One thing to consider, is that the camera’s metering system has some nonlinearities which are dependant on the native aperture of the lens.
This was a big discussion topic when the K10 came out, and I tested it on my *istD, K10D, K7D and K5D, and somewhere plotted and posted the results of exposure error as a function of F-Stop. I should really do it also with my K1 MKII as I suspect there is also some error, because I notice with my sigma 70-200/2.8 a tendency to over expose 1 to 1.5 stops when used with the 2x TC. This is due to the camera thinking the lens is F2.8 as the TC feeds lens data straight through without correction, but the actual aperture is double what is reported. This is exactly the same symptom I saw with my K10D and the same lens and TC combo.
Here is the plot I posted way back when.
Understanding the chart.
The chart measures the greyscale value of a frame, the test image was a uniformly sunlit block wall or paved surface, which the camera should interpret as a 18% grey card. Perfect metering is with a greyscale value of 128 (out of the range 0-255). Around the middle of the greyscale value each full stop is an increment of 40-45 greyscale. This linear relationship holds over a greyscale range of about 30-225. At each end of the curve it is “S” shaped, with the next 3 stops at each end having greyscale ranges of 15, 7,and 3 respectively as you approach the ends of the exposure curve