Good explanation, and the reference to the other forum was useful, but probably on another level than I am dealing with here.
What I have is a set of NiCds (100 cells to be precise) - 10 strings of 10 cells in series, which are in turn wired in parallel. 12V nominal, 10AH for the entire battery. It's a home-made DIY deal, and the packs that make up the final battery are all rather aged, but were "brand new in the box". Some perform better than others, and the imbalances that are inherently there cause stability issues when charging, standing, and discharging. I can't complain too much though, they were all absolutely free - I have all of about 20 bucks in the whole project, with half of that in just the box that contains it.
That said, my biggest problem is that in order to keep the NDV tendencies of certain selected cells (scattered throughout) from preventing the others from reaching full charge, I reduce the temperature of the entire pack so that they cannot go into thermal runaway. The lower temperature reduces the "formation" of the NDV in the cells that are (presumably) weaker than others and therefore reach full charge before their stronger counterparts.
The battery performs ok as long as I use the charge fairly soon after removing the floating voltage to keep it charged, so I know that there are some "leaky" cells that self discharge at a higher rate than some others as well. The two are clearly related (at least in my mind), and so I am doing whatever I can to mitigate the effects of the cell ages by keeping temperature as balanced as possible.
Problem is, from what I understand, the charge acceptance increases as 32F is approached, but then begins to degrade rapidly as the cell temps drop below this, and at 0F, the acceptance is apparently (according to what I can find about the subject) near nil at those temperatures.
But when I charge them at those temps, I seem to get maximum capacity, but at a significant price - they take FOREVER to charge because the charge rate drops to next to nothing and the terminal voltage shoots up rather high.
I'm using current limiting, and under normal float (after a lengthy time), it stabilizes at 13.7V or so with the current limited to 250mA (C/40 rate). The cells get barely noticeably warm (particularly toward the center of the pack) when holding this "pattern". Unfortunately, this current is alarmingly high for that terminal voltage (from what I can gather), and indicates just what I have found otherwise; aging is showing it's ugly head, even though the cells have only been cycled half a dozen times. They spent most of their "life" hanging on a shelf.
When I drop the temperature, the voltage begins to climb, and when they bottom out somewhere in the 5F range, I've seen them as high as 16.2V, and the current limiter begins to "lose regulation" - less than 200mA is typical. The voltage source is set at 19V before the limiter. Somehow, there's some odd math there. Can't put my finger on it.
From other things I've read, I should limit the current at those temps anyway, because apparently one of the bigger factors is excessive gassing, which at those temps do not readily lend themselves for recombination, and the pressure builds, which leaves voids in the electrolyte, further perpetuating the problem. I'd suppose that a "negative" runaway condition could potentially take place, but nobody seems to indicate any damage as long as very low rates are employed (which according to them, I am within).
So my question is then, how do I win this? Take the temperature down to 32F and regulate it there somehow? Easier said than done, really, even though I'm not currently worried about the energy consumption that it takes to utilize this particular battery. Capacity is the paramount consideration, with charge rate as a strong second.
There are some pics of the battery, just in case anyone has any ideas as to how to improve it given what's there. There are small gaps (~ 1mm or so, propped by toothpicks) between the packs to allow for air flow, but I realized that it was going to be very restricted in there, so I used two fans to pull the air through them. The fans are before a blocking diode, so that they only run during charge. They are also wired through a DPDT to allow me to switch between series and parallel for hi/lo.
Steve