Articles


 

SHARE:  

Understanding Sensors & Error-Proofing, Part 2: Failure Modes for Photoelectric Sensors

By: George Keremedjiev

George Keremedjiev is president of Tecknow Education Services, Inc., Bozeman, MT; 406/587-4751, www.mfgadvice.com.

Sunday, May 1, 2016
 

Fig. 1—This part, with multiple hole features and a unique shape, demands careful testing of photoelectric sensing before installing sensors in the tooling.
Fig. 2—Copper tubes provide air blasts to eject parts from the die.
 
Fig. 3—Here’s a typical through-beam photoelectric setup where a transmitter emits a beam of light across the exit path of the part to the receiver. The transmitter and receiver mount on the end of the die.

Fig. 4—Assuming equal and symmetrical bursts of air from the copper tubes, the part may start to rotate vertically about an axis (as shown by the two arrows). This rotation could drastically reduce the amount of part-material cross-section within the beam of light. This reduction may be random and unpredictable.

Fig. 5—If the air bursts are not of equal force, the part could start to spin about a second axis (as shown by the left-to-right/right-to-left circular arrows).
Fig. 6—Shock and vibration of the press and die can introduce unpredictability (signified by the centered arrows).

Fig. 7—This illustration depicts the part rotating left-to-right or right-to-left in a circular pattern. When combined with previously described multiple axes of rotation, this circular motion can prove indecipherable to a photoelectric sensor.
Frustrating to many a metal stamper is the seemingly magical reaction that photoelectric sensors can have to a moving stamped part. The sensor salesperson assures you that this particular sensor will detect your stamped part—even as that part exits the die rotating and gyrating into various flight paths on its way to the container. So, the sensor is purchased and mounted on the die to detect the exit of the stamped part. All seems to be okay for several hours, perhaps even for a full day’s run. Then it starts…a seemingly random and intermittent series of nuisance stops. The operator checks the die and sure enough—the part did exit but somehow the sensor failed to detect it. Why would the process work for many strokes of the press and then, with no apparent change in that process, the sensor fails to detect the part?

The sensor salesperson assured a flawless application, and it sure seems that way for 99.9 percent of the time. But production floors view that 0.1-percent failure rate with disdain, and rightly so as it can translate to dozens of nuisance stops in a given day. What to do? How to simulate all possible part position on a test bench to eliminate even the slightest possibility of a blind spot in a million possible exit configuration for that part? How should one test for conditions that seem impossible to predict? There is a way.

Consider Part Rotations, Orientations As they Exit Dies

Imagine a World War I biplane putting on an acrobatic performance in an air show. That’s akin to what happens to a stamped part as it exits the die. To help visualize this, consider a representative part in Fig. 1. The small, stamped bracket with three holes—two hexagonal holes on the ends and one oval hole in the middle—has a cross-sectional thickness of 0.040 in., with the holes accounting for about 20 percent of removed material. Let’s explore this part’s various motions as it exits the die, and how these motions are captured, or not captured, by the photoelectric sensor.

Fig. 2 shows two copper tubes using air blasts to eject parts from the die. The copper tubing is symmetrical and well-aligned with the part. This, of course, is not the case for many dies. Setters and operators make their favorite adjustments to the tubing, bending it in whatever manner necessary to get the parts out of the die. But for our purposes, assume that the two puffs of air are synchronized and that the parts eject symmetrically—exiting on a parallel path to the tooling.

Fig. 3 shows a typical through-beam photoelectric setup where a transmitter shoots a beam of light across the exit path of each part to the receiver. The transmitter and receiver mount on the end of the die. Ideally, the transmitter and receiver would be mounted within blocks of steel or aluminum to protect from damaging blows as the die is moved about the shop and serviced in the toolroom. The 0.040-in. part thickness is well within the detection capabilities of the photoelectric sensor. What could possibly go wrong?

Fig. 4 depicts the first challenge to the sensor’s capabilities. Assuming an equal and symmetrical burst of air from the two copper tubes, the part may start to rotate vertically about an axis (as shown by the two arrows). This rotation could drastically reduce the amount of part-material cross-section within the beam of light. This reduction may be random and unpredictable. There may even be circumstances where the rotation only presents the crown above the oval hole of the part to the light beam as the upside-down part exits the die.

What if the two air bursts are not of equal force. As Fig. 5 shows, the part could start to spin about a second axis (as shown by the left-to-right/right-to-left circular arrows). In this case, the part may shoot out of the die lengthwise, presenting a minimum amount of solid metal to the light beam. However, the very next part, with a different angle of rotation, would offer much more metal within the light beam…an unpredictable situation.

Shock and vibration of the press and die also can introduce unpredictability (Fig. 6). The part may jiggle as it leaves the die, and this could present material to the light beam that is not stable on the X and Y axes of vibration, making the oval and hexagonal openings truly variable and unpredictable within the light beam. Fig. 7 depicts the part rotating left-to-right or right-to-left in a circular pattern. When combined with the previously described multiple axes of rotation, this circular motion can prove indecipherable to a photoelectric sensor.

Simulate on a Test Bench

Personnel testing the photoelectric sensor must be fully aware of an exiting part’s potential rotations and vibrations. There is no excuse for not spending some productive R&D time to simulate all of the above conditions before committing to a given sensor application and installation. I suggest first testing the photoelectric sensor to ensure that it will not pick up any oil mist ejected along with the part. Should the part refuse to exit, the photoelectric sensor may be fooled into thinking differently as it picks up oil mist instead, and incorrectly issues a part-exit signal. This can lead to a double-hit die crash. Oil-detection testing is best accomplished by filling a plastic spray bottle with the same oil, with the same viscosity, as will be used to produce that specific part. This test identifies any sensors that could be fooled by the oil.

After oil-testing, conduct a meticulous test of what the photoelectric beam can and cannot detect related to rotations and vibrations of the stamped part. Use a test bench with the transmitter and receiver mounted on adjustable, track-mounted towers. Mount the transmitter and receiver to mimic their positions on the die. Then, take a set of tweezers in the case of a small stamped part, or a pair of long-nosed pliers for a larger part, and hold the part within the light beam. With much patience, rotate the part into every conceivable position within the light beam. Introduce slight vibration and jiggling by agitating the tweezers or pliers.

In other words, simulate every possible worst-case scenario regarding the position of the part within the light beam. You must find any blind spots that may arise with the part in a particular position, and then decide if the sensor can be adjusted electronically to compensate for such blindness. Increase sensor sensitivity to accommodate blind spots only after making sure that the increased sensitivity will not result in the sensor mistaking oil droplets for the part.

It’s Not Just Part-Out

Although this article focuses on part ejection, the very same concerns are valid for other photoelectric-sensor applications, such as feed detection, cam/block returns, part-feature detection, transfer operations, etc. Remember: All target motions, no matter the application, should be simulated on a test bench first, before any photoelectric sensors are installed within the tooling. MF

Part 3 of this series will detail the various part-quality-measurement sensors that can be used within tooling to perform 100-percent inspection−at running speeds—of critical dimensions.

 

See also: Tecknow Education Services, Inc.

Related Enterprise Zones: Sensing/Electronics, Tool & Die


Reader Comments

There are no comments posted at this time.

 

Post a Comment

* Indicates field is required.

YOUR COMMENTS * (You may use html to format)

YOUR NAME *
EMAIL *
WEBSITE

 

 

Visit Our Sponsors