Thanks, Rob. I have since learned that I'm not the only one with this problem. Most have given up, but It think it can be fixed. The ciruitry is very simple -- I may have to just trace out the schematic. It looks like they have a threshold too low on the LED solid state switch or the data slicer. The low cost designs often sacrifice beta and temperature stabilized biasing, which can be easily fixed (if that is the problem). When they ommit the bias stabilization, you get variations in performance with temperature and across production lots. I was hoping someone had already fixed this or at least, had a schematic to make my job easier.
Regards,
Dean