Applicability of ZigBee Technology to Electric Motor Rotor Measurements

Transcription

1 Applicability of ZigBee Technology to Electric Motor Rotor Measurements Ville Särkimäki, Risto Tiainen, Tuomo Lindh and Jero Ahola Department of Electrical Engineering Lappeenranta University of Technology P.O.Box Lappeenranta Finland Abstract The purpose of this work is to determine the applicability of the ZigBee technology to electric motor rotor measurements. Requirements for data transmission, electrical structure and powering of a sensor are discussed. A prototype wireless ZigBee-based torque sensor is built and tested. The results of the tests are analyzed. Index Terms wireless sensor, ZigBee, rotor measurement, sensor network W I. INTRODUCTION ireless technologies have gained foothold in many industrial applications, such as different sensors used in condition monitoring and maintenance as well as logistic applications. The cost and time for the installation and maintenance of cables can be substantially reduced when using wireless technologies. Retrofitting of cables is expensive and in harsh environments chemicals, vibrations, and moving parts exist that could potentially damage any sort of cabling. Mobility of machinery is increased and temporarily accessing any of the machinery in the plant for diagnostic or programming purposes can be done wirelessly by using mobile devices. [1] Also data transmission from rotating rotor or shaft of an electric machine is difficult to realize using wires. A possible solution is the usage of slip rings for data transmission, but again wireless can offer cheaper and simpler solution. Wireless data transmission in industrial is studied for example in [2]. Technologies such as IEEE b/g, Bluetooth and ZigBee have found their place in industrial applications. ZigBee is a low cost, low power wireless technology specially intended for medium data rate sensor networks. Its high density of nodes per network capability and simple protocol makes it suitable for sensor networks used for motor measurements according to [3]. In this article the applicability of ZigBee technology to rotor measurements is studied. In order to test the applicability of ZigBee, data transmission measurements are carried out from a rotating transmitter attached to the shaft of an electric motor. Also a prototype sensor for measuring torque from rotating shaft of an electric machine is designed and constructed. The sensor itself is attached on the shaft and is wireless, using ZigBee to transmit torque measurements to receiver. The receiver can be mounted for example on the stator of the electric machine or close proximity where connection to e.g. field bus is easily accessible. This kind of sensor has many advantages, it is easy to use and faster to install than wired sensor because no installation of wires is needed. Major advantages to use this kind of sensor are in applications where we require a torque or some other measurement, but no mechanical changes to installation are possible. For example measuring torque from cylinders of paper machine or from the generator located underwater in hydroelectric power plant. The remainder of this article is organized as follows. Section II introduces data transmission scheme and an introduction for ZigBee technology is given. Section III discusses about requirements of wireless sensor, such as reliable transmission and power issues. Section IV introduces the structure of sensor that is build to test the applicability of ZigBee for rotor measurements. Section V introduces tests that we carried out and the results that were achieved. The final Section summarizes the results of this work. II. STRUCTURE OF DATA TRANSMISSION In this chapter, simplified scheme for data transmission for condition monitoring of electric machines in industrial plants is introduced. It is based on the assumption that the lowest level area network is build using ZigBee-based sensors. The lowest level of the industrial plant information hierarchy is called field level and industrial processes as well as electricity distribution are located there. network can be thought to be a sublevel of field level. Upper levels from field level are process control level and management level where data from sensors is further processed and saved. In figure 1, overall picture of network hierarchy is given.

2 1 Information system of industrial plant Management level Database server Office workstation Local are network Gateway Process control level Control room Field bus they posses. As the name says RFD devices do not have all the operations that FFD devices have. This categorization makes possible to use simpler low-power microcontrollers for RFD devices and also makes power saving possible, because RFD devices need to be connected to the network only to check messages and transmit data. RFD devices can only work as end devices in a network. FFD device can work as a coordinator or router in ZigBee network. Possible network topologies in ZigBee are star-, cluster tree-, meshnetwork. ZigBee also supports networks as large as nodes. [5] Different network topologies as well as different roles of coordinator, router and end devices are shown in figure 2. Field level LAN (Ethernet) Gateway Field bus ZigBee Coordinator ZigBee Network Process M Fig. 1. Simplified picture of information system of industrial plant, showing ZigBee sensor network at the field level. network is used to collect measurements e.g. for condition monitoring purposes. s are attached on electric motors, pumps and other machinery where measurements need to be taken. One motor can have multiple sensors, so distances between sensors can be very short, less than one meter. However, all the sensors in field level can form together a large sensor network that covers the whole plant area. Some of the wireless transmitters can work as a gateway device between the sensor network and wired information system of industrial plant. So there is a need for expandable wireless transmission system, where transmission distances between two nodes can be something between 1-30 meters. If the wireless protocol supports routing of data from node to node, transmission distances that cover the whole plant can be achieved. There are also questions of transmission speed, reliability and power consumption when using wireless technologies; these issues are discussed in the next section. ZigBee is a wireless standard for low-power sensor networks based on IEEE Operational frequencies are 869 MHz, 915 MHz and 2.4 GHz. Transmission speeds at the physical layer are 20, 40 and 250 kbps, respectively. However, actual payload transmission speeds are lower because of the overhead from the protocol. Available frequency bands are divided into channels. There are 1, 10 and 16 channels available in the frequency bands mentioned earlier. ZigBee uses direct sequence spread spectrum technology (DSSS) where data to be send is first converted to symbols, then each symbol is mapped in pseudo-random 32-bit sequence which is then transmitted. [4] ZigBee devices can be divided in full function devices (FFD) and reduced function devices by what operations Star Mesh Cluster Tree Coordinator (FFD) Router (FFD) End device (RFD) Fig. 2. Different ZigBee network topologies, star-, mesh- and cluster tree network. III. REQUIREMENTS FOR WIRELESS ROTATING SENSOR The wireless sensor can be easily mounted to existing electric motors. When it s a part of larger wireless sensor network it can provide useful information from the specific motor or process in which the motor is part of. Requirements for this kind of wireless sensor are for example, reliable data transmission, small power consumption, easy of use and installation. One of the biggest problems in wireless data transmission its unpredictability, particularly in indoors. According to [6], in industrial environments general causes for reduced link quality are static multipath, time variant multipath, static and time variant interference. In the case of static multipath destructive interference can cause blackouts in transmission, if transceivers are located in particular positions. Moving obstacles such as humans and vehicles can cause time variant multipath. Interference from machines or other transmitters operating at the same frequency band, can block channels for a period of time or cause errors in data transmission. To overcome these problems, different technologies and techniques can be used. For example, protocols that support mesh networking can be used, so that data can be transmitted through alternative route if primary route is blocked. Spread spectrum technologies, such as DSSS, make the signal less vulnerable to noise and interference compared to the narrowband data transmission methods. ZigBee supports both mesh networking and DSSS. Power consumption is important aspect, because if the sensor is wireless, it must be operated with batteries or with

3 2 some alternative power source. There are three main parts in wireless sensor that require power; microcontroller, transceiver and actual sensor element. If measurements are needed only certain periods of time, some of these parts can be turned off during the measurements to save power. For example strain gages used for torque measurement can be powered only when measurement is taken. Same applies to temperature measurement. ZigBee also offers some means to save power, one of these is the previously discussed distribution to FFD and RFD devices. can work as RFD, so it is connected to coordinator only periodically to check messages and transmit data. During the sleep times, transceiver can be powered down and microcontroller can be turned to power saving mode. If operated with batteries, the operating time is limited and batteries have to be changed in the certain amount of time. In general there are alternative, sources of power available than batteries. There is possibility to harvest energy from the surrounding environment. These are discussed e.g. in [7]. Also there is means to transfer energy inductively to the sensor on the shaft. Torque measurement using strain gages is hard to implement, because sensor elements need higher voltages (10-20V, compared to 3.3V which microcontroller and transceiver needs) in order to get accurate measurements. Also more power is required. Basically this requires using e.g. slip rings to power the sensor. In other or temporary measurements that do not require so much power, such as measuring the rotor temperature, also batteries can be used. We might also have receiver unit and other sensors attached to the stator of the machine. These sensors also need power to operate. Without losing the advantages of wireless, we can probably in some cases use short wires to power sensors. In addition, in electric machines there are always power cables available. One solution could be harvesting inductively energy from power cables of electric machine and transferring it to the sensor located close to the power cables. This requires some additional circuitry and probably short wiring, but still installation is easier than expensive retrofitting of long power and data transmission cables. IV. SENSOR PROTOTYPE AND MEASUREMENT SETUP In the previous sections it has been made clear that our prototype of sensor uses ZigBee technology and overview of the data transmission was given. In this section general architecture of torque sensor is discussed. The torque sensor s main parts are transceiver and antenna, 8-bit microcontroller, amplifier, A/D converter and power source. The actual sensing elements are strain gages attached to the shaft. Also temperature sensor can be included for compensation purposes. Overview of the sensor is given in the figure 3. Inputs Amplifier 8-bit microcontroller A/D-converter Fig. 3. Overview of the torque sensor prototype. Transceiver Power source Radio transceivers in both sensor and receiver were Chipcon CC2420, which are fully IEEE compliant. Some of the parameters of the transceiver are in Table I, full description can be found from [8]. Inverted F- antenna manufactured on a circuit board was used. This allows sensor to be constructed in a compact size and radiation pattern is close to omnidirectional so direction of the antenna is not critical. However, because of the low transmit power and antenna construction long transmission distances aren t possible. Maximum distances that our sensors are capable of are less than 30 meters. TABLE I. PARAMETERS OF CHIPCON CC2420 TRANSMITTER [7]. Operating frequency MHz Nominal output power Programmable, maximum 0 dbm Sensitivity Typical 95 dbm Transmit bit rate 250 kbps Current consumption RX: 18.8 ma, TX: 17.4 ma Supply voltage V, on chip regulator Physical size 7 x 7 mm, QLP-48 package Microcontroller is running Microchip s ZigBee stack. This protocol stack in current version V3.3 is based on the ZigBee specification 1.0. All the functionality of ZigBee is yet not implemented, for example, some of the security related functions are unavailable. However current version has functionality for end devices, router and coordinator and all the network topologies are implemented. Specific application was programmed on top of ZigBee stack for sensor and receiver unit. To test the transmission, application that transmits 200 packets of payload of two bytes without acknowledgment was programmed. This simulates transmitting 200 torque measurements. Final application for torque measurement is similar, but measurements are taken only from a request from the coordinator. In our implementation measurement setup includes torque sensor, working as an end device attached to the shaft and receiver working as a coordinator located near stator of the electric motor. is battery powered, which suits for temporary measurements. Coordinator is the mains powered and connected through the RS285-bus to the computer. Connections to different buses can be easily constructed e.g. Modbus and Ethernet. V. RESULTS OF LABORATORY TESTS Measurements were carried out at the power electronics laboratory at Lappeenranta University of Technology. In a viewpoint of radio communications there were some sources of disturbance, such as other wireless transmitters

4 3 using the same frequency band in the laboratory. There were also metal structures and machinery nearby that can cause multipath propagation. The call of the test is to determine operation of the data transmission when sensor is attached on the rotating shaft of an electric machine. Modeling of the transmission channel is not an easy task, but we can define some properties of the channel. Small-scale fading is used to describe the rapid fluctuation of the amplitude of the received signal over a short period of time or distance. The place of the transmitter is constantly changing, as a function of time, due the rotation. This causes small-scale fading effects, such as, rapid changes in signal strength, random frequency modulation due to varying Doppler shift and time dispersion caused by multipath propagation delays [9]. Multipath propagation is illustrated in the case of rotating sensor in figure 4. Radio waves arrive in different paths with different delays to the receiver due to reflections from obstacles nearby. c Receiver b Obstacle a v Rotating shaft Fig. 4 An example showing multipath propagation when sensor is rotating. The small-scale variations of a transmitted signal can be directly related to the impulse response of the channel. Channel can be modeled as a linear filter with a time varying impulse response, where the variation is due to the sensor rotation. If we assume that the rotational speed is constant over a short time interval we can express the received signal y(t) as a convolution of the transmitted signal x(t) and impulse response of the channel h(t,τ). The variable t represents the time variations due to motion, whereas τ represents the channel multipath delay for a fixed value of t. [9] nature of the channel can be investigated by calculating Doppler spread and coherence time [9]. Doppler shift is given by f d v f d = λ cosθ, (2) and λ = c. f c (3) Where v is the speed of the sensor and θ is the angle between the direction of motion and the direction of arrival of the wave. f c is the carrier frequency. Because of the rotation of the sensor, received spectrum will have components in the range f c -f d to f c +f d. Now if the f m is the maximum Doppler shift, given by Eq. 2. Coherence time can be calculated as T 1 C f. (4) m If coherence time T C of the channel is bigger than symbol period, T S << T C (5) we can conclude that a signal undergoes slow fading. In this case the effects of Doppler spread are negligible at the receiver [9]. Transmission speed for ZigBee at 2,4 GHz are symbols/s, so T S =0,016 ms. In tests maximum radial speeds of the sensor are 14 m/s. Calculating coherence time from Eqs. 2, 3 and 4 we get T C 8,9ms. According to the Eq. 5 the effect of Doppler spread is negligible. Data transmission from torque sensor was tested by attaching sensor to the shaft of an electric machine. Data packets were send and packet loss was calculated. Network formation was star-network and location of the receiver was varied. Effects of different radial speeds were tested by using two different size electric motors with different size shafts and running motors at variable speeds. Figure 5 is from one of the test setups where torque sensor is rotating on a shaft of a 15 kw induction machine. y ( t) = x( τ ) h( t, τ ) dτ = x( t) h( t, τ ) (1) Delay spread and coherence bandwidth are parameters, which describe the time dispersive nature of the channel in a local area. These are depended on the surroundings where transmitter and receiver are located. However, they do not offer information about the time varying nature of the channel caused by rotation of the sensor. The time varying Fig. 5. Test setup, where ZigBee sensor is attached on the shaft of an electric machine.

5 4 Some of the results from the measurements from testing the data transfer are presented in Table II and III. In the first measurement, presented in Table II, sensor is attached on the shaft of 15 kw an induction motor. Receiver is placed axially 30 cm in front of the shaft end at the same height as the shaft. Radial speed is the speed of the antenna on the sensor. In every measurement 200 packets of payload of two bytes was sent, without acknowledgment. The Table II also presents average arrival speed of a packet and data transmission speed. Transmission speeds are low due to the overhead of the packet and small payload. In the second measurement, presented in Table III, the setup was similar to the first measurement expect that the receiver was located radially on the side of the shaft. Height of the transmitter and receiver was same and distance was 30 cm SM Poles 14,00 12,00 10,00 8,00 6,00 4,00 2,00 0,00 2 Radial speed [m/s] TABLE II. RESULTS FROM THE DATA TRANSFER TEST AT 15 KW INDUCTION MACHINE. RECEIVER LOCATED AXIALLY FRONT OF THE SHAFT. Number of Radial speed [m/s] received packets Avg. arrival Time [ms] Transfer speed [b/s] 1, , , , , , , , , ,0 362 TABLE III. RESULTS FROM THE DATA TRANSFER TEST AT 15 KW INDUCTION MACHINE. RECEIVER LOCATED RADIALLY ON THE SIDE OF THE SHAFT. Radial speed [m/s] Number of received packets Avg. arrival Time [ms] 1, , , , , , , , , ,0 318 Transfer speed [b/s] Similar tests were carried out changing the location of receiver and with larger machines. At the larger machine, radial speeds of 14 m/s seconds were achieved and the transmission was found to be working similarly as results shown in tables 1 and 2. These results were very promising and we can conclude that ZigBee is working fine in this kind of data transmission. In the figure 6 are shown couple of different IEC frame sizes of induction motors and their shaft radial speeds. Radial speeds were achieved by looking from datasheet shaft diameters and calculating radial speeds at the nominal speed of induction machines from the [10]. This figure is to show radial speeds of induction machines of different frame sizes. It should be noticed that the antenna is not necessarily located close to shaft, but probably little higher because of the sensor construction. In this case radial speed of the antenna is higher than radial speed of the shaft. Fig 6. Radial speeds of shafts of induction machines of different IEC frame sizes. CONCLUSION Wireless sensors will certainly have place in industrial plants. Fast and easy installation is just one of the points that speak for them. However, there are some problems to solve. Powering issues and reliability of data transmission are some of these and lack of suitable standardized wireless technology has been a big one. In this article promising wireless technology ZigBee was tested in operating scheme where wireless torque sensor was constructed as an example to show how wireless data transmission can give advantage over wired data transmission. According to results data transmission from rotating sensor works fine. Errors in data transmission were small and achieved transfer speed was more than was needed. Similar rotating sensors can be constructed to measure e.g. temperature directly from the shaft or rotor of an electric motor. REFERENCES [1] A. Willig K. Matheus, A. Wolisz, Wireless Technology in Industrial Networks, Proceedings of the IEEE, vol. 93, No. 6, June 2005, pp [2] T. Brooks, Wireless technology for industrial sensor and control networks. for Industry, 2001, Proceedings of the First ISA/IEEE Conference, 5-7 Nov Pages: [3] Patrick Kinney, ZigBee technology: Wireless Control that Simply Works, White paper, [4] IEEE Standard , Standard, [5] ZigBee Alliance, ZigBee Specification, version 1.0, Specification, June 2005, [6] Jay Werb, Michael Newman, Victor Berry, Scott Lamb, Daniel Sexton, and Michael Lapinski, Improved Quality of Service in IEEE Mesh Networks, Proceedings of the International Workshop on Wireless and Industrial Automation, San Francisco, California, March 7, 2005 [7] Edgar H. Callaway, Wireless Networks : Architectures and Protocols, ISBN , CRC Press LLC, [8] Chipcon, CC2420 datasheet, [9] Theodore S. Rappaport, Wireless Communications, Principles and Practise, second edition, Prentice Hall PTR, ISBN [10] ABB, Low and High Voltage Process Performance Motors, Catalogue,