What was your most difficult process-issue to handle?

This is NOT only for the more experienced programmers.

This is for ALL PROGRAMMERS at ANY level of experience. EVERYONE is invited to identify and dispel their demons!

Who knows... you might have developed a solution that no one ever thought of (and someone might wish they came up with the idea you found) or you might hear a comment regarding how someone else handled the same problem - only better.

I'm looking for unusual problems (unusual from YOUR point of view). I'm looking for those problems that made you wish you knew more about the laws of Physics and the relationships between "cause and effect". What was the method you used to "Beat the BA$TARD"!

An example might be something like "Open-Looped Motion Control" with or without a motion control card. That is, speed or position control using only discrete inputs.

Or,... Auto-Correction based on last-performance WITHOUT an analog indication of current-performance.

Granted, for all of you purists (Yeah, I'm one of them) there are more effective ways to handle a lot of problems with extra hardware... however, things being as they are for some folks, you have to work with what you got!

So please, lay-off on the hardware solution UNLESS that is the ONLY solution!

Scheduling several items (lots) through a linear process (Chemical baths) so that the times could be different for each item started and at each process but it would not get held up waiting for the previous item to finish anywhere down the line.

That's easy...

Dead band, dead time, non-linear systems and sloppy mechanical machines. Just as bad are customers that don't listen.
I have a customer who has sent a machine overseas to be installed. This is a big machine. It has hydraulic cylinders that oscillate or ring in a sustained way even in open loop or manual mode. They wouldn't fix it in the shop. I don't know how it will be fixed there.

Similar to DesertDog

A multiple substrate and finish plating system where, for example, at any time, random loading of the line would affect which pre-plate process to use and which finish thereafter.

When a random system is implemented, it has to be determined, programmatically, if a certain "loaded bar" can be introduced into the system because of the downstream effects upon the other loaded bar's in thier current process' throughout the line. Also, during the process of any other loaded bar we have to keep an "eye" on the whole system to determine if a bar has to move on to the next process without interfering with another process or the ability to be able to unload a bar if completed.

All this with keeping in mind that only one bar (in some space prohibited facilities) can be load and offloaded at one time, at which point, giving the operators/workers enough time to load/unload any give bar with varying parts which require different amounts of time.

Enough to drive you to drink! banghead

Edit... All the above with different control features for similar process like, Amps per Sq. Ft. per bar to control rectifiers, pH control for different parts, time in each process... it goes on and on.

My main process is composed of several high level modules. Each module is a large segment of the over-all process.

When all conditions allow, the entire process runs in automatic. However, anyone of those modules can be taken to a manual mode or a semi-automatic mode.

Each module needs to know the control mode of the previous and following module so that operation in the other modules can continue to be properly co-ordinated. While one module is in manual, the previous and following might still be in automatic.

The most problematical situation is when the operator takes a module to manual mode. In this case, the operator is not required to operate the module in the same way that the PLC would in automatic mode.

This means that the PLC has to be able to "watch" everything that the operator might be doing while in manual. The big "gotcha" is that there are some "blind-spots" in the process. These "blind-spots" are not too problematical when the system is running in auto or semi-auto.

Targets (product and/or product handlers) can be added to or removed from the process by the operators as they see fit. That freedom gives them the ability to be as productive as they can under the particular circumstances.

That makes it better for them... and living hell for me.

Whenever a particular module goes from a more manual mode to a more automatic mode, I have to handle the "blind-spots" first. That is, before actually resuming operation in a higher automatic mode, I have begin slowly to determine the condition of the "blind-spots". The module resumes the more automatic level of control only after the program has been able to make a reasonable determination of the status of the "blind-spots".

While "BEING THE COMPUTER" I have to "watch" everything that occurs and then try to make educated guesses as to what the operator is doing while he is doing anything in manual.

When the operator then gives me (the computer) the OK to go back to a higher level of automatic control... I practice the "Trust, but Verify" philosophy. If I can not make a positive determination then the operator is notified that he must do this or that to bring the module to a "known state". That is, the operator is notified that the program can not determine the state of this or that "blind-spot".

Considering all of the things that an operator might do is a huge task. They always seem to find those things that I have not considered.

K-Map? They're great... but they ain't always easy. Slowly but surely I'm applying K-Maps to all of the lesser modules. They take time and effort.

and they dumped it in my lap

An 8 spindle automatic screw machine with three (3) dimensional detectors that had to index the 'bad' part to the pick off collet and hold it so the operator could manually remove it. The three detectors are all on different spindles so that the part had to be tracked through the indexing process. More than one bad part can be in the system at one time so had to find a way of tracking any or all of them. Thanks to reading a thread here in the forum about bottling lines I visualized the parts as bottles and implemented the A-B BSL and it worked perfectly. I was passed this project from the day shifter who..."didn't have time" to do it, well I understood what that meant. It was great fun and I learned alot from it. Now to tackle the sorting machine that an OEM built for us using cascaded counters to track parts. I figure I can cut the program almost in half. Not a high-tech solution but one I'll remember...thanks for this thread Terry, one I'm sure I'll get many ideas from.

Bob