Choosing sensors for the application: Answers to audience questions

The Control Engineering RCEP webcast, “Choosing sensors for the application,” resulted in more questions than speakers had time to answer in the allotted time. Here each speaker provides more answers.

By Frank Lamb and Chris Thompson August 29, 2017

Sensor selection was the topic of an Aug. 22, 2017, Control Engineering webcast; the two expert presenters answer additional audience questions below. The archived version of the webcast, along with a link to a quiz for RCEP professional development hour (PDH) is available for a year at Upcoming Webcasts.

The webcast provides advice about how to choose sensors for automation and control applications, including how details about the implementation are important for selecting the right type of sensors. Webcast learning objectives follow. 

  1. Define what is to be sensed and differentiate between the two major areas of applications for sensing, machine position and conditions and quality of product.
  2. Define where sensors can be located and constraints.
  3. List methods, technologies, and considerations for sensor applications.
  4. Outline fundamentals of specifying process sensors.
  5. Describe several process sensor technologies and their application.
  6. Provide lessons learned from installed applications.

Webcast speaker information follows.

Frank Lamb is founder, owner, and manufacturing and automation business consultant, specializing in controls, programming, machine building and design at Automation Consulting LLC. Lamb also is a Control Engineering Editorial Advisory Board member.

Christopher J. Thompson, PE, is department manager, process engineering, Process Solutions Department at Matrix Technologies Inc. Matrix Technologies Inc. was named System Integrator of the Year by Control Engineering magazine in 2008, 2012, and 2016. 

More answers on sensor selection

Below, Lamb and Thompson provide additional answers to webcast audience questions about industrial sensors.

Q: Can you discuss communications protocols used for sensors?

Lamb: Most communications protocols are common to control networks rather than sensors themselves. For instance, Allen-Bradley’s Remote IO was one of the early options for connecting nodes in real time to a control system, but it required an intermediary device between the sensor and the controller. Later came DeviceNet [now managed by ODVA], which was an attempt at placing individual devices on a network. Profibus [PI North America, promoted by Siemens and others] and Modbus [Modbus Organization Inc., promoted by Modicon and others] also were early networks to handle I/O, though not on an individual device basis. HART protocol (carrying communications over a 4-20 mA linkage) was another early method.

Most sensors still are wired into a device that handles the communications rather than being connected directly, but almost every controls vendor has some sort of Ethernet protocol that they favor for devices. EtherNet/IP [also ODVA], Profinet [PI North America], and EtherCAT (EtherCAT Technology Group, championed by Beckhoff Automation and others) are all options for this. Sensor manufacturers usually make versions for the major platforms.

Q: Can you discuss application of radar and ultrasonic sensors and transmitters?

Thompson: Radar sensors are typically associated with measurement of liquid or bulk level in some type of container (e.g. tank, vessel, silo, or hopper). The sensor emits a microwave pulse that is either directed down a fixed probe (guided wave) or into the open area of the container (non-contact). In both cases, the transmitter then detects the reflection time of the microwave pulse as it bounces back from the fluid or build solid level in the container. Guided wave also can detect interface levels in liquid-liquid applications, and its measurement is largely independent of the property changes that occur at various operating conditions. Non-contact radar sensors can be inhibited by use with some solids, fluids with low dielectric, substantial surface turbulence, condensing vapor environments, and echoes from container walls or internals.

Ultrasonic sensors operate on a similar principal as the radar, with the sensor measuring time-of-travel or reflection time of the ultrasonic pulse emitted. The sensor correlates this time to the liquid or bulk solid level in a container, or flow through a pipe. The flow sensor is typically used with liquids and gases, and the emitted ultrasound must be capable of penetrating through or being reflected from the material being sensed. Ultrasonic level sensors are fixed, but flowmeters can be either fixed or portable, in-line or clamp-on. Ultrasonic level sensors can be inhibited by use with some solids, foaming, interference or echoes from container walls or internal, smoke and/or dust, and heavy vapors. Flow sensor applications are highly dependent on the fluid being metered. If there are any questions or uncertainties regarding the sensor, inquire into possible vendor testing, instrument trials or rentals, or an application warranty.

Q: Can you discuss sensor use in hazardous areas?

Lamb: To prevent explosions and combustion in hazardous areas, sensors are housed in an explosion-proof enclosure or intrinsically safe "NAMUR" rated sensors are used. These sensors operate on <8 V dc to prevent sparking and often are connected through an intrinsic safety barrier.

Q: Can you discuss accuracy considerations for gas flowmeters?

Thompson: With most gas flowmeters, accuracy will depend on the material being sensed, the process conditions encountered, the installation location, operating range, and the technology selected. Condensing vapors are a challenge for most applications, so operating in a range that will minimize the likelihood of liquid droplets forming is important. Depending on the technology, turndown could impact not only the accuracy but ability to use the meter (e.g. vortex, turbine, differential pressure). Particulate-laden gases also can pose a challenge for some meters, depending on the type and concentration of particulates.

Proper temperature/pressure corrections as well as proper flow distribution into the meter itself (e.g. sufficient conditioning or upstream/downstream straight runs) will impact differential pressure flowmeter accuracy, while thermal mass meters will struggle with streams containing variable composition, aerosols, droplets, etc. Flowmeter accuracy issues can be addressed by having an accurate description of the stream being sensed, an clear definition of the operating parameters, and an understanding of the potential technology limitations prior to sensor selection.

Q: Can you discuss electrical parameters (ac/dc, sink/source, voltage, and analog)?

Lamb: Discrete sensors are available typically in standard voltages; 24 V dc and 120 V ac, but they also may be available with a 12 V dc output. They may also use a contact closure, similar to relay contacts; you can use a variety of voltages with those. DC sensors come in 2 "flavors," sinking or NPN transistor-based devices, and sourcing or PNP transistor-based devices. Sinking sensors receive current flow from a sourcing-type input, while sourcing sensors provide current to a sinking-type input; they are therefore complementary. AC sensors general use a Triac (solid state ac device) output. Analog sensors generally are available in 0-10 V or 4-20 mA varieties; mA outputs are considered to be more immune to noise, but 0-10 V signals generally can be run over a longer distance.

Q: Can you discuss best questions to ask stakeholders when selecting sensors?

Thompson: In general, many of the components of the sensor specification itself may come from the stakeholders, such as process conditions, stream composition, accuracy requirements, etc., and buy-in should be obtained that the values specified are accurate. Some additional questions include:

  1. What’s the history with this application and sensors that have been used in the past?
  2. Are there any lessons learned to incorporate into our design considerations?
  3. Is there a preferred technology or make/model that has been identified at the facility or elsewhere throughout the organization?
  4. What are the accuracy, reliability, and range requirements for the sensor?
  5. Any access or operating limitations based on the installation location (e.g. internal interferences, maintenance accessibility)?
  6. What is the required sensor redundancy and what is the frequency of planned calibration?
  7. Have all stream components been considered and accurately characterized opposite the proposed materials of construction?
  8. If it is a replacement due to failure, has the root cause been determined and mitigated?
  9. What type of local or remote indication is necessary for process control, and is there any latency experienced from the instrument location opposite the control point?
  10. Into what environment will the sensor and transmitter be installed (e.g. electrically classified hazardous locations, temperature extremes)?

Q: Can you discuss the effect of noise on sensor measurements?

Lamb: Noise introduces inaccuracies into analog signal conditioners and inputs. Physical filters are sometimes used, as well as shielded cables to ground the interference. Shielded cables should be grounded on one end only for this to work. Digital filters sometimes also are used to remove 50 Hz or 60 Hz interference from ac devices, as well as filters that remove spikes of a specified maximum duration. Software filters also can be used to average or filter signals.

Q: What is the advantage of fieldbus systems when using sensors?

Lamb: Fieldbuses allow sensors to be mounted far away from the controller, reducing wiring. They can also transmit more information over fewer wires, so diagnostics also can be included with the signal.

Thompson: It depends on the application and the final goal. The proper platform architecture must be in place to support the communication, and the individual sensors installed must provide those capabilities as well. If starting from scratch, installation can require reduced control hardware, improved accuracy, and provide the end-user with options beyond just sensing (such as diagnostic and PID control functions). That being said, other options (e.g. HART) may be worthwhile to consider if starting with an existing 4-20 mA architecture.

Q: Can you discuss CO/NO2 sensors?

Lamb: I know nothing about these; I believe mass spectrography can be used.

Q: Does telemetry (wireless data collection) require a special sensor?

Lamb: Usually the sensor will be a standard dc sensor wired into a wireless device such as a wireless router or switch, or to an I/O device with wireless capability.

Q: How much more accurate is a 32-bit system than a 16-bit system?

Lamb: The highest resolution I have heard of using in an industrial environment is 16 bit. This could provide 65,536 different levels (32,767 for -10 to +10 V range where only positive values are used), which is generally more than enough. Most analog-digital (A-D) converters in standard industrial systems are 13-14 bits at best. If 32 bit A-D converters were used, it would provide over 4 billion values within the 0-10 V range, I’m sure noise would eat up most of the accuracy.

Q: Please explain about sensor redundancy, especially with temperature sensors. Should I use two for cost reduction or three to ensure a correct sensor reading?

Lamb: Redundant systems in the U.S. are used in case a sensor fails, rather than for accuracy. It is better to ensure the sensor is calibrated for accuracy than to introduce several temperature values and not know which one is correct. I would use two and ensure that they read the same, or as close as possible.

Q: I believe a current sensor is preferred over voltage sensors because of no signal loss and less noise, so when would it be appropriate to use voltage sensors?

Lamb: Most sensors use current for that exact reason. Voltage signals can generally be transmitted over a longer cable run. They are most commonly used for variable frequency drive (VFD) speed reference signals.

Q: For pure water, what are considerations in selecting the best flow measurement system?

Thompson: Pure water can really mean a number of different things, such as distilled, demineralized, and reverse osmosis (RO), so clearly defining the actual water properties is important. Correct selection of materials of construction is critical as pure water can be quite corrosive. Flow capacity, measurement range, and accuracy will also need to be identified. Coriolis meter, mechanical meters (e.g. turbine, positive displacement), rotameters are typical possibilities. Ultrasonic meters can be used, but may experience some issues without sufficient aeration (bubbles). Mag meters are ineffective in this application.

– Edited by Mark T. Hoske, content manager, Control Engineering, CFE Media, mhoske@cfemedia.com.

ONLINE extra

If you missed this webcast and would like to view it, it’s archived for one year after it’s Aug. 22, 2017, live broadcast. Find it and others at: Upcoming Webcasts

Control Engineering has pages for:

Sensors and vision 

Process sensors and actuators 

The New Products for Engineers Database has sensor-related product categories.