Monitor production. posted by on April 26, 2014

Load balancing and finding bottlenecks requires correct information.
I sometimes hear people saying things like “Why we have only 50% output when we have only 10% downtime in this line?” …needless to say ofcourse the (manual?) downtime records only include longer stops and the small ones goes without notice.

It is really easy to include basic recording in the PLC programs. Cycle time, accumulated downtime based on cycle time longer than xx seconds, cycle counters, event counters etc.
You write this function block one time and include it in every plc program you write.
You can easily retrieve this data every minute or so over the network from a server using any of the old simple protocols like ModbusTCP and dump the data in a database. Use a Web frame work (for ex Drupal) and some graphical libs (for ex Flot) to present the data from the database and you can also have routines that serves raw data tables, that you can retrieve from any spreadsheet software on the client side.

Worth noting when ot comes to making your production go full speed are cheap vision systems.

Example: Raspberry Pi and it’ s camera. Why? Because normally a vision system cost so much you need a budget even for just a single system and you need training for the software and still you don’t really know what it is doing.

When you have a system that costs just 1-2% of what an industrial vision system cost, it is no longer about doing the same things cheaper. You start to use it in a different way.

You can use these systems to look for problems during the production cycles and you can use Ethernet to talk to PLCs you can alarm for problems and you can also save the pictures from the problematic events reaching them from a web server. OpenCV can do the image processing for you either from C(++) or Python. LibModbus is your friend.
And yes you can adjust the focus on the raspi-cam, you just have to remove the glue that prevents the lens from rotating. I can easily go down to 50mm working distance.

Leave a Reply

Your email address will not be published. Required fields are marked *

By submitting this form, you accept the Mollom privacy policy.