timer accuracy AB

[thedeester1]

Firstly Ive not been using this page for long and this is my first post. Hello all and thankyou for the answers ive found here upto yet.

My question regards timer accuracy in allen bradley PLC's. Ive written a program to test a valve array. the program times the time each valve takes to open, how long its open and then how long it takes to close. Firstly the program it written on an SLC 5/03. I dont think it will be fast enough to time the events since it doesnt seem to accept a time base of less than 0.01s. Is this down to rs logix or the CPU limitations? Secondly if its the CPU do the compact logix CPU's have a 0.001s time base?
Ive used the SLC mainly to develop the program since theres boatloads lying around where i work. To use a different CPU would mean spending money and i need to know its right before i shell out on a new system. Ive tried reading the selection guides but they rarely quote timer bases.

Appart from this im worried that scan times may affect accuracy?
The program isnt that long (300ish rungs) but does include a large number of timers moving values into interger's which are then compared to alarm values testing for slow or non operation of the valves. Any input to this would be helpfull since i dont wanna waste any more time or money if its an impossible cause?

 

[mordred]

to anwer your first question its a limitation on the SLC I do not know compact so cannot help you there but 300 rungs shouldn't cause any problems with your scan times interferring for your timing I beleive you may have to look at CPU's that support high speed timers I'll check the AB products that I have the manuals for I know several Omron processors support them but not sure on the SLC product line

 

[mordred]

NO go from the looks of it on the SLC line accuracy is 0.01 +/-0 up to scan times of 2.5 seconds

 

[mordred]

looks like I could be wrong' What model of SLC 503 are you using as the C series allows me to program a time base of 0.001 sec???

 

[LJBMatt]

For compactlogix and controllogix controllers the time base on all timers is 1ms or .001s.

 

[kamenges]

Timer timebase resolution is just the first of your worries. How fast are your outputs? How fast are your inputs? As you already said, how fast is your scan? You can only resolve to the slowest of any of these. I really doubt that an SLC will scan 300 rungs in less than 1 ms. In fact, I don't think I've ever seen an empty SLC run that fast.


Even the CompactLogix may not work for you. I think the RPI is fixed at 2msec over the backplane. If you decide to go with the CompactLogix make sure you use the high speed input modules and put the timers in a periodic task so you can ensure the scan rate.

Keith

 

[mordred]

I agree with everything you've said thus far Kamenges however I do have one question on this subject. According to the SLC instruction set manual
on the fixed and SLC 501 processors the scan times do not have an effect unless they exceed 2.5 sec is this in fact true or are their other factors I'm missing?

 

[kamenges]

Look through this thread. I linked you into post 15 because it is a pretty good explanation of what an AB timer really is.


The 2.5 second thing has to do with how an AB timer keeps a record of the last time the timer instructions was serviced. This is an 8-bit value with a 10 millisecond basis. The actual value is every 2.55 seconds. If you don't update the timer instruction at least this fast you will exceed the range of the time from the previous call and you will 'lose' 2.55 seconds of time.

The way to think about it is that the scan time has no effect on the absolute time value of the timer. The scan time does, however, have an effect on when that absolute time is compared to the preset. As an example, assume a 30 millisecond scan time and a 2 second preset. Further assume that the timer is serviced at accumulated time 1.99 seconds. The timer will not interrupt the program scan when it is at exactly 2 seconds. In fact , it doesn't know it is at 2 seconds until the timer instruction is scanned again. At this point the timer accumulated value will be 2.02 seconds.

Keith

 

[thedeester1]

Thanks for the Answers. Another day of reading it seems. Firstly the processor is a 5/03 ser D revision 4. This is just what i borrowed from work to blow a program into and run it to debug (dont have emmulate). Someone said it might be the RSlogix software (timer base problem). This is a project ive taken on my self and the help ive recieved from work is minimal. I was given an old laptop with logix500 ver 5.xx that belonged to someone (no-one knows who) who has left the company. As such I have no way of updating the software. With regards to the speed of outputs / inputs speed i cant see a problem as long as they are a constant. the valves should react within a set time. I can set the alarm value to whatever i want to show slow operation. Our current programmers are involved in a large scale modernisation of the plant and as such, I as a lowly electrician have taken on the project. My boss says if i can prove it to work he will support me but otherwise he wont want a budget hit on a white elephant.

The main aim of the project is to show valve operation at a pressure switch. From the information i have gathered when one is slow its visible to the nakid eye when the process is watched by an operator. Its an ongoing problem that repairing the valve array is difficult since its slow operation and not none operation that causes the problem in the process. At present a man connects a supply to the coils and sees if air comes out. The arrays are then fitted only to be removed 2 hours later becuase production complain of a slow process event.

I suppose its gonna be a case of show and tell. Show what I have done so far and see if anyone wants to take it further.

 

[mordred]

good informative link Kamanges Its obviously something one should be aware of thanks for the info I guess one could sum it all up by saying the timer(not timer) is only as accurate as your ability to poll the timer (not timer)

 

[NOP]

You might want to try using the free-running clock (s:4) rather than timers, I think that this updates at 100uS on all revisions of CPU. The issue with variability of scan time affecting accuracy still remains though.