IBM 610 workstation computer 3378

Surprisingly, in some applications, it works just ducky. The *input* to the PLC is integer - 12 or 16 bit or such. Any analog outputs are integer. Well, why the heck should the math in between be floating point?

Especially when floating point isn't available on the system :)

More to the point - the water utility (I do mostly water-sewage-transit stuff) could care less if the poowell has 5.00 feet of poop or 5.00000001 feet. Makes no difference - they want the pump 'on' at 5.00 feet. If it's off by a fraction of a foot, no biggie - setpoints get adjusted up and down anyway. You're not going to be spot on anyway - the instrumentation inevitably never works as close as it should (for whatever reason).

In any case, the 'ok' and 'oh poo!' points are always far enough away - the backup is a mechanical float (generally), which isn't even close to being accurate.

Just because you CAN calculate out to 50 decimal places doesn't mean you should or need to. Integer math on a PLC is cleaner and and less bug prone, and gets the cycle time down, which is more important anyway.