Conan makes some valid points, and these are the things that have made documenting the ULA so difficult. I have spent a year doing research and following clues, and have uncovered new facts and information.
The ULA resin has been removed and the surface imaged with a high power optical microscope after grinding off the layers.
I have the documentation for the 5000 and later ULA versions, which I got by researching the Ferranti Archive and other sources. This has taken a year of intense research. Now I have the optical images of the 6C001 ULA silicon layer, I am able to document the 6000 series structure, including resistor values. Note that these values have been derived from later ULAs that have the same propagation delay, power rating and cell structure as the 6000 ULA. Note however that the exact resistor values are not required because it is their ratio that is important. I think the 6000 series ULA was designed and produced only for Sinclair.
It has taken me a month to document one quarter of the circuit at the transistor level.
Yes there have been several versions of the ULA. The 6C001 ULA has a different metal layer to the 5Cxxx ULA because the cell structure is different. The circuits will be almost identical however, and through careful analysis of propagation delay and signal timing, it is possible to show where and how the circuits are different.
I can only afford to image one ULA, because it costs $500 for one hour and takes a whole day.