Sorry for dredging up an old post but this line had stuck in my head for a couple days and the reason why coalesced this morning... I wonder if the problems the Quad Float have are due to those tweaks; Their attempting to wring the nth degree of performance out of them means avoiding some of the design compromises that ensured reliability.
Similarly, the original Quad ESLs are some of the best speakers ever made, but only within fairly proscribed parameters: That they will only be used at moderate volumes, that there is appropriate space for them that is neither too large nor too small and can accommodate their best positioning, that the room is neither too dry nor too humid, that your preferred music isn't too bassy... Speaker designs that attempt to work around the ESL's shortcomings tend to end up being other things entirely. It's arguable that there might not be a singular optimal speaker design, but instead a range of optimal speaker designs, and Quad happened to discover one of them very early on.
Interesting enough, I was talking to another head-fier who had quite a bit of experience with the original Floats, and he said the problems with the Float QA I described sounded very similar to the problems the original Floats exhibited from time to time. He described some inherent design flaws with the original models, though to be fair many of those were purposeful and part of Jurg Jecklin's philosophy going in to get their sound a particular way. He described some potential fixes and thinks that these needn't be compromises any longer. Just from chatting up some folks who know a lot more than I do, it seems like they haven't tweaked the original design enough. It's still pretty dated and apparently it's a matter of some simple fixes.