Which instr is better ADD or CTU in SLC 5/05

[AB2005]

Dears

I have been programming with SLC5/05. Here is an input which freq is almost 5-8c/s and connected with PLC input card 1747-IB32. For count up, which instruction is better, ADD or CTU? If I program the both instructions, i found a difference between accumulated value of both.

 

[rdrast]

Do you use a one-shot prior to the ADD instruction? The CTU has a built-in one-shot on it's count input.

What is the discrepancy? Which count's more? Is the difference consistent?

 

[Ron Beaufort]

as my distinguished colleague rdrast has mentioned, there is a built-in "one shot" function for the CTU - but there can be a difference in the operation of that built-in "one shot" and the operation of a regular OSR "one shot" particularly when the processor first goes into the Run mode ...

also ... how high are you counting? - and how - and at what point - do you RESET each of the two "counters" ? ... also ... in the ADD type counter are you using an N-type integer - or an F-type floating point location to hold the value? ...

if you'll answer these questions - and those from rdrast - we should be able help you sort this out ... and most important - can you post your RSS program file? ...

 

[AB2005]

1)[FONT=&quot] [/FONT]Yes, I am using OSR inst in series of ADD.
2)[FONT=&quot] [/FONT]The input address is same but the accumulated value in ADD always increases time by time than in CTU.
3)[FONT=&quot] [/FONT]The max frequency of input is 26HZ.
4)[FONT=&quot] [/FONT]I am using F integer in ADD (ADD 0.1247 F8:0 F8:0) which accumulates max 4millions and resets after month end.
5)[FONT=&quot] [/FONT]In CTU (say C5:0) Instruction, after 30000 counts, counter resets and Add 1 in an Integer. Normally, after 2.5sec time interval I calculate as (CPT F8:1 ((N7:0 * 30000) + C5:0.ACC)) * 0.1247). I found F8:0 of ADD > F8:1 of CPT.

Hope this info help you.

 

[OkiePC]

Since you are adding floats, then the total is accumulating rounding errors, therefore your second method of adding integers and doing the scaling to a float at the end of the calculation will yield superior results.

It isn't an issue of CTU vs OSR+ADD, rather an issue of floating point accuracy that is making the difference in your case.

 

[Ron Beaufort]

yes, my distinguished colleague OkiePC has nailed why the numbers come out "different" ...

if you were to use the ADD to "add one" each time it "counts" and THEN/AFTERWARDS do the math calculation on that accumulated value, then you should get the same results with the ADD arrangement as with the CTU ... but - just remember that with floating point math in a PLC, there are usually some rounding off errors - so that EITHER WAY the final mathematical value that you're calculating/computing might not always be precisely "correct" ...

in other words, NEITHER of the methods you've listed might be 100% correct - but they might still be "close enough" to work in your application ...

 

[AB2005]

Thanks,
I have understood the real cause.