Not Sensing versus No Sensor

[dmargineau]

I would strongly recommend replacing the photoeye type; most of the major platforms provide optical sensors with both functionalities built-in.

It is a classical application of detecting presence/absence while ensuring detection availability without laborious and expensive "fixes".

 

[Jsu0234m]

I've had a problem like this in the past and what i did was create a startup screen on the HMI that forces the operators each morning to flag the photoeyes before the line can be started.

I created a HMI screen that had the locations of the photoeyes around 20 and before the machine could be started they all had to be flagged. Once they were flagged, the system could be started. It worked well and it also forced operators and maintenance to get familiar with the location of all the photoeyes.

 

[slackjaw]

I would strongly recommend replacing the photoeye type; most of the major platforms provide optical sensors with both functionalities built-in.

It is a classical application of detecting presence/absence while ensuring detection availability without laborious and expensive "fixes".

Thanks. I totally agree. Sadly, management and QA seem to REALLY not like this idea...

Frankly, I am surprised that anyone would use a "Dark On" sensor, but that is what they have, and what has been qualified for use. Though maybe there are pitfalls that I have not yet thought of...

 

[Doug-P]

However, the line is periodically dismantled and then reassembled for maintenance activities. Recently, the line was reassembled, but the photoelectric sensor was not plugged in (yes, I am serious).

I'll make a big assumption here since I never dealt with critical stuff like pharma but, could you not add to the restart procedure/checklist an item stating 'sensor X' will be checked for proper operation before bringing the line back into operation? It really could be as simple as someone flagging the sensor while another monitors the input module.

 

[dmargineau]

Though maybe there are pitfalls that I have not yet thought of...

...such as: The system will keep operating when any (or all) of the below listed scenarios are true:

Defective/Damaged/Misaligned/Covered Emitter

Defective/Damaged/Misaligned/Covered Receiver

Defective/Damaged Emitter electrical circuitry

Defective/Damaged Receiver electrical circuitry

Faulty system Input point.

By switching to Light Operated sensing you will ensure the detection station functionality at all times.

Relate these thoughts to Upper Management and especially to QA Lab...

 

[slackjaw]

I'll make a big assumption here since I never dealt with critical stuff like pharma but, could you not add to the restart procedure/checklist an item stating 'sensor X' will be checked for proper operation before bringing the line back into operation? It really could be as simple as someone flagging the sensor while another monitors the input module.

Due to mechanical limitations, there is no way to trip the photo sensor. Otherwise, yes, that is what we currently do - an operator has to verify that the little green "on" light is lit.

The trouble is that our startup checklist continues to grow and grow. We almost literally spend more time verifying the checklist than we do producing tablets. I have been tasked with automating as many of these checklist items as possible, with as little paperwork as possible.

 

[slackjaw]

...such as: The system will keep operating when any (or all) of the below listed scenarios are true:

Defective/Damaged/Misaligned/Covered Emitter

Defective/Damaged/Misaligned/Covered Receiver

Defective/Damaged Emitter electrical circuitry

Defective/Damaged Receiver electrical circuitry

Faulty system Input point.

By switching to Light Operated sensing you will ensure the detection station functionality at all times.

Relate these thoughts to Upper Management and especially to QA Lab...

Ha! Yes - I may have to circle back around and have this discussion with them again. It seems to me that using a "Light On" sensor would essentially be a fail-safe design. If at startup there is no signal, then either the sensor is malfunctioning, or there is an obstruction in the line. Of course, perhaps this sensor provides better sensing than it's opposite counterpart.

But what I meant to say was, I may not have considered all of the potential problems that a "Light On" sensor might present.

 

[Dravik]

Mm, We typically like to use NC sensors and use edge detection inside the logic. A product passing by the sensor should have a falling edge transition followed by a rising edge transition in some reasonable time frame.
The really critical stuff uses redundant sensors, dual output sensors, etc depending on the risk level.

 

[keithkyll]

Think of how your car computer knows when a valve (solenoid) is unplugged. It uses NPN logic to drive it. Senses the low side of the coil. When it's off and plugged in, there should be 12 Volts on the return line. Unplugged is zero.
I expect the current draw to be low on your sensor. Add a relay contact to break the power to the negative side. When the relay is off, see if the current draw is enough to (reliably) trip a PLC input. Try a high speed input - it should be more sensitive.
If that doesn't work, then use an analog input. You may need a resistor divider to limit voltage. A 9.1 Volt Zener should be enough.

At powerup, test for connection, then turn on relay to power sensor. You could do this test periodically. It would only take a second or two.

 

[papejo]

im a kind of new to the plc world, is it possible to use a t off timer, as soon as you start the machine the t off start and when the sensor detects , the sensor reset the t off , if the sensor dosent send the signal then the machine stops and a buzzer gets activated.

correct me if i am wrong ,please

 

[slackjaw]

Mm, We typically like to use NC sensors and use edge detection inside the logic. A product passing by the sensor should have a falling edge transition followed by a rising edge transition in some reasonable time frame.
The really critical stuff uses redundant sensors, dual output sensors, etc depending on the risk level.

Yeah, the existing logic already DOES utilize edge detection and a timer. The issue is that the transit time is variable enough that the delay timer does not reliably work. It is currently set for one minute.

What is needed is a verification of the sensor at build time, or at the very least, at line initialization.

The line can take a matter of days or weeks to be dismantled and then rebuilt. Once rebuilt, the line can sit idle for hours and usually days. So, a timer against either of these milestones is not feasible.

I am reaching out to the sensor vendor as well as the original system vendor to discuss the option of replacing the sensor with a "Light On" type.

Thanks again!

 

[Keystone]

If your model of sensor has both a NO & NC output you could wire the unused output to a PLC Input and do a check to make sure at least one of the contact are high. (See or No See).
If the photo eye is unplugged you would not have either signal.

 

[slackjaw]

Think of how your car computer knows when a valve (solenoid) is unplugged. It uses NPN logic to drive it. Senses the low side of the coil. When it's off and plugged in, there should be 12 Volts on the return line. Unplugged is zero.
I expect the current draw to be low on your sensor. Add a relay contact to break the power to the negative side. When the relay is off, see if the current draw is enough to (reliably) trip a PLC input. Try a high speed input - it should be more sensitive.
If that doesn't work, then use an analog input. You may need a resistor divider to limit voltage. A 9.1 Volt Zener should be enough.

At powerup, test for connection, then turn on relay to power sensor. You could do this test periodically. It would only take a second or two.

Thanks!

This was more along the lines of what I thought I might be required to do - it just felt a bit too unorthodox, and I was worried that I was going to wind up with some unworkable contraption.

 

[slackjaw]

im a kind of new to the plc world, is it possible to use a t off timer, as soon as you start the machine the t off start and when the sensor detects , the sensor reset the t off , if the sensor dosent send the signal then the machine stops and a buzzer gets activated.

correct me if i am wrong ,please

No, you are correct. While the line is in operation, there is a timer that is enabled to verify that product arrives at the sensor within a reasonable time. The problem is that the transit times are not reliable or reasonably repeatable.

Also, once we have the alarm, we are facing a mountain of paperwork to unwind the issue, and the loss of many tens of thousands of dollars of drug product.

 

[slackjaw]

If your model of sensor has both a NO & NC output you could wire the unused output to a PLC Input and do a check to make sure at least one of the contact are high. (See or No See).
If the photo eye is unplugged you would not have either signal.

Yes, when I first heard that the line had been assembled without the sensor, this was what I had hoped to do. Sadly, the sensor is not configurable, and has only a single "Dark On" output.

I can only sense when a tablet passes the sensor. And there is no means to obstruct the sensor at build time - it is entirely inaccessible.