CMOS for photometry

Likely gone over again and again, so my apologies. 

I am investigating upgrades to the observatory and have been seeing more and more CMOS sensor based imagers out there, 16bit, great cooling, and pretty significant QE, example, the FLI KL4040... expensive, yes.... but good for accurate photometry? We've been using SBIG STL units for years and done great VBR photometry with them.  So, thoughts on CMOS? Stick with CCD? 

Thanks a ton in advance.

John Blackwell (BKL)