I know, but this just seems like where I'm going...
Operating voltage:
The lithiums will never overcharge, there are 10 of them, and with panel OTV being 39V, the resulting maximum voltage per cell (when balanced) is 3.9V/Cell, worst case.
Balancing:
The balancing resistors are 150R and will deplete the cells somewhat when the sun isn't shining, at a rate of something right around 25mA. When full, a cell disconnects, with the balancing resistor carrying the base charge current to the other cells. As they all fill up, they all become disconnected until they sense a load. Thus, their tendency to stay balanced will be fair to good. I am also looking at a nightmare-ish MOSFET bypass design that I ran across a while ago... Something else entirely...
Protection:
They will be augmented by the cap array to help divert spikes (that might damage the protection circuitry), and I wouldn't have tried this if they did not contain those internal upper limit cutoffs (trigger @ 4.3V) in each cell. So I'm not too worried about overcharging. The KA3525 also has a shutdown pin, which I intend to actuate by a design based on the lower cutoff for the lithium ions.
If there IS an explosion/fire:
They will never be inside, only part of a controller box that will be in close proximity to the panels (designing the system as a portable quick-deploy power source)... Even if one blows out and takes the pack, sure it will be pretty, but damage will be minimal. They will be 'off to the side' so property damage will be minimized.
Why the design has gone where it has:
As I said before, I'll be looking hard at various designs, including the one you suggested, but ATM this is where it's at. This design is actually the one that I am familiar with; I built the power supply for my car amp some time back which runs a 60W amp, and hasn't had the first hiccup yet. I wonder sometimes about frequency (osc/toroid) issues that arose during testing, but I was stressing it far beyond what it ever sees in the car, so it hasn't been a problem. That's another thread however (in fact I think I posted about it last year or so), and only applies to this in the sense that I was happy with the design. The major difference as I recall was that I believe that version is controlled by a TL494.
Why I can't do much else:
The parts are free, aplenty, and with the exception of the complexities of dealing with lithium ion cells, it's very simple in design. This was my primary goal. Taking it where it leads me. If I get a 1:2.5 or better current gain out of this, I will be satisfied completely, and might even put up the schematics with parts lists, maybe even a kit.
Why I'm just appending the ongoing thread:
I haven't started a new thread because this one accurately shows the lineage of my thought process, as well as input from others. When I have a satisfactory design in use, I will terminate this thread and post a new one, or two or three... depends on what I come up with.
The final straw that led to all of this:
ATM, this is close to a synchronous supply, only with push/pull instead and a secondary. The boards that I have available pretty much dictated this, not so much my desires. I went to Radio Shack earlier to try and get a boost converter (for laptops) so that I could modify it and change it to a buck converter. What I walked out with looked promising (already in 'buck' configuration), but turned out to be a let down; a 317 with 4 voltage settings. I was irritated to say the least. Hence the heavy digging into what's available and free...
The very next move:
After initial balancing and testing/tweaking (balancing is running as I type), I am going to look into sensing the battery line and use that to kill the PWM as well, so I don't have to dump the surplus.
Steve