COP Studio5000 order on a UDT

[maexio]

Hello,
I have been looking for a reference for the order that a UDT is copied in the COP instruction.

More specifically - I have a set of code that used an unprotected COP on an array of UDTS that were changed in a higher priority event task. Due to the size of the array and udts a CPS was not used (to prevent motion issues) and I stupidly didn't think or know of a UID / UIE pair to prevent data integrity issues.

My code now correctly uses a UID / UIE pair before cop of the data, but the symptom of the problem will only be corrected if the UDT data is copied in a specific order, which I can't find anywhere.

More Specifically i have an array of UDT's of the type 'Board' and it has approximately 30 DINT subtags ordered as follows :

ID (Position 0)
Tag 1
Tag 2
Tag 3
Tag 4
...
Tag 20
Tag CriticalCondition
..
Tag 29

If I do a COP Board[5] to TestBoard with a Len 1 -> and an event task interrupts the COP and moves Board[4] -> Board [5] (after moving board [5] -> board[6]...)

Depending upon where the COP is in the interrupt - this could explain my problem. BUT only if the COP copies in the following order :

Tag 29 - Tag 29
Tag 28 - Tag 28
....
Tag 1 (ID) - Tag 1(ID)


That way i could be doing a check that ID NEQ OldID -> MOV ID OldID and have the Tag 21 (CriticalCondition) not be representative of the data i'm trying to 'one shot' with my stored 'OldID' condition.


If however it copies Tag 1 -> Tag 1 then Tag 2 - Tag 2, and so on - My problem still exists and I want to solve it before updating a bunch of locations its used in.



That said -> I know the COP copies data in 32 bit groups, but not the order it copies within a UDT and can't find anywhere on KnowledgeBase or in google searches that clarifies it.

Much thanks in advance, as this issue has me quite distracted trying to find another source of the issue.

 

[baref00t25]

For cases like this I have used a circular queue. Don't move the data, move the pointer of where the data is being used. You can find more details in this post.

https://www.plctalk.net/qanda/showthread.php?t=131077

 

[maexio]

Thanks baref00t25,

I use circular queues as well for some items, but in this case it doesn't work with most of the other handoff and trigger positions as easily. The vast majority of interactions with the array were data safe, just this one wasn't and it now is. The only problem is either the integrity issue corrects another symptom that is very intermittent or it doesn't. As opposed to waiting weeks or months and having it occur or not, I am hoping to stay in the headspace of this issue and verify that it is solved, but the order a udt is copied in affects that.

Thanks

 

[plvlce]

Good question. I know that an array will be be copied starting from the lowest member.

I would assume that since the COP instruction functions over a set of contiguous bits this would be the same regardless of data type, but I admit I don't have any actual evidence or documentation regarding the order it operates on UDTs.

 

[drbitboy]

It should be easy to test, summat like this:

Continuous task, three rungs:

COP baseudt udt 1

GRT udt.first_dint udt.last_dint CTU ctrgrt 32767 0
LES udt.first_dint udt.last_dint CTU ctrles 32767 0
​

High-priority Timed-interrupt (1-5ms?) task, three rungs:

ADD baseudt.first_dint 1 tmp_dint

XIC first MOV tmp_dint baseudt.first_dint MOV tmp_dint baseudt.second_dint ... MOV tmp_dint baseudt.next_to_last_dint MOV tmp_dint baseudt.last_dint

XIO first MOV tmp_dint baseudt.last_dint MOV tmp_dint baseutd.next_to_last_dint ... MOV tmp_dint baseudt.second_dint MOV tmp_dint baseudt.first_dintBND

​

Toggle the first bit, and see which counter increments.

 

[maexio]

Thanks - I"ll try that tomorrow. I called tech support and they inferred that it copies from first tag -> last tag based on 32bit 'buffering' example in :
https://rockwellautomation.custhelp.com/app/answers/answer_view/a_id/52154

So either they are correct, and my 'impossible' situation is still impossible, or they are wrong (<fingers crossed>) and the problem is now solved...