How to Design for Reliable and Accurate Presence Sensing

As both consumer and industrial systems become smarter, more interactive, and more autonomous, they need to be able to sense the presence of an object, user, or passer-by. While basic presence sensing techniques and technologies are well established, designers are under pressure to perform presence sensing more accurately, efficiently, cost effectively, and reliably.

The typical applications that are demanding such improvements include machinery and manufacturing, where gear teeth or bottles, respectively, need to be detected. In robotics, the presence of nearby objects or people needs to be detected so the robot can either avoid them or stop moving, and in smart devices and kiosks, the system needs to be able to detect and respond to user interactions.

This feature will describe some of the main presence sensing technology selection criteria to consider for various applications, before introducing suitable devices and how to design with them.

Presence sensing design requirements and technologies

The performance of presence sensing systems is dependent on real-world, application specific issues such as indoor vs. outdoor setting; target object size, material, and consistency; the confinement and consistency of the situation; power requirements and available power; and packaging/placement requirements, to cite a few factors.

Also, many presence sensors are in physical settings where they are exposed to both a harsh environment as well as unintended (or even intentional) abuse from people and objects hitting against them (such as cars grazing walls), so physical mounting and ruggedness are issues to consider.

Designers also need to consider their mix of sensor techniques. Using a single sensor to detect with 100% accuracy that someone has walked across a private driveway is challenging. There can be false alarms due to animals or wind-blown leaves if the arrangement is too sensitive. On the other end, if the arrangement is not sensitive enough, it may miss the presence of small children.

As a result, both the selected approach and chosen sensor need to be evaluated carefully and may even require a redundant, independent, alternate sensing approach for higher accuracy with respect to “misses” as well as false positives.

Three widely used approaches for presence sensing use optical, ultrasonic, and inductive proximity principles. There are also other options such as radar, but this has some complex RF design issues as well as regulatory and possible licensing concerns. Another option is a video-based approach which requires complicated signal processing and algorithm development, so it is often not the best candidate for the low-cost reliability and ruggedness that many presence sensing applications require.

Another factor to understand is the beam width of the sensing signal. Depending on the technology, this can be a narrow, tight beam or a wider, broader one, and the application defines which one is preferable. For example, a narrow beam is needed to sense individual bottles on a production line conveyor track, while a broader one is a better choice for sensing people walking up to a kiosk.

Among the many design considerations when selecting a presence sensing solution are:

• Distance from sensor to sensed object
• Beam width
• Interference (physical impediments between sensor and sensed object)
• Consistency of overall sensing situation and the person or object being sensed
• Concern about both false positive indications as well as false negative indications
• Environmental setting (indoors, outdoors; daytime and nighttime (for outdoor settings); chemicals and oils,
• Physical abuse (deliberate or accidental)
• Location options for sensor electronics versus sensor transducers
• Power requirements
• Long-term reliability
• BOM cost and installation cost

Use multiple sensors for accuracy and reliability

An application installation is not restricted to using just a single sensor or even a single sensing technology. Although costlier, it is not unusual to install two or more sensors, often with differing technologies to provide a more reliable indication of presence.

The option to use multiple sensors, whether the same or different types, means that the designer must address a question about system priorities. If the prime concern is to ensure that no real targets are missed, then one solution is to use the sensors in parallel (the logical OR function) for overlapping coverage. That way, a positive indication from any one sensor can signal the presence of a target.

Or, is it more important to ensure that there are no cases where an object is detected even if none is actually there (a false positive)? In that case, the multiple sensors can be used in a series arrangement, to implement the AND logic function for a “unanimous” vote approach.

Both OR and AND functions can be implemented by hardwiring the outputs if they are compatible. However, a software-based approach is more flexible in combining different types of sensor outputs, and can use different algorithms to filter the individual sensor signals. The use of software also allows weighting of a series of sequential sensor readings, and even their combination in both AND and OR operations to best minimize both false negatives and positives. It also provides flexibility in modifying the logic-based decision process and adding conditional decision making if needed.

Start with optical sensing

The use of a focused light or laser beam, usually in the infrared (IR) spectrum, is an obvious way to determine presence. A typical sensor consists of an emitter and receiver which may be mounted in opposition across the area of interest, or they can be co-located with a mirror or retroreflector at the far end. Using co-location simplifies installation and wiring, but also means that the optical path length is doubled, which reduces received signal power and increases the signal-to-noise ratio (SNR).

The emitted optical signal is usually pulsed to distinguish it from interference, as well as from nearby lighting, reflections, interference, and other optical noise sources. The light beam is also tightly focused so that its off-axis beam width is restricted.

Optical detection can be used outdoors, but often has problems related to sunlight induced overload as well as the need for wide dynamic range as it must cover both darkness and daylight situations.

Overall, the optical sensor is a reliable approach in controlled and constrained settings. It can detect targets down to a few millimeters in size, and can sense small objects across an area of tens of meters. However, dust, fog, steam, and nearby reflection can prevent sensing or cause false indications, while something as simple as a dirty mirror (if used) can cause the reflected signal power to drop below a minimum threshold.

Optical systems now have a unique advantage compared to almost all other presence detectors: by using optical fiber, their electronics can be located at a considerable distance from the exposed placement of the optical window of both emitter and receiver. A pair of fibers is used to carry the LED/laser light to an emitting site, and the received light to the receiver phototransistor. Only the relatively rugged fibers are exposed to the harsh environment where they are exposed to temperature extremes, oil, dirt, and abuse. Also, additional protection can be provided for the fiber more easily than for the electronics.

A good example of this use of optical sensing is the Panasonic FX-500 series (Figure 1).

Figure 1: The control unit of Panasonic’s FX-500 optical sensor, with its relatively sensitive electronics, can be placed several meters away from the exposed ends of the more rugged emitter and receiver optical fibers. (Image source: Panasonic Industrial Devices)

This sensor places all the circuitry in a unit which also includes topside buttons for setup of mode, output intensity, threshold setting, and test functions (Figure 2). This unit can be conveniently located tens of meters away from the point of use.

Figure 2: The conveniently located control unit of the FX-500 also contains user operated pushbuttons which are used for setup, threshold setting, and test. (Image source: Panasonic Industrial Devices)

A pair of optical fibers, each just 2 millimeters (mm) in diameter, carries the emitted and received optical signals. The fibers come with threaded bushings for easy and rugged attachment to a panel or surface (Figure 3). Due to their tight, 25 mm bend radius, the fibers can be easily routed around or through obstacles between the electronics and the placement of the exposed fiber ends.

Figure 3: The sensing end of the FX-500 consists of two thin optical fibers housed in rugged bushings for minimum physical exposure. The thin, horizontal, red line on the left is a “visualization” of the unit’s IR beam. (Image source: Panasonic Industrial Devices)

Ultrasound-based presence sensing

As with light beam, the use of ultrasonic energy is another time-tested technique for presence sensing. The principle is simple enough: a piezoelectric transducer is pulsed, and then it “listens” for the return echo from the object of interest. The echo power which must be detected is a function of transmitted output, distance, and the nature of the echoing object (a car versus a person in a coat, for example).

Ultrasonic sensing has a practical range of up to several meters, and a relatively wide beam width. It is best suited for larger targets with a relatively large cross section that are operating in controlled, indoor applications as any wind will affect the signal path in both directions, and thus affect consistency of performance. Changes in temperature will also affect the propagation of the ultrasonic energy.

The UM18-218 ultrasonic sensor from SICK AG is representative of a unit which can be used in both benign and harsh environments (Figure 4). The transducer is mounted in a threaded bushing (18 mm in diameter) and has an IP67 rating so it is protected from dust and capable of withstanding water immersion between 15 cm and 1 meter (m) for 30 minutes. It runs off a 10 to 20 volt DC supply, and dissipates under 1.2 watts (W) while operating at 200 kilohertz (kHz) with an 80 millisecond (ms) response time. Its output is an easy to interface, open collector PNP transistor with a 200 milliampere (mA) load capability.

Figure 4: The SICK UM18-218 200 kHz ultrasonic transducer comes in a rugged bushing and meets IP67 requirements for resistance to dust and water. (Image source: SICK AG)

The UM18-218’s beam width is shown in its detection range graph (Figure 5). For designers, this defines the performance envelope for detection width versus distance (left axis, up to 1.5 m) for both detection (operating) and reach (maximum).

Figure 5: The graph which defines the detection area envelop versus distance of the SICK UM18-218 for two basic target types is a critical tool for assessing the viability of this sensor type and model under optimum conditions. (Image source: SICK AG)

Note that these ranges are a function of the target specifics; in the graph, the targets are a 500 mm square plate at right angles to the beam (a favorable target with a large cross section), as well as a 27 mm diameter pipe (a far less visible target).

The type of target not only bounds the distance, but also the shape of the zone, so run evaluation test results will differ for each target type.

Inductive proximity sensing: an option for ferrous objects

A third option is the inductive proximity sensor, which is triggered when a ferrous metal object comes within close proximity of the sensing coil. The inductive proximity sensor is especially well suited for detecting small ferrous objects at close range, such as gear teeth, keypad buttons, and fan blades; it can also be used at longer distances to detect larger objects such as cars. As a non-contact, magnetic-based sensor, it is reliable and unaffected by dirt, oil, moisture, light, or other interfering contaminants, and can be housed in a rugged, non-metallic enclosure.

Until the availability of ICs which could work with inductive coils, the design of this class of sensor was complicated and required many discrete components. The challenge of building this type of presence sensor with consistent performance is now greatly eased by ICs such as the LDC0851 from Texas Instruments (Figure 6).

Figure 6: The Texas Instruments LDC0851 inductive proximity sensor manages a reference coil and a sensing coil; when a ferrous material disrupts the sensing zone, the inductance of the sensing coil changes which then changes the frequency of its associated L-C oscillator. (Image source: Texas Instruments)

This IC is an inductance “comparator” that uses two user-provided coils. One coil is used as the sensing coil and the other as a reference coil. In operation, the sensing technique exploits the fact that the actual inductance of an open field coil will be affected by nearby ferrous objects. Note that this does not apply to shielded or toroidal coils, with no external field. If the coil is part of an L-C resonant oscillator, the free-running frequency of the oscillator will be a function of the inductance.

In operation, the output of the LDC0851goes low when the sense inductance drops below the reference inductance (as measured by changes in its nominal 200 kHz oscillation) and returns high when the reference inductance is higher than the sense inductance. To eliminate chattering of the output due to vibration of the system, the IC includes user-adjustable hysteresis around the reference coil threshold.

The sense and reference coils are critical to successful implementation of this IC, as they aid in maintaining a consistent switching distance over temperature and to compensate for other environmental factors. The coils can be external discrete passives, or fabricated as part of the pc board. The latter lowers cost, but may increase pc board size; for many low-cost applications, this is an acceptable tradeoff.

Ideally, the two coils should be identical, but issues of placement with respect to each other, orientation, and other factors may require fine-tuning of one or both coils for optimum and consistent performance. In some designs, pc board size limits require that the coils be placed on top of each other, while in other cases they are located side by side.

To get the best sensing range with stacked coils, the spacing between the sensing coil and reference coil should be as large as possible on the multilayer pc board (Figure 7). For this IC, the target being sensed should be located at a distance which is less than 40% of the reference coil diameter.

Figure 7: The sensing coils of the LDC0851 are key to reliable operation and can be fabricated as part of the pc board layout; they can be located side by side, or stacked with their turns directions as indicated. (Image Source: Texas Instruments)

Note that the inductive proximity sensor is a variation of the magnetometer, a passive device which senses changes in the earth’s magnetic field. The magnetometer is well suited for larger ferrous objects at a distance up to several meters, such as vehicles. However, the proximity sensor can also be used in this role with larger coil sizing and calibration.

Conclusion

Sensing the presence of a person or an object – small or large, ferrous or non-ferrous – is a common requirement in many systems. While optical, ultrasonic, and inductive approaches are well understood, designers are under constant pressure to meet improved accuracy and reliability requirements at lower cost and greater efficiency.

Vendors are responding with various solutions to help, but before making their decision, designers need to factor in issues such as the operating environment, consistency of the target, interference from nearby objects, and ambient conditions. That decision may include the use of multiple sensors of the same or even different types for added detection accuracy, as well as the adoption of software-based presence sensing approaches.