DO820 cyclic switching issue after upgrade from 800xA SV 5.0 SP1 to SV 6.0.1
Good day, everyone. During an upgrade process from an 800xA SV 5.0 SP1 to 6.0.1, a PM861A CPU showed certain anomalies in its behavior; specifically, 95% of all the DO820 that went through Modulebus and had at least a channel in an active state would switch every second said channels to “0” for approx. 200ms and then back to “1”. The valves connected to these outputs were supposed to stay active because of the programmed logics, but due to this issue they started switching off and on intermittently.
This issue started approx. 5 hours after the upgrade process on the controller was completed, and it only affected this specific CPU. After a modification on the hardware configuration in CBM was made (deletion of an unused I/O module), the issue disappeared. Our CHP thought that it might have had something to do with the task time allocation and the CPU load, so they adjusted the delays a bit to avoid any overlapping between them, but once again, after 5 or 6 hours, the same behavior reappeared.
•301-304 DO820; 305 DI810; 306 DO820; 307 AI845; 308 AO810
•401-402 AI810; 403 AO810; 404-406 DI810; 407 DO820; 408 AI810
◦1-4 DI810; 5-8 DO820; 9 AI845; 10 AO810; 11 AI810
◦CPU load: 75% during regular production
◦CPU load: 85%
After multiple logics modifications, due to ValveUni input and output variable changes, CPU load increased to 100%
•Controller shares applications and tasks with a PM864, but taking the brunt of the load (two different applications, the most connected modules)
The only pattern we have observed is that, while running SV 6.0, every 4 to 6 hours the same issue reappears. As said before, only this CPU is presenting this problem. So at the end our CHP had to downgrade the CPU to SV 5.0
One thing they would like to know is that if it would be out of the realm of possibilitiy to think that, having no Modulebus scan cycle time set and seeing as the CPU load gets so high, either the tasks are not ending or there is no time to write to the outputs, so they go to their default (“0” in this case).
The rest of the system (servers, workstations and other controllers) are all running SV6.0.1, it is only this particular one that has had issues.
Please, let me know if any more information is needed. Thank you all in advance.