Which sensor will work best for this?

WBA Consulting

I have an odd issue with a flatbelt conveyor at the end of one the SMT lines at work. The current sensor used to detect objects at the end of the flat belt uses an IR/IR sensor set aimed across the surface of the belt to detect when a board is sitting about 16 inches from the end (indicating one or more boards are backing up on the belt). If a board is present for more than 20 seconds in this position, an alarm is sounded to alert the operators to prevent additional boards from backing up to the reflow oven's chain conveyor, thuis resulting in damaged boards.

Here is the issue: We have a particular board that has a very low profile due to the fact that it is a 0.062" fab and the tallest part is 0.074", making the total board height 0.136". When the sensors are set low enough to detect this board, we get occasional false alerts. (probably because most of the IR is getting reflected off the surface of the belt). We are trying to some additional adjustments, but....

My thoughts are to use a different type of sensor to drive an alert system. The issue is what sensor and how to mount it?

• 28015 PING sensor? If used overhead, can it detect a distance difference of 0.136"?
• 28317 Linescan Imaging Sensor? overhead or
• 28380 ColorPAL overhead? But I think the ColorPAL needs the "item" up against it's light shroud.
• 27401 QTI Sensor
• 28302 Color Sensor Module?
• something else?

Phil Pilgrim (PhiPi)

How wide is the belt? What color is it? Is it shiny or dull? How much clearance do you have above the belt to mount a sensor?

-Phil

Dave Hein

Can the belt be painted with flat black paint, or covered with a non-reflective surface?

TinkersALot

Could a lever bar be placed across the belt that would 'swing' whenever a board was going by? And that lever be used to simply close a switch contact?

WBA Consulting

• Belt is 24" wide
• It is a rubber coated fiber belt
• It is black and mostly non-reflective.
• It cannot be painted because it is a specially made rubber that meets ESD requirements.
• We do have a lever bar at the end, but the design of the belt is such that it is ok fro a few to stack up end to end, the problem doesn't arise until enough boards stack up to reach the reflow oven exit.

The IR sensor was replaced with a laser and laser detector that are housed in small pieces of copper pipe. These appear to be working along with a spacer bar placed under the belt at the sense point to make the belt as flat as possible at that point. These are linked to a DIN rail delay module and the output drives a flashing yellow light and alarm.

However, I would like to use something a little "industrial" as this needs to be very permanent and have a low chance for error.

Adam Wiesler

I don't know what your price range is, but my dad uses vision systems a lot at work. He likes to use the ones made by cognex, but they might run you a couple thousand a pop. Another options is, if the boards are a different temperature than the belt, parallax has a couple of IR temp sensors that you might check out.

erco

Laser.

Any type of laser, laser diode, laser pointer, voltage lowered so it will last forever, aimed horizontally across your belt 0.100" above aimed into a phototransistor.

Phil Pilgrim (PhiPi)

WBA,

How much clearance do you have above the belt for a sensor?

-Phil

Tracy Allen

>>something a little "industrial"

Keyence is one company that makes an "optical micrometer", a laser beam that is formed optically into a wide flat parallel beam, say 1mm x 10mm, so that it can detect exactly how big an object is passing through, down to a fraction of a mm. It can be made to adapt to slow changes. The receiving optics are either a photodiode behind a lens, or a CCD.

http://www.keyence.com/products/measure/micrometer/micrometer.php

WBA Consulting

Adam: Yes, the PCBs will be much hotter than the belt, so the IR Temp Sensor sounds like a really good idea.

Phil: There is nothing above the belt except the Z-bar which is at 10 feet, so plenty of room.

Erco: We are using a laser now and it appears to be working well. Only a few false alarms caused by the belt deformation. Even though it is tight on the rollers, it's "flatness" is easily +/- 0.10" which causes issues with getting the laser down low enough to detect the 0.136" height of the troublesome board without also detecting the belt. We could increase the delay, but that creates other risks.

WBA Consulting

Tracy: That would be a really nice solution, but a little out of my cost target. I have used Keyence photo-eyes in the past and have a few in my project boxes from torn down machines. That sensor would work perfectly if the flat belt didn't vary in position so much to throw off readings of the current setup.

The explanations of the belt have started me to think about the belt itself. The adjustment options for the belt are very simple (all you can really do is tighten the rollers and make them parallel) which means that a belt that has imperfections will not ride perfectly flat. The belt itself is very expensive (probably because the conveyor is not an active product of the manufacturer) and that makes finding a hardware solution feasible. My plan is to resolve the issue for under $100. I have all of the necessary hardware except the sensor, so technically, the sensor could run $100.