Micrologix 1500 increment by .1 issue

Hey fellas,

I have a Micrologix 1500 ser.c on a bxb base.
I found this weird issue today to run by you guys. I have a momentary toggle switch wired to an input that is used to increment a (float) measurement field. I have a standard little one liner in RSLogix with a ons instruction to add .1 to my float each time the button is toggled. It works great but at certain numbers it flips out and shows ex. .8000001 then increment up it goes to .9000002 but then for some reason straightens up and goes to 1.0 , 1.1, 1.2, 1.3, then maybe 1.4000001 ect.

Really odd, anyone ever see this? I don't think I'm getting any bounce because it works flawlessly if I use 1 as my incremented value. The .1 flips this thing out.

Thanks

Search for IEEE Floating Point Precision

This has been discussed many times. Everything is working exactly as it should. Poke around this website and many others for an explanation that will help you understand. Good luck.

And remember that the MicroLogix 1500 supports the Long integer data type, so you can use a 31-bit-plus-sign value to run your accumulator.

1 is always 1 in Integer-land, but 0.000001 is sometimes zero in Floating-Point world.

Flopro & Ken,

Ah... Got it!
Thanks guys, I appreciate the hints.

Have a great weekend!

Excellent !

It's worth mentioning that a Long Integer is a signed value, so one bit is used to indicate the sign. So an "L" type data table element in a MicroLogix has a range of -2^31 to +2^31.

That's -2,147,483,648 to +2,147,483,647.

If your accumulator is counting by ten-millionths and only has one digit to the left of the decimal point, you can make each of those increments a 1, and still only be in the 10 million units range, much smaller than the over-2-billion maximum value for the integer.

Thanks Ken,
For this application (thankfully) the highest value I will ever need is around 1000 (with my decimals staying in the tenths), so I wont be anywhere near the max range of even the integer element.

I simply set up a temporary (N) address to hold my accumulated values then added one instruction to divide that result by 10 and sent that result to my original Float address and its done!

I'll keep your suggestion in mind for the future... Thanks a lot!