S7~OB35 Interupt Problems

[SimonGoldsworthy]

I've seen scans increase where I/O was missing.

I have seen this too. I normally use a 317 and running the program in the CPU without the remote i/o connected does not significantly change the scan time. On the occasions where I have used a 315 it is a different story as the scan time increases dramatically when the error OB's are processed. Note that from the diagnostic buffer entries posted by STL there is no mention of any rack failures or I/O access failures so it appears the I/O is OK.

I use the error OB's so the CPU doesn't stop but I always flag up the various OB faults via the HMI.

 

[[FIN]Pete]

Hi
This doesn't help You find the memory "eater" but this might help you decreasing the cycle time.
I don't know Your process and how "big" the PLC program is but because the program cycle time varies so much I'am questioning the program structure / phasing of the block calls.
1) On OB35 -> Do You have some functions, what the program calls every 100ms cycle even it is not necessary -> like analog measurements, PIDs ...?
2) Is there some huge calculation functions and what is the purpose? If they generate "some general" information only for the scada system -> the block call once per second is fast enough.
3) Even in OB1 -> is some part of the machine program what you can call every other program cycle??
* If the program includes alot of indirect addressing, address register handling, any-variables and loops -> these "kills" the small CPUs quite easily.

-Pete-

 

[PeterW]

Fun part is it is 100% STL

Regards

Don't find STL a problem, prefer ladder for main part of program if only for maintenance but subroutines where the real work is done is better in STL.

Where I feel sorry for you is simply STL needs to have good documentation to follow it, as it in German!!!

Where I'm working at the moment, one of the systems is provided from Germany, nobody told them the code is supposed to be in English!! its all STL about half know how protected (easy to remove BTW) and all over complicated with indirect addressing (actually set some pointers in datawords to I/O i.e. DW#16#8100034B) and using UC instructions for FB's that need instance DB's, all clever stuff but leaves you asking the question WHY??

Glad I'm not going to be the maintenance guy on this stuff.

One of the German engineers actually (proudly) told me that when a new engineer is employed he has to train for 3 years on their systems software before being allowed to make changes.

Why do companies do this?? just not clever to me at all.

 

[geniusintraining]

Glad I'm not going to be the maintenance guy on this stuff.

I am the guy maintaining it...it sucks, getting better on the STL, but the German terminology, is tuff at times, my programs are half German and half broken English...flashlight=lamp, lifting machine=lift table, pusher=puller...

Why do companies do this?? just not clever to me at all.

I have no idea, but it does make us better in the end

I did just order a copy of PG-2000 it will convert STL to ladder (most, not all), I wish I had it a few years ago, I have only used it with the S5's but it's good

 

[PeterW]

I've translated bits using asta vista's babel fish, one word came out as Infiltrator, had me going for days that, what the hell was an infiltrator??

In the end I realside it was Inductor

 

[STL???]

Update

Hi Guys,

I found the problem quite quickly once i had typed in the translated text. A block called positioning, which was one of the calls inside OB35 was the culprit!

Although the station positioning generally worked quite well, If it over/undershot, it would reverse/forward to sort itself out, however the start speed sent to the inverter was to high, so it would never achieve an in tolerance position.

The code for this, gives the direction/release command and jumps over the other direction which caused the loop to keep going over and over when an acceptable figure couldnt be achieved.

I have lowered this value and everything is working well and the scantime (although high) is well within the 150ms default limit.

Interestingly all the Inverter settings can be altered from the PC with "Admininstrator" privileges, No one knows the password apparently so i changed it from within the block itself.

Once again thanks for the assistance/ideas you gave me!

Regards

STL