I'll have to test this again in 1.2.3, but using my design in this: http://forum.industrial-craft.…page=Thread&threadID=5236 For what should be either 3.5 or 3.6 seconds of redstone signal off I see 3 seconds of produced power; but at either 4 or 4.1 seconds (earlier in the video) I saw a positive thermal co-efficient instead of a slightly negative one as the numbers indicate should exist; I unfortunately did not examine the power produced during that event as the reactor was shut-off by a minor bug in the output when partly full MFSU setting.
I wonder:
a) what the startup delay between redstone off and reactor is (or if it simply checks once per second at a specific offset).
b) How many cycles of thermal rise are associated with the pulse of 'off' including regions during which no power is consumed.
c) If any fuel is used during the non-power production time periods.
I know already how I plan to setup test cases for B when I finally have time; that's easy enough with the correct setup. A is more difficult; however sufficient observation should lead to workable conclusions. C is fairly difficult to measure without examining probably NBT tags in inventory... which is likely the easiest way of looking at it, but that's going to be annoying.
If anyone wants to investigate these before I again have some time in about 3 weeks please do so.