CRC 16 in SCL language

smiffy

This is generating the CRC by calculating it at runtime, which is very inefficient.

This is true, but in this instance being able to document the function is more important than having the function optimized. The CRC calculation will be part of Safety Related part of control system, and thus the need to document the code. Easier to document the short calculations than the content of the datablock full of values.

I'm still unclear about the reasons why the SCL code (not the code using table) appears to loop nine times as opposed to eight times. for j := 0 to 8 do

Edit: Clarified which code

I have now converted to SCL and tested the function provided in the first page, (the short but CPU heavy function, the one not using tables) and it does not work. One glaring bug is that there is a comparison "if ((crc or $0001)<>0) then..." This will always be <> 0, as the $0001 constant brings the <>0 to the comparison.

Even when I tested different variations on that comparison, I did not manage to get the code to work.

I compared the results to Lambert online calculator and an existing code I have that works.

https://www.lammertbies.nl/comm/info/crc-calculation.html

Attached is a test project with the two CRC calculations side by side.

In short: I do have a CRC function that works, but I want to understand how come the Delphi code could have ever worked? Is it intended to work for some other CRC and not for the selected polynomial? I'm using CRC-16-IBM (modbus) polynomial. (https://en.wikipedia.org/wiki/Cyclic_redundancy_check)

Edit: I found this topic useful when I created the working CRC calculation on my own:

https://support.industry.siemens.com/tf/ww/en/posts/ccitt-crc-calculation-in-s7-1200/51128?page=0&pageSize=10

It includes two manuals that describe methods for calculating the modbus CRC, and it also performs the loop only 8 times, not 9 times.

Edit2: The source of the manual I mention above: http://www.modbus.org/specs.php

smiffy

The lookup table will use considerably less CPU time as a lot of the work has been done already in the table data. The only drawback is that you will need to use a bit more program memory to store the table data. It is a trade off between CPU time and memory, between the 2 methods.

Table - Low CPU time but more program memory required.
Calculaton from scratch - More CPU time but less program memory required.

Generally in todays modern PLC's, memory is a lot cheaper to buy than CPU time. So it makes sense to use the table method.

The choice is yours.

 Hello Smiffy
I can't get it to work on TIA Portal V14. Can you please help me?

A.Remon

 

smiffy

The lookup table will use considerably less CPU time as a lot of the work has been done already in the table data. The only drawback is that you will need to use a bit more program memory to store the table data. It is a trade off between CPU time and memory, between the 2 methods.

Table - Low CPU time but more program memory required.
Calculaton from scratch - More CPU time but less program memory required.

Generally in todays modern PLC's, memory is a lot cheaper to buy than CPU time. So it makes sense to use the table method.

The choice is yours.

 Hello Smiffy
I can't get it to work on TIA Portal V14. Can you please help me?

Hello again, fellas.
I found a very good CRC Calculator for Modbus 16

http://www.tansoo.cn/?p=969 with polynomial
y
http://www.tansoo.cn/?p=976 with table
I hope that it will help you.
It's in Chinese, but it translates easily

If you want the whole story behind the CRC calculation and want to document the code yourself, it's better to go to the source. Attached is the diagram from the document I linked earlier in this topic showing what must be done to calculate the CRC.

Hi Smiffy 

Many thanks for sharing this code. One question. I guess the tables are made for one value of the polynomial. How would it be if I need to use polynomial 0x8005  ( CRC_16 Modbus).  

Many thanks!! 

smiffy

Err, I have just noticed a boo-boo in the post  made. The polynomial is 8009 not A009.
So are you after the Modbus CRC? If so use code as supplied.
If you are after the CRC-16/ARC CRC them simply use the same table but set the initla values of CRCHi and CRCLo to 0.

 Hello Smiffy

I saw you saying that the polynomial is 8009, is that correct? I think you mean 8005. I used a function with a table/8005 to generate a crc for the following data:

device no: 0x01

function: 0x04

start address: 0xF7

no of registers: 1C

the CRC I got is 1FB0, so my complete frame will be like this (hex bytes): 01 04 00 F7 00 1C B0 1F

I have 313C CPU with CP340 card, the setting for the card can be seen in the attached screenshot, I tried to use Send_P2P and P_Send functions to send my frame to the slave, for both I didn't succeed.

In case of I use SEND_P2P, when I assign the DB name to the buffer input I got the following error code:

16#8328 (BUFFER is a pointer to BOOL with a number of bits not equal to n * 8)

and when I change it and point to the first byte in the DB and assign 8 to the length input, I got the following error: 

I tried to change the length but I still have same error.

In case of I use P_SEND,  I put the DB number and specify the length as 8, but I dont get anything.

In both cases I dont see the TXD led blinking, so no send at all. I also tried to swap the CRC bytes but I got the same results. 

BTW, I used the (dongles-less software supplied by TIBI68) that you suggested to me in another post, it works with the settings shown in the attached screenshot, however, I want to set up mine to have better understanding of how the modbus working etc. 

So, what am I doing wrong here? I use TIA V13.

kind regards

Ebrahem