Contrologix RTOS accuracy

[curlyandshemp]

Is there anyway to improve the accuracy of the Real To String conversion instruction in contrologix?

using the instruction for a test:

Even the online help shows the inaccuracy of the instruction :

I have to print scaled data from a mass flowmeter scaled as S.G. and send to a serial printer. Since I am dealing with S.G. the range of my numbers will be 0.999 to 1.999, so I can limit the length of the converted string to 5 characters

 

[Oakley]

Most likely it is due to how real numbers are represented in binary.


Can you utilize building the string using DTOS instructions?

 

[curlyandshemp]

Most likely it is due to how real numbers are represented in binary.


Can you utilize building the string using DTOS instructions?

I was thinking doing that, taking the REAL # X.XXX multiply by 1000, store as a DINT ,then convert DINT to string and insert a '.' after the first digit. But then I would have to ensure the number is not 0.XXX

seems like a lot of work

 

[jimbo3123]

I'd have to think for a bit about the exact details, but I think you're on the right track.

I was thinking doing that, taking the REAL # X.XXX multiply by 1000, store as a DINT ,then convert DINT to string and insert a '.' after the first digit. But then I would have to ensure the number is not 0.XXX

seems like a lot of work

 

[Contr_Conn]

I was going to offer DINT solution right before I saw your later post.
For many years I used DINT and INT to display numbers with fixed decimal point on HMIs, particularlyfor motion systems where 2 or 3 decimals used:
same idea - take Real number, multiply by 1000 (or 100 if only 2 decimals) and send it to HMI. Most of HMI will let you specify where to put decimal point.

In your case to compensate for 0.xxx use after decimal inserted:
if STR.LEN = 4 CONCAT (ZEROSTR STR STR)
ZEROSTR will have string value of "0" - CONCAT can't use constants

 

[curlyandshemp]

I was going to offer DINT solution right before I saw your later post.
For many years I used DINT and INT to display numbers with fixed decimal point on HMIs, particularlyfor motion systems where 2 or 3 decimals used:
same idea - take Real number, multiply by 1000 (or 100 if only 2 decimals) and send it to HMI. Most of HMI will let you specify where to put decimal point.

In your case to compensate for 0.xxx use after decimal inserted:
if STR.LEN = 4 CONCAT (ZEROSTR STR STR)
ZEROSTR will have string value of "0" - CONCAT can't use constants

Thanks

Kinda sucks though you have to do this considering we are in the 21st Century. Maybe Rockwell needs to consider a Double Precision Floating Point number such as what Siemens had with the S5-135u PLCs in the early '90s

Thanks for answering considering you must be stuffed with turkey

 

[Contr_Conn]

Thanks for answering considering you must be stuffed with turkey

Yep, huge turkey is gone along with appetizers, cheesecake and other goodies my 200+ pound overweight body doesn't need...

As for math - decimal point rounding has nothing to do with 64 bit math no matter what brand you have.
Sure with 64 bit Reals mantissa will have much better accuracy than 23 bit we have today with 32 bit REAL, but do you really want to see your string as 15.300000000001 ?

Having abilitiy to specify how many decimals you want would be a nice feature...