On high-speed packaging
machines, what you can’t
see can hurt you. Here’s
what you need to know
about sensing clear
Package designers, though they probably won’t admit it, take great pleasure in throwing curves to system engineers. Impossible demands run down hill, so they say, and there’s a long list to prove it. Today’s equipment must handle a wide range of package sizes and materials, it must be retoolable in software, it has to be fast and precise, and it can’t cost much more than the sum of the parts.
Now consumer goods manufacturers want machines that can handle nearly invisible, or clear, objects. Recent product development and marketing campaigns — primarily in cosmetics, pharmaceutical, and beverage industries — are hammering on the message that clear means clean and pure. This strategy has paved the way for a parade of translucent and clear soaps, gels, soft drinks, and foods, and is largely responsible for the recent surge in the use of clear plastic and glass packaging.
Clear objects may look nice on store shelves and in homes, but they are a real challenge to system engineers because they are difficult to sense. Ordinary photoelectric sensors (transmitted- beam or retroreflective types) are made for opaque objects. They are typically set for maximum sensitivity and light variation (high margin). In a photoelectric sensor, the unblocked light signal is many times stronger than that required to energize the output. This translates to higher detection reliability even if dirt and dust accumulate on the sensor or reflector. But there is a down side.
If a clear object enters such a sensing field, it probably won’t be detected. The reason is that clear materials typically don’t change the light intensity enough to trigger the sensor. Even though there are several losses in the optical path — light must pass through the object, bounce off a reflector, then return through the object to the sensor — the high margin on the sensor all but prevents the output from changing. Basically, the object is ignored.
This isn’t to say ordinary photoelectric sensors are totally immune to clear objects. If the light beam is polarized, the direction and extent of polarization will be altered to a greater or lesser degree depending on the optical properties of the material. In some cases, the variation could depolarize the light sufficiently to reflect a detectable signal — but if it happens, it would merely be by chance. In fact, many unsuspecting system engineers have confused irregular sensor activity stemming from random polarization for intermittent operation, wasting hours, if not days, barking up the wrong tree.
Materials that greatly change polarized light are called “optically active.” Most plastics used for containers and packaging are optically active. For this reason, photoelectric sensors often incorporate polarizing filters to block depolarized reflections from shiny surfaces on bottles, plastic wrap, and other objects.
Something else about standard photoelectric sensors that makes them less than optimal for detecting clear objects is that they typically have a large (20 to 40%) hysteresis. Hysteresis, the difference between turn on and turn off points, prevents the output from “chattering” due to noise generated by electromagnetic interference (EMI), radio frequency interference (RFI), and ambient light. It sensitivity is tuned to energize the output with the reflector in the field-of-view, a passing object must reduce the signal by at least 20 to 40% to trigger the sensor. This is too much to ask of a clear object.
When light enters perpendicular to the surface of a clear object, it loses approximately 5% of its energy due to reflection. The same amount is lost on the way out. Two passes through an empty bottle, therefore, involves eight surfaces for a total loss of about 40%. A single glass plate, on the other hand, constitutes two surfaces, through which two passes add up to a 20% loss.
It follows then that a sensor with 20% hysteresis would just barely turn off and back on again for a single glass plate, and only if perfectly adjusted. In reality, however, one would not seriously attempt such an application because it doesn’t allow for any sensitivity drift or misadjustment.
Clear-object detection applications, thus, call for a special type of low-hysteresis sensor, having a hysteresis range of 8 to 10%. This ensures a stronger “off” condition. Granted, a sensor with 20% hysteresis would work in an application that sees a 40% light loss, but it only allows for a gain margin buffer of 12.5% between the “on” and “off” thresholds.
In the special case of bottles, 40% is the theoretical minimum light loss that occurs at the center of the bottle, where the surfaces are perpendicular to the light beam. Away from the center, where the surfaces tend to be more parallel with the light, the loss due to reflection is greater. This discrepancy (large variations in reflected light as bottles move through the field-ofview) explains the undesirable sensor chatter, or “off-on-off” response, frequently observed in bottle handling applications.
Continue on page 2
Making matters worse is that surfaces, and hence optical properties, on clear bottles are not always uniform. Some areas, for example, may pass light at a different refraction angle than others. This lensing effect changes the nature of the application, often increasing the signal instead of reducing it. Because of this randomness, each clear object detection application should be reviewed using all possible target orientations and positions.
Tests should also include new bottles (fresh from the mold) as well as handled ones (touched by humans). New bottles have different optical properties than bottles that have come in contact with machines or people, even if there are no apparent scratches or irregular surfaces. Simply put, bottles used in tests should match those on the line during production.
Putting on blinders
To accommodate the gamut of optical conditions possible in an industrial application, clear object detectors often rely on help from additional components such as coaxial optics and linear and circular polarizers.
Linear polarizers help reduce primary and specular reflections. Usually it takes two polarizers oriented 90° apart. One polarizer, in front of the source, orients the incident light to a particular direction. How the light returns to the detector determines whether it gets through or not.
If light bouncing off a reflector becomes randomly oriented (depolarized), some of it will get through to the photodetector. Light reflected from a shiny object, however, remains polarized, never reaching the detector.
Now if a clear plastic object gets in the path of the sensing beam, part of the light passes through the object while part is reflected back to the sensor. The reflected light is blocked. What happens to the light that gets through the plastic is a bit more complicated. It turns out that light passing through plastic may become depolarized. Because the randomly oriented components may add to the depolarized light bouncing off the reflector, the sensor (detector) may end up getting more light than it would have from the reflector alone. Thus, linear polarization alone isn’t sufficient for sensing plastics.
Circular polarizers were initially developed to remove ambient light glare from cathode ray tubes. Their extension to photoelectric applications is rather recent.
A circular polarizer is a laminate consisting of a linear polarizer and a quarterwave plate. The plate has an optical axis parallel to the flat surface of the polarizer and is physically oriented 45° from the linear polarization axis.
Going through a circular polarizer, nonpolarized light is first linearly polarized, then it passes through the wave plate, from which the X and Y wave components emerge 90° out of phase. This light forms a helical pattern from which the name “circular polarization” is derived.
When circular polarized light is reflected by a specular surface, such as a clear bottle, the polarization is reversed. If, in this state, the light re-enters the circular polarizer from the opposite direction, it must first pass through the quarter- wave plate, making it perpendicular to the linear polarizing surface. So, reflected light from the bottle is entirely blocked.
An actual reflector, on the other hand, returns randomly polarized light, some of which is bound to get through the quarter-wave plate and linear polarizer. Consequently, light from the reflector is not blocked. Circular polarization helps prevent this “double pulsing” by making it more difficult to depolarize the light.
Get in line
Coaxial optics, where all components lie along a common optical path, is another way to improve clear object detection. The advantages over the side-byside optics used in standard photoelectric sensors are many.
One advantage arises in close quarters. When a standard sensor and reflector are close together, the emitted light is reflected directly back to the source, preventing the receiver from seeing any transmitted light. This creates a blind zone directly in front of the sensor, the severity of which grows with the separation between source and receiver lenses.
The use of coaxial optics reduces the possibility of not coupling the emitted light to the receiver because the source and receiver share the same optical path. Thus, with coaxial arrangements, sensors and reflectors can get as close as required without concern of a “dead zone.”
Another advantage stems from how coaxial optics treat variations due to imperfections. Some bottles, because of warping and molding variances, may bend light in such a way that it increases the coupling, and hence gain margin, between the source and receiver. It’s possible then that even a “clear object” sensor may miss them. (Hold any bottle against the light and you’ll see the variations that cause this sort of thing.) With coaxial optics, however, the margin can’t increase because the coupling between the source and receiver is already maximized.
Other advantages of coaxial optics owe to a narrower than usual light beam and overlapping source and receiver areas. In general, coaxial optics produce a beam that’s contained within the receiver’s field-of-view. Because the receiver captures the entire beam, any bending of the light would tend to deflect the beam out of the receiver path. This, in effect, increases the sensor’s ability to see clear bottles because no returned light indicates the presence of an object.
Polarizers and coaxial optics developed to sense clear materials were first used with capacitive proximity sensors. Applications were only marginally successful. Capacitive sensors have too many limitations to adequately do the job. The primary issues are range (2 to 10 mm just isn’t enough) and sensitivity to moisture (packaging plants can be fairly wet).
Ultrasonic and photoelectric sensors were next in line to be outfitted to detect clear objects. Initially, ultrasonic sensors were more dependable and accurate, but over the last few years, improvements in photoelectric sensors have made them the primary choice. Some of their advantages include better edge detection, higher resolution, and indifference to target size, shape, and temperature. Not only that, costs continue to drop while performance heads in the other direction.
Jeffrey R. Holman is marketing manager for photoelectric sensors at Rockwell Automation, Presence Sensing Business, Chelmsford, Mass.