Woodman1,
Looking at your logic I see one problem not related to counting, but to the reset of the counts. The DL05 does not have a real timeclock, so the use of V7770 and V7767 will not give you useable results.
Back to the counting issue. Presuming that all your inputs have the same common. One test if the length of signal on time (pulse catch) is the problem is to put a value of 50 (BCD) into V7633 and a value of 5 (BCD) into V7634. This will setup X0 as a pulse catch input. With this configuration if X0 turns on during any time during one PLC scan it will place a 1 in the address. If the input was configured as a normal input it would only report 1 if it were on (and if filter conditions were met) at the moment the PLC transferred the status of the input table.
Now disconnect all but one of the photo eyes from the inputs and place jumpers to the disconnected inputs so that only one of the photo eyes fires the programmed inputs. Run bottles for a day and note any differences in count. Do this for each photo eye until you discover if one of the signals is the one with the short pulse.
The bad news is that only one of the inputs (X0) can be configured for "pulse catch".
You have indicated that your application is bottle counting. You have also indicated that the signals you are receiving are good. I would encourage you to double check this. The classic photo eye to use for bottle counting is a polarized retro reflective eye. The problem occurs when using one that does not have the ability to offer NC or NO outputs. The pulse catch will not work on a fast moving bottle if the photo eye turns on the input when it "sees" the reflector.