Split long word for average function in function block

[paulusB]

I'm building a function block to sum flow and production hous and calculate average flow based on a REAL input.

For avering I use /L function.
The Result R is a LWORD

To get a more acurate result I want to convert the result to a REAL.

The first two words of the LWORD is the Quatient in DINT
The last two words of the LWORD is the Remainder in DINT

How do I read out the remainder in a function block.

A bit shift is nog working (NASR/NSRL) is only availabe to 32 bits,'.

When not in a function block and using de D channel you simply pick the 3rd channel

LWORD = D100

D100 = DINT for QuatientA
D102 = DINT for remainder.

m3_test.opt

m3_test.bak

m3_test.cxp

 

Edited May 4, 2020 by paulusB
attached function block

[paulusB]

Figured it out. You have to work with arrays and move them (as discribed in the help )

[Crossbow]

Why do math on integers and then convert to real?  That's not accurate.  Convert the numbers to real before doing to math so decimals are not rounded off.

[paulusB]

Because the total amounts can become big and real then les accurate

[banker]

Why not 64 bit real ?

[paulusB]

Thanks for the replies. I thought 64 bit real math wasn't available in CX-programmer. But it is.

Edited May 6, 2020 by paulusB
but help says only 32 bits...?

[Crossbow]

Those are 32-bit real number commands.  I think only NJ/NX have 64-bit floating point.

[Michael Walsh]

I see you are using a CJ2M PLC.  I have shown below how to use double real numbers in CXP with a CJ2M PLC.

Symbol creation:

Choose LREAL as the data type.  Be aware that this data type uses 4 consecutive 16 bit addresses.  So in this example, the variable would use D0, D1, D2 and D3.

Here are some of the double floating point instructions:

[BobB]

CJ2 has long real so does the CP1L.

[paulusB]

I totally missed the "D" instructions. Thanks

[Crossbow]

My bad... I am way behind on my knowledge of the legacy platforms, pretty much a Sysmac guy now...