Floating point in FX1N

Hi all. Although the FX1N allows floating point instructions, the one that's missing is a means of converting a value to an integer. Has anyone found a workaround for this one? Could I maybe just use the first word of the pair, or is that just plain stupid? Thanks.

Floating Point in FX1N

According to the programming manual the FX1N doesn't support floating point functions at all. If you try and use one of the 'E' commands (FNC110 to 129) in GX Dev it doesn't allow you to.

Thanks, Michael. I'll have to find another way...

Floating Point in FX1N

Try posting your problem. There will be someone here that has come across it before.

DairyBoy said:

Could I maybe just use the first word of the pair...

Click to expand...

Even with FP support, using the first word of the pair will not give you a number that remotely resembles what you want. A 32 bit floating point number consists of a sign bit, followed by 8 bits that represent the exponent normalized to 127, and then 23 bits that contain the mantissa of the float.

For example, if we take a look at the value of PI in a single precision float we would see that the high order 16 bit word has the integer 16457 and that the low order word has the integer 4058. Neither one remotely resembles 3.

If all you want to do is a comparrison you can actually get away with a bitwise comparrisson between two sets of words storing the floats, but anyting else is going to be more tricky.