I have been given the task of programming the PLC code for a label rewinding machine. They want to detect if a label is missing during the rewinding process. The machine is capable of running up to 1000 ft/minute. I am using a Banner and/or Keyence label sensor. The sensor is working OK, I am getting pulses into a register. But I am having a terrible time manipulating the pulse data to detect when a label is missing. The label are 2- 1/2" by 2- 1/2" with a 1/8" gap between each label. They want to stop the machine if a label is missing, that's not a problem. I have been trying to compare two slices of time, but the two time slices are usually the same, even if labels are not present. Any help would be appreciated.

 

You did not indicate which sensor (voltage etc.) but AC sensors can only count about a dozer signals per second but DC sensors can generally count about 1000 per second. There are high speed input cards for counting how many counts since the last time it was scanned. Simply taking an input into a counter fuction block will not give the number of counts as the inputs are looked at at the beginning of the scan, the logic is solved then the outputs are written to. High speed inputs hold the value and it is continuously updated. This may solve your problem of working at low speed but not at high speed. The website you want is http://www.modicon.com

 

Quite a few ways to do this. One is a classical missing pulse detector. That is a one shot that is reset continuously by the label pulses but will time out if not reset when a label is missing. The problem with this is that the time interval set neccesarily implies a range of label speed with too fast missing gaps and too slow constantly triggering. If the labels are pinfeed you can use coincidence methods. For example, every nth sprocket hole sensed should coincide with a label pulse. This is easily done in logic. In fact this will work with any reference that is an integral multiple of the label pulse. An even simpler method is to look at two places for labels. If one pulse is there and the other isn't, obviously there's a label missing. This can be strengthened if you can get multiple scans during a pulse so you can ignore the edges where you might catch small timing differences or wrinkles. If you want to use expensive fast counter modules, the last scheme could be done by simply using two counters and stopping on a miscompare keeping in mind thay will be 1 or more counts apart. I can think of several more of hand and would be happy to help with circuitry or logic as I'm bored to death at the moment.
Hope this helps

Regards

cww

 

If you provide your proper name, you are more likely to get a response as some people here are reluctant to discuss a subject with someone if they don't know who they are. I occassionally see messages here from people I do business with, and I don't want to unknowingly end up conducting private business discussions in public.

However, I think it is safe to address this subject.

1000 ft/min = 200 inches/sec
1/8" = 0.125 inches (gap if label present)
2- 12" + 1/8" + 1/8" = 2.75 inches (gap if label missing)

(0.125 inches) / (200 inch/sec) = 0.000625 seconds (time if label present) (2.75 inches) / (200 inch/sec) = 0.01375 seconds (time if label
missing)

I don't know which label sensors you are using, but these devices often have a "pulse stretcher" option to extend the registration pulse out long enough for slow devices (e.g. - almost anything which uses software) to see it. This is also typical for "ring" sensors which detect rivets or screws in blow tubes. You are likely seeing the same length pulse for both cases because you are seeing the "pulse stretcher" output.

If I were registering labels in this application, I would prefer to use two sensors spaced a set distance apart. If both sensors were on simultaneously, then a label would be missing. This would not be sensitive to differences in speed, only differences in label size (something which is much easier to set up and trouble shoot).

However, you would have to be careful that the "pulse stretch" output (or even just the physical charaterisitics of the raw sensor itself) do not cause the pulses to overlap at high speed when they shouldn't. What you may need is access to the raw signals, and some special interface circuitry which will "and" the signals for you before going through a "pulse stretcher".


Another alternative is to find or build a sensor which passes the light through the label twice a set distance apart. This would require either mirrors, prisms, or fiber optic, and a sensor which has enough excess gain to penetrate the label backing twice, but not the label itself. It would also require a sensor which is fast enough to turn on in the time available (this will depend upon how far apart the sensing points are - but this shouldn't be difficult).

If this sensor ever goes on, a label is missing. You will have to test the sensor for function regularly as it will only go on when the condition is "bad" (the label is missing).

You may also still need a "pulse stretcher" to make sure your PLC (or whatever you are using) will see the "bad" signal from the sensor. This can typically be provided by an off-delay timing relay which has a solid state output.


--

************************
Michael Griffin
London, Ont. Canada
************************