RS-485 COM20020 RS-422 24AWG 75176B RJ-11 S09-300-444 S39-102 RS485 75LS176

Rev. E, May 1994 RS-485 CABLING GUIDELINES FOR THE COM20020 UNIVERSAL LOCAL AREA NETWORK CONTROLLER (ULANC) AND EXPERIMENTAL

TECHNICAL NOTE 7-5 Rev. E, May 1994 RS-485 RS-485 CABLING GUIDELINES FOR THE COM20020 COM20020 UNIVERSAL LOCAL AREA NETWORK CONTROLLER (ULANC) AND EXPERIMENTAL PROCEDURE FOR VERIFICATION OF RS-485 RS-485 CABLING GUIDELINES INTRODUCTION TO EIA RS-485 RS-485 EIA RS-485 RS-485 is a specification for the support of a multi-drop differential serial digital data network. RS-485 RS-485 came about as microprocessors and the use of distributed intelligence became popular in the design of industrial systems. The implementation of such concepts created a demand for a standard method of communicating serially in such environments. The actual RS-485 RS-485 specification came about as a result of the shortfalls of the RS-422 RS-422 standard. RS-485 RS-485 is virtually identical to RS-422 RS-422 except in two respects, increased receiver sensitivity and support of longer line lengths. The basic RS-485 RS-485 specification standardizes the electrical characteristics of each transceiver and provides some basic guidelines for establishing a network. By definition, a basic transceiver shall have a minimum input resistance of 12 Kohms and handle a +/-7V common mode voltage regardless of whether power is applied or not. When the transceiver is powered it must present the minimum input resistance and present less than 50pf of capacitance at its input terminals. In addition, each driver must be capable of providing a minimum level of 1.5V in the presence of 32 transceivers and two 120 ohm terminating resistors. The 120 ohm termination results from the use of twisted pair cable, the preferred media in many industrial applications because of its wide availability and low cost. Another critical requirement of RS-485 RS-485 is that the receiver must be capable of detecting levels down to 200mV which is of great advantage when long line lengths are needed. RS-485 RS-485 does not specify a modulation method or a maximum data rate. This gives the system designer great flexibility in creating a low-cost high-performance network. The combination of long line length, high node count (32 nodes), and support of low-cost media has made the RS-485 RS-485 specification the primary choice as a data communication standard in industrial applications. CABLING GUIDELINES FOR RS-485 RS-485 INTERFACE WITH THE COM20020 COM20020 The following cabling guidelines provide a basis for establishing a low cost Local Area Network (LAN) based on the ARCNET protocol for use with the COM20020 COM20020 Universal ARCNET Controller with a differential RS-485 RS-485 driver. The guidelines presented are for unshielded twisted pair cable modulated with the COM20020 COM20020's backplane encoding scheme. All testing and experiments were performed using a 24AWG 24AWG copper twisted unshielded 2 pair cable with a characteristic impedance (Zo) of 120 Ohms. The topology used in all experiments was a daisy-chained configuration with no stubs (i.e. no drops). TRANSMISSION LINE EFFECTS IN LOCAL AREA NETWORKS Transmission line effects often present an obstacle in obtaining high performance in data communication networks. Among the problems that plague high data rate LAN's are reflections, signal attenuation, and D.C. loading. Taking into account all three parameters when designing a network can result in a faster and more reliable network. A) REFLECTIONS IN TRANSMISSION LINE A reflection in a transmission line is the result of an impedance discontinuity that a travelling wave sees as it propagates down the line. To eliminate the presence of reflections from the end of the cable you must terminate the line at its characteristic impedance by placing a resistor across the line as shown in Figure 1. TERMINATION OF A NETWORK - Reflections are caused by a discontinuity in the line DIRECTION OF PROPAGATION INCIDENT WAVE Z = 120 ohm SOURCE Z = 120 OHM REFLECTED WAVE R = 50 OHM Z = 120 ohm DISCONTINUITY PROPERLY TERMINATED NETWORK INCIDENT WAVE DIRECTION OF PROPAGATION Z = 120 ohm R = 120 OHM DRIVER R = 120 OHM Z = 120 ohm Figure 1 It is important that the line be terminated at both ends since the direction of propagation is bidirectional. In the case of unshielded twisted pair this termination is 120 ohms. Note that all reflection measurements were made with no stubs on the network (i.e. no drops). Theoretically, a properly terminated transmission line would produce no reflections at all. However in a real network, small reflections are produced since the characteristic impedance of the cable cannot be met exactly due to variance in the manufacturing process of the cable. Another primary source of reflections is the impedance mismatch between a data transceiver and the line, which can cause problems on a data network by creating perturbations on the line during an otherwise idle state. Reflections affect the network by triggering false transitions (bits) on the line receiver's input translating into possible framing errors and CRC errors. A measure of the relative strength of the reflection generated by discontinuities along the line is called the Reflection Attenuation Factor (RAF). This is a measure of the strength of the reflected wave to its incident wave. The RAF can be obtained by comparing a reflected wave to its incident wave. The magnitude of the reflected wave can be measured by sending a burst of sine waves down a transmission line and observing at the sending end the magnitude of the wave after the burst has ended (see Figure 2). The reflection measured at the sending end is the reflection generated by a discontinuity at the receiving end of the line. The magnitude of the incident wave can be measured at the receiving end of the cable, since this is the wave from which the reflection is generated. It is important to compensate for line loss when measuring the reflected wave because the reflected wave, measured at the sending end of the cable, has lost some amplitude due to line loss. MEASURING REFLECTIONS IN A NETWORK Z = 120 Ohm Z = 120 Ohm 2.5Mhz SINE WAVE 10 NODES Z = 120 Ohm R T = 120 Z = 120 Ohm MEASURE INCIDENT WAVE HERE MEASURE REFLECTION HERE Figure 2 2 Measurements were made for twisted pair cable and are summarized in Table 1. The following relationship was used in calculating the Reflection attenuation. Reflection attenuation = 20 log (Vref/Vinc) where Vref = reflected voltage (compensated for loss) Vinc = incident voltage measured at receiving end of line REQUENCY Reflection Attenuation Table 1 625 MHz 1.25 MHz -33.19 dB -28.89 dB 312.5 KHz -35.12 dB 2.5 MHz -24.52 dB 5.0 MHz -17.83 dB These above numbers can be interpreted as follows: Assume a +5vp-p incident wave at 2.5MHz, a reflected wave will be generated that travels to the incident source at an amplitude of: -24.52dB = 0.059 therefore the reflected wave is 0.059 * 5V = 0.297V. In practice, the amplitude of the reflected wave might be smaller because discontinuities are generated throughout the line and are not in phase with each other, thus providing a canceling effect and decreasing the magnitude of the reflection. There are several methods for minimizing the effects of reflections such as squelch circuits and D.C. biasing. For the small reflection levels observed during experimentation, the recommended choice is D.C. biasing for its simplicity and minimum parts count (2 resistors). Biasing the network may cause some duty cycle distortion but the ARCNET protocol is insensitive to duty cycle or jitter. The biasing network will be discussed later in this guide. B) SIGNAL ATTENUATION IN TRANSMISSION LINES A second transmission line effect that has a bearing on the performance of LAN's is signal attenuation. A transmission line can be modeled as a combination of the distributed capacitance of the line, the distributed inductance, and resistance (see Figure 3). TRANSMISSION LINE MODEL R' C' ZO G' R' Z = O R' L' C' G' R' L' L' L' R' = UNIT RESISTANCE OF LINE L' = UNIT INDUCTANCE C' = UNIT CAPACITANCE G' = UNIT ADMITTANCE L' C' Figure 3 The capacitance of the line is formed by the parallel conductor pair. At the distances used in LAN's (100's feet), the resistance of the cable is negligible and contributes very little to line loss. The majority of line loss comes from the LC combination that acts like a low pass filter and tends to attenuate the signal as frequency and distance go up. For twisted pair cable, the attenuation rate is given in Table 2. These are measured values. FREQUENCY Attenuation per 100ft. Table 2 - Signal Attenuation 312.5 KHz 625 KHz 1.25 MHz -0.4 dB -0.6 dB -1.0 dB 3 2.5 MHz -1.3 dB 5.0 MHz -2.0 dB C) D.C. LOADING IN RS-485 RS-485 NETWORKS The third parameter that affects network performance is D.C. loading. The D.C. load presented to a line driver is a combination of three parameters: 1) loading effect of termination resistors 2) loading effect of biasing resistors 3) loading effect of RS-485 RS-485 transceivers EIA RS-485 RS-485 specifies that a line driver must be capable of presenting a 1.5V signal differentially at its outputs under the loading of 32 receivers and two 120 ohm termination resistors. Each receiver or passive transceiver is to provide a minimum input impedance of 12K ohms. The total parallel combination load impedance is 51 ohms, which includes the receiver load, termination resistors, and biasing resistors. Since the worst-case specification is restrictive, it may not be practical to design to this worst-case because, quite often the typical RS-485 RS-485 transceiver can drive substantially more than the worst-case load. Typically, many RS-485 RS-485 drivers can drive quite a bit more than the 51 ohms, sometimes as low as 20 ohms. If typical characteristics are used, a network can be composed of many more nodes than the 32 specified by EIA RS-485 RS-485. From laboratory experience, SMC recommends using typical transceiver characteristics when determining the D.C. load. In very high reliability applications (i.e. medical electronics, avionics), SMC recommends the use of the worst case RS-485 RS-485 parameters. The D.C. biasing of the network is essential to providing reliable operation at high data rates. The D.C. bias offsets the NULL voltage of the network when all RS-485 RS-485 transceivers are in their TRI-STATE mode. A small offset is needed because small reflections will occur that will cause spurious transitions on the RS-485 RS-485 receivers input. This would cause the COM20020 COM20020 to receive an undesired extra bit, thus causing framing or CRC errors and corrupting the data. It is only the negative half of the reflected wave that is of concern, so the effective reflection is .15V. The largest reflections seen on the network are approximately 0.3Vp-p differentially. This is more than enough voltage to trigger a transition on the receiver's input since the receiver has a hysteresis of 50mV and is thus quite sensitive (200mV differential sensitivity). By installing the biasing network of Figure 4 at each node, the network will be offset enough to keep the network in an inactive state. The +0.35V differential offset provided, will keep the line in an inactive state even in the presence of a 0.3Vp-p reflection (see Figure 5) which could cause a false bit to be recognized by the 20020 and cause data rejection by the receiving node. SMC recommends a pull-up and pull-down resistor (see Table 3 for correct value) in the circuit diagram shown in Figure 4. BIASING NETWORK RS-485 RS-485 TRANSCEIVER +5V R1 HI LO TO NETWORK R2 Figure 4 If a larger offset is desired, the offset voltage can be calculated as follows: To calculate the correct biasing voltage, the maximum reflection should be known: From laboratory measurements maximum Vref = 0.3Vp-p. We are only concerned with the negative portion of Vref = 0.15V so Iref = 3.33ma To get the proper bias resistor values the following relationship can be used: +5V = Ibias(Rpull-up + Rpull-down + (Rt1 || Rt2) This yields a bias resistor value of 720 ohms for the entire network which gives a bias voltage of 0.35V. EFFECTS OF BIAS ON REFLECTIONS DIFFERENTIAL SIGNAL W/ NO BIAS DRIVER TRI-STATES +V 0V -V REFLECTION OCCURS CAUSING FALSE TRANSITION (.3V) DIFFERENTIAL SIGNAL W/ 720 OHM BIAS RESISTORS +V DRIVER TRI-STATES .35V OFFSET -V REFLECTION OCCURS RAISES REFLECTION TO .1V (.3V) Figure 5 In practice, it is better to provide a larger bias resistor at each node so that the parallel combination of all nodes can provide a proper offset voltage. A lower resistor value than the suggested may be used if a large amount of noise is present in the system or if larger reflections are encountered. It should be noted, that the biasing arrangement adds to the total D.C. loading of the network, and yields a total load of Rtotal = Rt1 || Rt2 || (Rrcvr/N) || (Rbias/N) where Rt1 = Rt2 = termination resistance Rrcvr = receiver impedance Rbias = bias resistance N = Number of Nodes CALCULATING THE NUMBER OF NODES AND LENGTH OF CABLE Three parameters must be taken into account when deciding upon the configuration of the network (i.e. line length and number of nodes). These parameters are: 1) D.C. load 2) Cable attenuation 3) Noise margin The first two parameters have been previously discussed and the third parameter, noise margin, will be addressed now. The noise margin is the minimum voltage above the 200mV receiver sensitivity limit prescribed by EIA RS-485 RS-485. All further calculations will assume a 0V noise margin. Your individual application may require a greater noise margin then that prescribed by the EIA RS-485 RS-485 specification. The following relationship may provide Vend = .8(Vdriver - Vloss - Vnoise - Vbias) where Vend = voltage at end of line 5 Vdriver = driver output voltage Vloss = voltage loss due to cable Vnoise = noise margin Vbias = D.C. bias voltage applied to network at each node (typically .4V) .8 = Derating for cable tolerances (± 20%) The driver output voltage is a function of the D.C. load presented to the driver by the network and can be calculated as described above. For convenience, Table 4 contains the number of nodes vs. D.C. load using the typical transceiver characteristics for the 75176B 75176B RS-485 RS-485 transceiver. Vloss can be found by determining the driver output voltage for a given load and using Table 2 to calculate the line loss for a given distance and frequency. Usually, either the number of nodes required or the maximum cable distance are known prior to designing a network. By using the above relationship all the remaining variables (i.e. number of nodes or maximum cable length) can be determined and combined to implement a working network. BIASING THE NETWORK From experimental results, a good value for biasing resistors is 810 ohms. This value can be implemented in two ways. One way is to provide one set of resistors for the entire network with a separate power supply for the bias. The second option is to provide separate bias resistors at each node, thus providing greater flexibility in adding nodes and increased reliability. This has the effect of increasing bias as nodes are added, automatically compensating for the additional reflections generated, but decreasing the dynamic range due to loading. Table 3 shows the recommended values for different numbers of nodes. Note that no one value is optimal because as the node numbers change the loading of the network will increase due to biasing. Table 3 - Bias Resistance vs. Number of Nodes for 2.5 Mbs NUMBER OF NODES BIAS RESISTANCE 1 - 10 2.7K 11 - 20 12.0K 21 - 30 18.0K 31 - 40 27.0K Table 4 contains the estimated load and corresponding driver output voltage based on the typical characteristics of the 75176B 75176B transceiver and the bias resistance from Table 5. Note that these are typical characteristics and your driver may vary significantly from these numbers. Note: SMC recommends locating the bias resistors at a single location if possible as shown in Figure 6. 470 1 MASTER 2 470 Figure 6 Table 4 - D.C. Load Table ESTIMATED VOUT NUMBER OF (Volts) TOTAL LOAD NODES (Ohms) 1 57.41 2.5 5 48.97 2.4 10 48.70 2.4 15 52.17 2.4 6 3 NUMBER OF NODES 20 25 30 35 40 TOTAL LOAD (Ohms) 50.00 49.65 48.00 47.89 46.55 ESTIMATED VOUT (Volts) 2.4 2.4 2.4 2.4 2.3 Table 5 takes into account line loss and shows Table 4 with the maximum distance that the network can be driven using the various data rates supported by the COM20020 COM20020 (i.e. 312.5Kbs, 625Kbs, 1.25Mbs, 2.5Mbs). A 0.1 volt noise margin was used and a 0.200V receiving end voltage were used in establishing the criteria for cable length. The maximum distance driven is relatively independent of load for the node numbers of interest (