see now, after the break in we have 2000 cells at 4 hours each = 8000 hours run time, and then you have to analyze the data.
The next question involves the quality of the cells. If I move what I consider "value" from the price of the cell to the consistency of the cell, How much is this "value" cell going to cost and can I afford it?
The next question has to do with ethics. Should I follow what the re-labelers are doing with batteries and simply inflate the runtimes expected from my battery pack, or should I advertise a minimum runtime with a replacement offered if the pack does not meet the minimum runtime? Along with this comes the question of how much is it going to cost me to replace battery packs if people are not satisfied with them.
I am sure there are more questions, but that should be enough to get started.
Also, is the current draw fixed? Is it possible to shave it down a little?
As you can see, this can get a little involved. I suppose you could say that your device will run for at least 1 hour on the battery pack and have people praising you because they are getting 3-6 hours off of their packs...
Tom
