apparently, the message got clipped when I originally posted to the boards, here's the begining of the message again, as i remember it:

First, I wondered which sources you had used to work up your autocalibration algorithm. I've found the Bell et al (MER Pancam Investigation) and Maki et al (MER Engineering Cameras) greatly informative with regards to the methods of autoexposure calculation onboard the rover. But, the real 'holy grail' of calibration documents to me, would be a JPL document referenced by the Bell paper:
Bell, J.F., III et al., Mars Exploration Rover (MER) Project Pancam and Engineering Cameras Calibration Report, JPL Document D-19826, MER-420-6-700, Jet Propulsion Laboratory, Pasadena, CA.

I've yet to be able to find this one available anywhere publically, though I'm curious if anyone else has had better luck.

In regards to the autocalibration done on the rovers itself, I've found some great references to what's being done within the datasets of the Maestro program (http://mars.telascience.org/home). These datasets include some, but by no means all, of the individual image frames, included along with their PDS tags (stored in timestamp.msml files) which contain many interesting states of the rover at the time the images were taken. I'll use, as a reference, something from the newest dataset, Post-Brushing, Pre-Ratting Humphery from sol 55 spirit. The L2 frame of the color set is: 2P131251640EFF1147P2583L2M1 and L3->L7 follow sequentially afterwords.

For this autoexposure set, all 6 frames were set to
OBSERVATION_REQUEST_PARMS.AUTO_EXPOSURE_DATA_CUT = 2500
which is the point (2500 out of 4096 for the onboard 12bit image, i'm assuming) in the histogram by which all pixels above (brighter) are counted towards the numberator of the next value, the OBSERVATION_REQUEST_PARMS.AUTO_EXPOSURE_PERCENT. The value of exposure percent was set to 10 for the L2 frame, and 20 for the rest of the frames. This would mean that the L2 10 % of pixels above ~ 2500 DN when the image is cut off, whereas the remaining frames have 20% of the pixels above that same value.

I wondered, because of this, how your autocalibration could accounts for these differences between the autoexposure factors?...