Photoelectric sensors with built-in microprocessor- based intelligence offer easier and more accurate adjustments, which optimizes their sensing reliability. These smart sensors, which are rapidly becoming the norm rather than the exception, also deliver advanced self-diagnostics and the ability to interface with sensor networks.
To find out what these smart devices can do, let’s first look at how conventional photoelectric sensors work. Then we’ll see how adding intelligence to these sensors expands their capabilities.
Conventional operation
Every photoelectric sensor detects changes in the amount of light it receives and responds to them by changing state, or switching On or Off (thus triggering accept/reject, go/no-go, or start/stop actions). The key to operating a photoelectric sensor is to maximize the contrast ratio between two light levels, commonly called “light” and “dark,” as seen by a photoelectric receiver. “Light” means the receiver clearly sees an unobstructed light beam from the sensor’s light source and “dark” means that the light beam has been broken by an object so that the receiver sees no light. A high contrast ratio between light and dark, generally 10 or more, gives high sensing reliability.
In a simple application, for example, a cardboard box on a conveyor breaks a light beam between a sensor’s light source and its retroreflective target or receiving unit. The box completely blocks the beam so the receiver sees no light, thus creating an infinite contrast between light and dark, hence, maximum reliability.
But, not all applications are that simple. A web sensing application, for example, may produce a low contrast if the web material (clear plastic or thin paper) lets much of the sensing light pass through. In this case, there may not be enough contrast to switch the sensor On and Off. Therefore, an operator must carefully align and adjust the sensor to obtain the highest contrast possible.
The point at which there is enough light at the receiver to cause the sensor to change state (switch On or Off) is called the switching threshold. A sensitivity adjustment knob on the sensor moves the threshold up (more light required to switch) or down (less light required) to suit the application. The ideal setting places the threshold midway between the light and dark signal levels. With this optimum setting, the sensor can still change state after lighting conditions deteriorate.
Conventional sensors typically provide little feedback to indicate when a sensitivity adjustment is correct. This feedback is usually limited to switching an output status indicator at the threshold. Therefore, the operator must have an experienced “feel” for making this adjustment. Or, as is often the case, the operator simply adjusts the sensor until it works — without trying to optimize the setting — and hopes for the best.
Because of this imprecise adjustment, thresholds are sometimes set too close to either end of the contrast range where small changes in lighting may prevent them from switching. Thus, the length of time sensors remain working is often a matter of chance.
Smart sensor adjustments
To simplify adjustment procedures, new photoelectric sensors with built-in intelligent microprocessor chips continually monitor received-light signals and display their values, using LED bar graphs or LCD digital readouts, Figure 1. By monitoring these readouts, an operator easily fine-tunes the sensitivity adjustment to place the threshold midway between light and dark levels. Thereafter, the display of signal strength gives machine operators and maintenance personnel an instant visual indication of how well a sensor is operating.
Switching stability
When the contrast ratio for a sensor approaches one (no discernible difference between light and dark levels at the receiver), the change in light level approaches a zone called the hysteresis of the receiver’s amplifier. Hysteresis can be thought of as an inherent characteristic that effects stability of the sensor’s switching function. With zero hysteresis, a received-light signal that hovers right at the switching threshold can cause the sensor to wildly oscillate between On and Off, a condition called buzzing or chattering. To avoid this unstable condition, there must be enough signal change between light and dark to overshoot, or move past the threshold (and the hysteresis zone) before the sensor switches.
Heating thermostats provide a good example of hysteresis. A thermostat set for 72 F typically requires that the room temperature drop to 71.5 F before it energizes the heating system, and the room must warm to 72.5 F before the thermostat turns the heat off. Otherwise, the thermostat would constantly switch the furnace On and Off.
Hysteresis in conventional sensors is a fixed value, typically 10 to 20% of the threshold level. However, digital circuits in smart sensors let users adjust hysteresis to accommodate specific sensing contrast and ambient conditions.
Diagnostics
Conventional photoelectric sensors can fail to operate for a number of reasons. For example, when dust gets on the lens, or the sensor and receiver get out of alignment, the light level at the receiver may be so low that the sensor doesn’t switch properly. In such cases, a technician needs to inspect the sensor to determine the cause of failure before it can be corrected.
On the other hand, many intelligent sensors monitor signal strength and contrast readings, then report impending problems when they detect abnormal values, Figure 2. In some cases, light indicators or displays identify the cause of a problem, thereby saving substantial troubleshooting time.
When a light signal is too low, an alarm output may send a warning signal to a system controller or a machine operator so corrective action can be taken. In a common example, a sensor operates where there is airborne dust (or oil mist or caustic fumes). As dust accumulates on the sensor lens, less light reaches the receiver. Sensor self-diagnostics, located in a microprocessor chip, send an alarm signal to a controller and flash a status LED when the light level approaches the sensing threshold. Typically, this occurs when the signal drops from its normal value, which is usually several times the threshold value, to within about 150% of the threshold setting.
These alarm outputs are sometimes used to automatically correct a problem. For example, where dust buildup on lenses is a known problem (corrugated cardboard processing), loss of the light signal triggers an alarm output that energizes an air valve, causing it to blow off the dust. This use of an alarm output is known as a “dirt alert.”
Diagnostic alarms or alerts are early warnings rather than failure signals. They let users make corrections before a sensor fails, thereby greatly reducing downtime.
Sensor diagnostics can monitor and evaluate several parameters including:
• Low light signal — caused by a dusty lens or misalignment between sensor and receiver.
• High dark signal — increasing background reflections in a normally dark condition cause erroneous switching.
• Low contrast level — approaching the sensor’s hysteresis level. This can be a problem in sensing clear glass or plastic, Figure 3.
• Output overload — failure of a load or input circuit.
• Low supply voltage (below specified minimum).
• High temperature (above specified maximum).
• Severe moisture or condensation — found in food processing applications.
Networking
Connecting conventional sensors to a control system requires wiring between each sensor and a dedicated controller, Figure 4.
But some intelligent photoelectric sensors contain the electronics and firmware (software program on a microprocessor chip) necessary to interface with a sensor network. With such networks, all of the sensors connect to a single low-voltage dc bus cable along with the controller. This greatly simplifies the wiring because only one input is required at the controller, rather than an input for each sensor.
In a typical network, each sensor is assigned an “address” for its load, or switching output. A sensor with diagnostics has an additional address for its alarm output. The host controller polls the various sensors for both alarm and load data in a programmed sequence.
Sensor networks typically support peer-to-peer communications. With this capability, sensors can control other sensors without instructions from a host controller. For example, a sensor that detects the presence of a carton can trigger a second sensor (on the same bus) that checks to make sure the flaps are closed.
As with most emerging technologies, several sensor networks are vying for dominance so they will ultimately become the standard. Furthermore, two networks were recently introduced that link not only smart sensors, but also controllers and other factory floor devices (PTD 7/94, p. 19). Sensor users should be flexible during this period of network proliferation and should select a network configuration that will most likely fulfill their specific requirements, both current and future.
This article is based on information provide by Robert H. Garwood, senior applications engineer for the Banner Engineering Corp., Minneapolis.