Don,
Admittedly I'm not experienced, but I'm studying on the subject. That's why I'm asking for your advice too, and I appreciate it. It has made me question what I've read and do more thinking about it. So far, I still can't square your thoughts with everything I've found.
What I've read:
The worst case is a hot sunny day, with batteries at a high state of charge. How much will the panels themselves sag? Some people say 1.5-2.0v, on 18v panels. That would be good to know, as I'm only going by what I've found in reading. I do know people have provided actual readings showing a variance of 20% in panel voltage from cold to hot weather. That's probably +10% one way (cold), and -10% hot (my guestimate).
There can be losses from hot wiring, and from connections, and the controller's overhead, and the wiring to the bank. I suppose all of those shouldn't total more than 0.5v on a good system. Then if there is any small shading, the panel voltage drops. In the worst case, add all those losses together and you're down more than 2v before the losses from the wiring we are discussing. It doesn't seem prudent to me to add another 10% loss into that.
In an email to me, Bob said the target voltage at the controller should be at least 15.5, and shooting for 16.0 is better, and things still could be marginal on hot days. I don't recall seeing a lower number than 15.13 in his blog. ?
Thanks again for the good discussion.