Counting / shifting problem

Parts enter a chamber through door A,=20
exit through door B.
Parts are moved from A to B on a single
speed belt.

Geometry of the part is such that it normally travels from A to B in about 1.6 seconds, and passes a fixed point in about 0.080 seconds.

End user desires to know if:
"a part goes in, but does not come out"

(gets stuck, falls off belt)

Obviously, a PLC is out of the question, since
this end user has '$tandardized on Allen Bradley",
and will balk if I suggest the best choice (Keyance)

I'm thinking count-in, count-out, and if I've counted in, and the count_total does not change for 1.600001 seconds, that equals 'stuck/jammed'

On the other hand, I'm thinking that there is a
packaged hardware solution out there that guys
in the packaging / pharm. businesses use all the time.

Well, what is it?

Jeffrey D. Brandt
[email protected]
http://users.supernet.com/jdbrandt/

 

> Obviously, a PLC is out of the question, since
> this end user has '$tandardized on Allen Bradley",

Why have you ruled out AB PLC's? Have you checked
out the MicroLogix 1000? It has high speed inputs
built-in. You would have no problem solving your
application with this product, or several other
equivalent PLC brands.
Mark

 

It isn't obvious why a PLC is out of the question. This is a simple problem. Figure out how many parts can be inside the process based on line speed, etc. Throw a SLC or MicroLogix with 2 high speed counter inputs watching parts in and parts out. Have the 'IN' sensor increment the count, and the 'out' sensor decrement. If the
value of the counter is ever higher than the max parts that can be inside, trigger your alarm.

By using this, you can also pick up first part/last part problems, which you would miss using the method you describe below. Also, at startup, you would nuisance alarm on the first couple parts until your machine was full.

This should work, unless there is something else that you aren't telling us that would prevent a PLC from doing this. In fact, you may not even need the High Speed Counter inputs. 80 Msec
should be long enough to see the part. You could also use tricks like Immediate hardware interrupt on the input, or, using fiber optic style sensors, angle the emitter and receiver so that you can stretch your beam interrupt time if you experience missed signals. I really think that 80Msec is long enough though. You may just
need to dedicate a small controller to the task. (MicroLogix 1000)

--Joe

 

The above "ring counter" is a great solution, I have used it for the same issue with pc boards running through a reflow oven. Be sure to write some error recovery logic and cover the photo-eyes to avoid accidental increment and decrement by shirtsleeves, labcoats, curious george types etc...
I agree there should not be a need for a high speed counter. A dedicated micrologix should scan well below 20ms, which is 2x the resolution you need (40ms) for an 80ms pulse rate.