y^-^^. "'^^""^"c>^" ,».. ■'^^.*""°\\^'' ,,„ °'^^'''''" .^f" ,.. S",* V " "; 'NO- V^ ^"-^^^ ^1 ^' A ^. 'o-,7' ,0^"^ '\/"'^^'*a'^ '^<.'''o^^*'_,&^'^ \>J^^^'\^'^ %^'°^S^\g«-'^^ ^"^^ ^v ■<<. o/ '^^. "• is} <"">. %> o > a^' ,-^^ ..1^^^/^^ -^a^ ^°--^. <■, ..." ^VJ 'Key %. "^ ^f° 5^°-^. .0^ ..!°.i.'. ^> (^ ■*» V ;•- v..**' .•^■•/%..*\.i&*;-/V..**'/aK'- '*-.,>* \-;«S&-.X./' V^^ vr'\^ 3 V 3 -VS. ''• %/ A >' ^^ (P- * o « o \ A-^' V o V :* ^ '^<'. c-?^" ^'^m^' t<. ^^ V . K 1 • ' .* .0^ ^°--^ -^o^ J^ o ^°-^^. t-o^ x«>^ ..^'N o V - '^ -r. '- -^^0^ o^ ^■^ /\j 0" o ^ 'O A'' .^'o., <^^ *oV" ^°-^. -*^ Q V ^ o * O c <^°^ • **,.<-'^ .-iSM-. ■\,^* /flife'-. **.>* -^V*-- "■^" -'* V\^ ■ ' ° '^c^fa ' : < e / T " J, " ' ^ .0* »lVL'» V •r.. ^:-^';y' ^ -^^0^ »«$^Ei- -^-oV^ :mi^'r.\ ^^^^ l^^%.,<^^''^.iVA^'\ .<^'*^'' o . » » -G '-^ * »'»»'* ^^■^ life ^^-n^ t>^ o_ •. ^v ... °^/ "■>-'<>' -0 -^o^ o^ 'o . t « A v^ .»^:i^'*. c^^ ^^0 ^•>^a.^^' <> *'..'^ ..0 ~^n c-?^" »'^^#'a:-. "^^ .5.^ ' »'«^1K'. \. ^^ \<^' -oV" ^^-n*.. C" * ■'U-o^ o. 'o.'.' ^A <. -^o^ ° ^^-^^^ "^z ■« ^^0^ ^-lo^ \'^^^l ^t^. -^i^^^km- .-iq ^ aV -^^ -% ••-•■ ^<^ .c°\^' \- ' < /. •^^^^ 3, v* . y -^%^m^.^ji.- -%''f.r'\^'' 'V^'^I^Vo^^ '%, C° *' ;* JO %.^' «. '^,s*P3^!ks'" "O. a"^ •• — •» O ^"""^^ .^ »!.*_°'/'> «^ ""I- .■?.^ -^*^ .l^t^-^ iO-jS. •^_ V .^':^'* > c^\c^^% 'h. '* "^0 0^ cl'f' ^?- '•.,^'> . r.s* ,0 .v.^^'\'^^Wv^^'% • ■.^ rV "O. 'o ^\ 4 • • ;* ^^^ '•"' ^"^ <. *'T.s* ,0 .* <.^' "^i ^°-^^. v .^ij;^'* <^ •^^ A^ »^fQe^'. ''^^?>. A"^^ /, ^ A^^ '*4 <^^ * " "-^--0^ c° . 'bt 'b 'o . . " A <* */T. .^%. .,-r , ^-^^ ^°^^^'> A/^•^iX ^°^:^^'> •^0^ 4{5^ - •■--./'' " V^'^O-^ ^ X'*"^"^^*,/^ "o^ '*^'* ^0-^ ^^^ ' • - • ^^■ ^oV 4?-n^. 7 -^o^ b^ l-^ "-.,^^ .^ "v.. <.^''\ '°}^W: A^% ^^..^^ IC ^°^^ Bureau of Mines Information Circular/1985 A Low-Cost FSK Modem Network for Polled Communication Systems By Richard A. Watson UNITED STATES DEPARTMENT OF THE INTERIOR 1751 'Wines 75TH ax^"^ i Information Circular 9021 A Low-Cost FSK Modem Network for Polled Communication Systems By Richard A. Watson UNITED STATES DEPARTMENT OF THE INTERIOR Donald Paul Model, Secretary BUREAU OF MINES Robert C. Horton, Director Libtaty of Congress Cataloging in Publication Data: no. ^0^1 Watson, Richard A., 1948- A low-cost FSK modem network for polled communication systems. (Information circular ; 9021) Includes bibliographical references, Supt. of Docs, no.: I 28.27:9021. 1. Mine communication systems. 2. Modems. I. Title. II. Title: Low-cost F.S.K. modem network for polled communication systems. III. Series: Information circular (United States. Bureau of Mines) ; 9021. -T! < CONTENTS Page Abstract I Introduction 2 Purpose and requirements 2 General description 3 Theory of operation 4 Basic overview 5 LSM operation 6 RCCM operation 9 Other applications 11 Conclusion. 12 Appendix A. — Frequency selection and filter design 13 Appendix B. — Modulation design and calibration 18 ILLUSTRATIONS 1 . Modem frequency spectrum 4 2 . Network diagram 4 3 . LSM block diagram • 3 4 . RCCM block diagram 6 5 . Waveform timing diagram 6 6. LSM schematic 7 7. Line splitter schematic. 9 8. RCCM schematic 10 A-1. Chebyshev bandpass filter 14 A-2. Filter response 16 A-3. RCCM filtered response 17 B-1. Demodular circuit 18 B-2. Modulator circuit 19 B-3. RCCM calibrator schematic 21 B-4. LSM printed circuit layout 22 B-5. Line splitter printed circuit layout 23 B-6. RCCM printed circuit layout 23 UNIT OF MEASURE ABBREVIATIONS USED IN THIS REPORT Bd baud (bits per second) Mfi megohm dB decibel ms millisecond dB/mi decibel per mile yF microfarad ft foot ys microsecond h hour a ohm Hz hertz pF picofarad kHz kilohertz V volt kn kilohm w watt MHz megahertz A LOW-COST FSK MODEM NETWORK FOR POLLED COMMUNICATION SYSTEMS By Richard A. Watson' ABSTRACT A frequency-shift keying (FSK) modulator-demodulator (modem) network has been devised for the Bureau of Mines mine-monitoring systems. This network permits an unlimited number of remote stations to communicate with one central station over the same pair of wires. The network oper- ates at a speed of 4,800 Bd at distances up to 5 miles. This report describes the development and operation of the modems. The theory of operation, schematic diagrams, and printed circuit board layouts are provided. Although the modems were built for a specific ap- plication, a more general use is discussed whereby the network would op- erate with a typical serial channel format. 'Electrical engineer, Pittsburgh Research Center, Bureau of Mines, Pittsburgh, PA. INTRODUCTION The mining industry is finding that the use of monitoring systems can increase safety and production. Remote monitoring gives management real-time information about the underground environment and equipment status. It also provides the ability to remotely control equipment such as belts and pumps. With this in- formation at hand, hazardous conditions and equipment failures are quickly recog- nized and appropriate actions are more immediate. As the benefits from these monitoring systems become documented, their growth can be expected to increase; with this, the distance and number of monitoring and control functions will increase. Two im- portant criteria must be considered: system speed and data security. The Bureau has researched monitoring systems for the purpose of establishing performance guidelines. Out of this re- search, a special FSK modem has been de- vised for use in the Bureau's monitoring systems. This modem permits operation over standard telephone wiring at dis- tances up to 5 miles, with no limit on the number of remote outstations. It maintains the high speed of the baseband link used for shorter distances and pro- vides for lower error rates. Although this modem was developed for a special application that may be limited, another application has been considered. It entails modifying the modem to operate with a typical asynchronous 11-bit serial channel. This modification is being tested and is discussed in some detail in the section "Other Applications." PURPOSE AND REQUIREMENTS The existing communication link used in the Bureau's monitoring projects consists of baseband signals transmitted over a special heavy conductor cable. Two of these cables are used in the intrinsi- cally safe mine-monitoring system (ISMMS) installed at the Lucerne No. 8 Mine, Clarksburg, PA. 2 Each cable is used to communicate with and power environmental transducers. When expansion of the belt-monitoring system was planned, another method of communication to the underground was de- sired. Modems were considered, because they would employ available pairs of wir- ing in the underground telephone network. This would eliminate the costs and ef- forts of installing more cable. The spe- cial cable was not required since belt monitoring was not included in intrinsic safety operation. ^Watson, R. A. An Intrinsically Safe Environmental Monitoring System for Coal Mines. Paper in Proceedings of the Sixth WVU Conference on Coal Mine Electrotech- nology (Morgantown, WV, July 28-30, 1982). WV Univ., Morgantown, WV, 1982, pp. 345-360. There are about 15 belt drive control- lers in the Lucerae No. 8 Mine. With the addition of several pump-monitoring sta- tions, it was obvious that the number of remote modems should not be limited. Each remote modem must connect with the local modem at the central station on the surface. The modems should be transpar- ent to the monitoring system, receiving and transmitting compatible signals at each end. The picture that was rapidly developing was of a system similiar to that of local networks. With local networks, many us- ers have the ability to share the re- sources of a centralized computer via a common communication link. Some networks use carrier sense multiple access with collision detection (CSMA/CD) and rely upon remotes' transmitting on a random basis, with provisions for detecting col- lisions. Other networks use a token passing scheme where information is sent around a loop and each station takes or injects what it needs. These, as well as other types of networks, are complicated, costly, and not directly compatible with the polling scheme used in the Bureau's monitoring systems. In order to have an unlimited number of remote modems connected onto the same line, each remote must be seen as a high impedance. The receiving section should be high-impedance-buffered, and the transmitting section should be discon- nected when not required to transmit. The line is then terminated, at the re- spective ends, with a resistance equal to the characteristic impedance of the line. Both local and remote modems should be guarded against electrical noise, spikes, and surges on the line. The degree of protection should be high, as mine elec- trical disturbances are known to be con- siderable. Additionally, the local modem connection to the monitoring system should be optically isolated from lines that enter the underground. GENERAL DESCRIPTION An FSK modulator converts binary digi- tal values, representing logical "ones" (1) and "zeros" (0), into two frequen- cies. They are known as the mark and space frequencies, respectively. The de- modulator reconverts these frequencies into digital values at the receiving end. Exar Integrated Systems, Inc.,3 manufac- tures integrated circuits (IC's) that do the conversions with a small number of external components. Two of these IC's were selected because of their low cost, ease of use, and well-documented applica- tion notes. They are the XR2206 modula- tor and XR2211 demodulator. The pair can accommodate signals over a 60-dB dynamic range of frequencies as high as 300 kHz. The modulator and demodulator IC's are used in both the local and remote modems. In the local modem, the carrier is always on. A crystal-controlled clock synchro- nizes timing with the transmitted message and "opens" the receiver when the receiv- ed message is due. This modem will be referred to as the local synchronous modem (LSM) . In the remote modem, the carrier is turned on only when activated by the station responding through the modem. This modem will be referred to as the remote carrier-controlled modem (RCCM). Although the mark and space frequencies of the modem IC's cover a wide range, the choice is not completely arbitrary. ^Reference to specific manufacturer or trade names is for identification only and does not imply endorsement by the Bureau of Mines . Selection of these frequencies is based upon the channel bandwidth, line attenua- tion, and the system baud rate. Addi- tionally, if a full duplex channel (simu- ltaneous transmission in both directions at the same time) is selected, the mark and space frequencies of the two groups must be separated by enough distance to prevent interference. This separation is determined based on the filtering re- quirements of each group. Figure 1 is an illustration of the two groups of frequencies. The mark and space frequencies transmitted from the LSM are fl and f2, respectively. Those received from the RCCM are fj, and f/^. The selec- tion of these frequencies and design sol- utions are covered in the appendixes; the frequencies are given here as 12.6, 16.2, 23.4, and 27.0 kHz, respectively. The selection avoided too high a frequency where line attenuation is greater and line balancing and matching become important. Separate pairs of wires are used for the transmit and the receive frequencies. A line hybrid could combine the two groups onto the same line but would be incompatible with the switching technique of the transmitter. A hybrid is used to balance a line and prevent the transmit- ted output from feeding back into its re- ceiver input. The balance is based on a fixed line impedance as seen by both the output and input and is used where modems are connected one to one. When many re- mote modems are connected to the same line and are being switched on and off, the line impedance is not constant. f '2 '3 FREQUENCY SPECTRUM FIGURE 1. - Modem frequency spectrum. (H(jw) = frequency response of a network function; H = frequency response in the bandpass.) With separate transmit and receive lines, it is not neccessary that the two groups of frequencies be separated, so long as line and circuit crosstalk are low. However, given the rugged environ- ment of the mining industry, the line conditions cannot be guaranteed. Bad splices and line leakage could be expect- ed with time. Because of these consider- ations, the groups were separated and filters were included at each receiver to guarantee at least 24-dB attenuation to out-of-band signals. THEORY OF OPERATION Operation of the modems requires one LSM modem connected to at least one RCCM modem, a 4,800-Bd serial computer channel (monitoring system) , and a remote trans- ponder. ^ The monitoring system communi- cates two 11-bit words (back to back) to the transponder through the LSM and RCCM, as shown in figure 2. The transponder then enables the carrier of the RCCM and communicates its reply of two 11-bit words to the monitoring system through the RCCM and LSM. There are no limits to the number of RCCM's connected to the LSM as each are seen as a high impedance. The number of transponders connected to each RCCM can also be quite large and is ■^The transponder used in the Bureau's monitoring system is a special device. Upon recognition of a given address and command, it issues a synchronizing pulse followed, 1 1 bit times later, with two data words. It is essential to the oper- ation of the LSM and RCCM in the original development. limited only by the supply current, line length, and definition capabilities of the monitoring system. LSM Terminators 4-wlre cable 2 transmit 2 receive RCCM ^, © ® RCCM ^ <::> I 4-wire cable +V -V Transmit Receive MS LSM RCCM KEY Monitoring system or any serial computer channel Local synchronous modem Remote carrier-controlled modem Transponder FIGURE 2. - Network diagram. BASIC OVERVIEW Figure 3 is a block diagram of the LSM. Communication between the monitoring sys- tem and the LSM is by way of RS232 speci- fied connections. Optical couplers pro- vide electrical isolation of the signals. The transmitted signal is passed to the modulator and the synchronous circuit. Mark and space frequencies from the modu- lator (f ] and £2) are coupled to the line through a transformer and line protection circuit. Precise timing of the synchro- nous circuit opens the channel gate when the transponder's reply is due. The responding mark and space fre- quencies (f3 and f^) from the RCCM are coupled to the demodulator through the transformer and line protection circuit and a bandpass filter. A reference oscillator (at frequency f ^) keeps the demodulator chatter down should a transponder fail to respond in the re- quired timeframe. The demodulated signals signify reception to the synchronous circuit and are then passed through the gate to the monitoring system. Figure 4 is a block diagram of the RCCM. The mark and space frequencies from the LSM (f , and f2) are coupled to the demodulator through a transformer and line protection circuit, a high-impedance buffer, and a bandpass filter. Demod- ulated signals are then passed directly to the transponders. The original RCCM, as given in the section "RCCM Operation," relies on the transponders to issue a synchronizing pulse that triggers the carrier and to delay the data to be transmitted. The updated version, discussed in the section "Other Applic- ations," provides the trigger and delay directly on the RCCM. Although this increases the parts count and complexity of the RCCM, it provides for a more general use. Modulated data from the RCCM (fj and f ^) are passed to the line through a solid-state switch and transformer and line protection circuit. The switch pro- vides greater than 50 dB of isolation in Mine monitoring system CPU TXD COMM RXD Optical coupler * -> Synchronous circuit Optical coupler Modulator Reference oscillator Gate Dennodulator Trans- former and line protection Bandpass filter Line out FIGURE 3. - LSM block diagram. Line in- Line out- Transformer and line protection High- impedance buffer ^- Bandpass filter Demodulator — ►D ^ Carrier control * ^ .^ Switch ^ Modulator ^ ^ Delay ^ * Data in FIGURE 4. - RCCM block diagram. the OFF state and less than 0.3-dB loss in the ON state. Thus, the line is not loaded by the output impedance of the modulator when not in the transmit mode. LSM OPERATION This discussion deals with the original LSM. Refer to the timing diagram of fig- ure 5 and the schematic diagram of figure 6. The updated version under test is discussed in the section "Other Applica- tions." The appendixes can be consulted for specifics on the operation of the modulator, demodulator, and filter. The transmitted signal enters the LSM from the monitoring system at connector Jl pin 3 or from a portable tester at J2 pin 2. Waveform 1 of figure 5 shows the transmitted signal as seen at test point TP2. This signal enters the modulator, IC2, at pin 9. Potentiometers R55 and R57 adjust the frequencies f, and f2. R61 sets the output amplitude to dB as measured at TP3. Diodes D3 and D4 and resistor R9 provide line protection for surges below the threshold of the gas discharge tube on the secondary side of transformer T2. The transmitted signal also enters the synchronous circuit at IC7 pin 6. Pin 5 is high at bit time (beginning of I I I I I I I I l | I I I I I I I I l| I I I I I I I I l | I I I I I I I I l | I I I I I I I M | I M I I I I I I Transmit 1 1 Transmit Receive I Receive 2 I 1 1 1 1 1 1 I ill,,, 20 30 TIME DIAGRAM, 40 bit time 50 60 FIGURE 5. - Waveform timing diagram. I u LU Z) o transmitted message) and can be measured at TP7. Waveform 4 is the signal at TP7 and is known as the channel-open pulse. With coincidence of the transmitted sig- nal, a flip-flop of IC5 is set and pin 2 goes low as shown in waveform 2. An oscillator circuit, running at 2.4576 MHz, is used as a clock input to dividers IC3, IC4, and IC,3. With a low on pin 11, IC3 begins to count clock pul- ses until bit time 16 occurs. At this time, a high at pin 3, waveform 3, sets two flip-flops of ICg at pins 6 and 8. ICg pin 2 goes low and thus opens the channel for reception and blocks the transmitted message from setting IC5 with data pulses. ICg pin 12 also goes low, as shown in waveform 5, and begins the count of the time-out timer, IC13. Shortly after bit time 16, IC7 pin 3 goes high and resets ICg at pin 4. Meanwhile, IC , 3 continues counting for 16 more bit times and then produces a time-out pulse at pin 3. This pulse is shown in wave- form 6 and is the reset to ICg pin 10. At bit time 16, the channel is open (waveform 4) and the RCCM should be re- ceiving the second word of the message shown in waveform 1. At the completion of this message, the reply from the RCCM is expected. The reply from the RCCM can be monitor- ed at TP4. Dl, D2, and RIO provide low- level protection for the received lines. After passing through the buffer and bandpass filter of IC,2j the mark and space frequencies are delivered to the demodulator, IC^. R27 adjusts the center frequency between f3 and f^, D5 indi- cates a carrier level sufficient for reception. ICj4 is another modulator set to oper- ate at the receive space frequency, f/^. R38 adjusts the frequency, and R39 is adjusted to provide about a -50-dB level as measured at TP9. This modulator serves as a reference oscillator to the phase-locked loop of the demodulator. It prevents chatter from appearing at the demodulated data output, pin 7, when no response is received and the channel is open. Normally, pin 6, tied to pin 7, holds the data output off when no carrier is present at the input, pin 2. However, the tight bandwidth of the filter atten- uates out-of-band noise and increases the probability of momentary phase lock on in-band noise and, thus, data chatter. The demodulated data output is measured at TP6 and shown as waveform 7. The pulse at bit time 10.5 is intermittent and occurs about half the time. It is a result of the carrier's being just turned on at the remote RCCM. Although the RCCM turns on at the same frequency as the re- ference oscillator (f4), phase differ- ences cause the momentary pulse. This pulse is the primary reason for the synchronized channel-open pulse. Typical modems, connected one to one, usually have this pulse only during connect time. It is commonly known, and software proto- cols account for its presence. As there was no control over the software for this network, the solution was provided in hardware. The demodulated data at TP6 are coupled to the monitoring system through Jl pin 2, or to the portable tester at J2 pin 1, and are also fed to the synchronous cir- cuit. It is logically OR'd with the time-out pulse and logically AND'd with the Inverse channel-open pulse at ICg pin 1. The resulting pulse sets IC5 flip- flop at pin 8. A low at IC5 pin 12 en- ables IC4 to count for 24 bit times. At the end of that time, IC7 pin 10 goes high and resets the counter and channel- open pulse. Waveform 8 is actually the Inverse of the 24-bit time at pin 12 of IC5, and shows the OR'lng of waveforms 6 and 7 to set the time for waveform 4, the channel-open pulse, to end. The synchro- nous circuit is now reset and awaits an- other transmission. For implementation at the Lucerne No. 8 Mine, the transmitted signals needed to be split into two directions, one for the south mains line and one for the north. To accomplish this, the circuit of figure 7 was connected between the LSM and the RCCM's. The gas discharge tubes were left off the secondaries of transformers Tl and T2 of the LSM. Instead, the transformers were connected to Tl and T2 of the line splitter, and here the gas tubes were placed on the secondaries. The transmitted signal is equally split, TBI-I > TXD TBI-2> Shield TBi-3> TBl-5> Tl 3 G1 I G2 $ G3 I G4 5 Vv^A R2 -W\A- R1 R4 R3 -v\AA R6, , -^AAA- R5 R8 — nAAA- R7 < TB2-I < TB2-2 < TB2-9 North TXD South TXD < TB2-8 < TB2-5 Shield < TB2-6 < TB2-7 < TB2-4 South RXD < TB2-3 North RXD Material list Resistors R1-R8: 10ii,2W Gas discharge tubes G1-G4: Tl I 339 Transformers T1,T2: TRIAD SP-69 Terminal board TB1: RDI3PCV-05 Terminal board TB2: RDI3PCV-09 FIGURE 7. - Line splitter schematic. and the 600-fi output impedance of the LSM modulator is transformed to 150 fi. This impedance more accurately reflects that seen in underground wiring. Since all RCCM's reflect a high impedance, a 150-fi resistor is connected at the end of the transmit lines in both the north and south mains. RCCM OPERATION The following discussion deals with the operation of the RCCM. Refer to the schematic of figure 8 and waveforms of figure 5. The mark and space frequencies from the LSM are received through transformer Tl. Surge protection is provided by the same methods as for the LSM. The received message is delivered to the demodulator, IC,,. through the buffer and bandpass fil- ter of IC5. R24 adjusts the center fre- quency between f j and f2. D6 indicates a received carrier of sufficient level for reception. The demodulated data are passed to the transponder through Jl pin 2 and can be measured at TP3 as that of waveform 1. The reply from the transponder can be measured at TP4 and is the inverse of waveform 7. The pulse at bit time 10.5 triggers IC4 at pin 2. This trigger causes pin 3 to go high and results in DIO going off and IC3 going on. IC3 is a solid-state field effect transistor (FET) switch with four stages in parallel. The total ON resistance is about 10 Q.. At this moment, the carrier is on and t^ is being transmitted back, to the LSM through T2. IC3 will remain on until IC4 times out. This time is controlled by R26 and C16 and is approximately 9 ms. This is enough time, including component toler- ances, to leave the carrier on for com- plete transmission of the transponder's reply. 10 u u a: O 11 The reply from the transponder is also fed to the modulator IC2 at pin 9. R32 and R34 set the f^ and f^ frequencies while R30 sets the output level as mea- sured at TP6. When IC4 times out, DIO goes back on and IC3 goes off. With IC3 off, the output line becomes high imped- anced and the next RCCM that transmits is not loaded by the output impedance of IC2, reflected through T2. OTHER APPLICATIONS As mentioned previously, the circuits presented here are those of the original development, which rely on the synchro- nizing pulse from the transponder for proper operation. These transponders are located in the same enclosure as the RCCM, and the synchronizing pulse has a short distance to travel. If the trans- ponders were to be located at a further point down the line, the pulse (about 20 s long) would be dampened out by cable capacitance. Another application has been encoun- tered that could make use of the modem network. But, in this network, the transponders would be located at dis- tances up to 8,000 ft from the RCCM. Operation of the circuits could not rely on the synchronizing pulse. The RCCM must gate the carrier on at the beginning of the reply and then delay the reply until the line stabilizes. This requires the addition of a clock circuit and data shift registers but provides operation with a typical asynchronous serial chan- nel format. The main concept of the network still remains the same. The LSM receives two data words from the monitoring system (or any serial computer channel) and returns two data words in reply. The words must be back to back at 4,800 Bd, with a posi- tive start bit into the LSM and a nega- tive into the RCCM. A breadboard version of the modified RCCM and LSM has shown successful re- sults. Although the modification is dis- cussed here, full documentation cannot be available until field testing is com- plete. For more information concerning this modification, consult the author. Refer to the RCCM schematic (fig. 8) for a brief discussion of how the RCCM is modified for testing of the breadboard version. The reply from the transponder enters the circuit through diode D9. The leading edge of the reply toggles a flip- flop that gates the carrier on, via IC3 , and enables clocking of a 4-bit shift register and a counter. A clock circuit, identical to that of the LSM, is divided down by the Q9 output of the counter and fed to the clock input of the shift reg- ister. The reply is then clocked through the registers and arrives at IC2 , 3.5 bit times later. A Q15 output of the counter resets the flip-flop and thus resets the counter and the shift register, and turns the carrier off. The modifications to the LSM can be performed more simply. Refer to the LSM schematic (fig. 6) and the timing diagram (fig. 5). TP6 is the demodulated data received from the RCCM. In the modified version, the intermittent pulse at bit time 10.5 will now occur at bit time 21.5 (the moment the transponder begin 's its reply and turns on the carrier) , and the reply will arrive at bit time 25 (3.5-bit delay from the shift register of the RCCM). The LSM can be modified to open the channel at bit time 24, thereby miss- ing the intermittent pulse and enabling the reply. The channel can then be closed 24 bit times later using the ex- isting 24-bit timer. To open the channel, the pulse of wave- form 3 is moved to bit time 24 by AND'ing the Q13 and Q14 outputs of IC3. The re- sultant pulse can then be fed to IC7 pin 2, ICg pin 6, and IC5 pin 8. The output of IC7 pin 11 should be disconnected, and the time-out circuit of IC13 is no longer needed. 12 CONCLUSION The modem network described in this re- port is an inexpensive modulation tech- nique for communication systems operating in a polling mode. As such, it can com- municate to an unlimited number of re- motes at a relatively high baud rate. The circuitry presented can be easily modified to accommodate different baud rates and word lengths. The procedures here and in the indicated references can be used to calculate component values for the modulator and demodulator circuits. The filter circuit is more complicated and may not be necessary for many applic- ations. If independent lines are used for transmit and receive, and if circuit crosstalk and noise are reasonably low, then the demodulator will operate with a reasonably low bit-error rate. With an output power level of dBm, the dynamic range of signal detection is 50 dB. Laboratory tests show that with a signal-to-noise ratio (SNR) of 20 dB , the error performance is on the order of 10"^. Preliminary field testing with five remote modems distributed along the north and south sections of the mine revealed line losses averaging 6 dB/mi and error rates about 10"^. If the 6-dB loss per mile is typical, distances be- yond 5 miles would degrade the perform- ance and result in error rates greater than 10" 5. The effects of impulse noise induced into the communication lines from load switching or other sources could also re- duce the SNR enough to increase error rates. As in any modulation scheme, the circuit is never assumed free from error and some type of information protocol should be considered. 13 APPENDIX A.— FREQUENCY SELECTION AND FILTER DESIGN FSK signals are commonly used to trans- mit digital information over telephone lines. In this type of modulation, the carrier signal is shifted between two discrete frequencies to encode the binary data. These two frequencies are produced by a function generator that is keyed with the binary data. This generator is then referred to as the modulator. A de- modulator, such as those of phase-locked loop (PLL) circuits, is then centered be- tween the two frequencies and reproduces the binary signal as the frequencies shift back and forth. The specifications of the function gen- erators and PLL circuits include guide- lines for selecting the two discrete fre- quencies. These guidelines are based upon the baud rate and carrier frequency, which are determined by the available bandwidth of the communication channel. The typical communication channel for a monitoring system utilizes telephone wir- ing. Twisted-pair conductors of No. 16 or No. 18 gauge are very common sources of mine wiring. In an environment free from electromagnetic inteference, this type of wiring could carry 4,800 Bd com- munications over a 50-kHz carrier for many miles. But documented evidence^ shows that electromagnetic interference in average mines is significant enough to limit this kind of coimnunication to a mile or two. Shielding twisted-pair wiring improves the immunity to electromagnetic interfer- ence but increases the overall cable ca- pacitance. This increase may be more detrimental to a 50-kHz carrier in the way of line loss than would the interfer- ence. The environment then must be con- sidered when selecting a cable, the baud rate, and the carrier frequency for a communication channel. In designing sys- tems for general purposes, severe or worst cases should be assumed or a ^Bredeson, J. G., J. L. Kohler, and H. Singh. Data Security for In-Mine Trans- mission. Final Report — Part I (contract J0308024, Univ. OK). BuMines OFR 76-81, 1981, 99 pp.; NTIS PB 81-221988. variety of communication channels should be designed. The function generator and PLL circuits used in the mine monitoring system modem network are the IC's XR2206 and XR2211, respectively. The specific design using these IC's is covered in appendix B. However, several operational characteris- tics of the PLL must be considered here in first selecting the frequencies of op- eration and filtering requirements. Application note AN-Ol^ gives the fol- lowing guidelines for calculating non- standard frequencies: • The lower frequency, f,, must be at least 55% of the upper frequency, f2 (less than a 2:1 ratio). • Select fi and f.2 higher than the baud rate, fp, for minimum pulse width jitter. • For maximum fj and f2 spacing (where the ratio is close to 2:1), use the rela- tionship (f2 - fi)/fp > 83%. • For narrower f^ and f2 spacing, use the relationship (f2 - f])/fp > 67%. Athough "narrow spacing" is not too well defined, some worked-out examples of the application note for standard modems result in the following frequency ratios and the relationships used: Baud rate Frequency ratio Relationship, % 300 300 1,200 1.10:1 1.18:1 1.82:1 67 67 83 The communication channel at the Lu- cerne No. 8 Mine would make use of two available pairs in a multipair telephone cable. Most of the pairs in the bundle are used for telephone communications to the underground section. For this rea- son, the frequencies selected for modem operation should be above the baseband of voice signals. Since the pairs available for use are direct wires (no telephone loading coils or other circuitry), the ^Exar Integrated Systems, Inc. (Sunny- vale, CA) . Application Data Book. June 1981 , 80 pp. 14 upper limit on the frequencies could be based on propagation losses. In the frequency spectrum (see figure 1)-^ for the available communication chan- nel, f ] and the general shape along the bandpass curve to its left should be above the voice band, and f^ should be as low as possible to minimize line losses and standing waves that might occur as a result of mismatching: preferably below 50 kHz. From the guidelines and worked-out ex- amples given above, the consideration that the carrier frequencies may lie around the 20-kHz range, and the 4,800-Bd rate of the monitoring system, a rela- tionship of 75% would work well. Therefore, (f2 - f ,)/fp = 0.75. (A-1) Then the frequency spacing, df , is nine resistors. The general layout of the circuit is that of figure A-1. This circuit is used with the demodulators for group 1 (fi and £2) and group 2 (f^ and f4). Using the monograph in the applica- tion note, a bandstop-to-bandpass ratio of 2:1 for this filter would yield a min- imum bandstop attenuation of 24 dB. This will be sufficient attenuation for out- of-band noise and for crosstalk inter- ference between the two groups of frequencies. This application note also recommends that the filter bandpass be twice the mark-space separation. Then the band- pass, bp, from equation A-2 is bp = 2 df = 2 * 3,600 Hz = 7,200 Hz, (A-3) and the bandstop, bs , is df = f2 - f 1 = 0.75 fp = 0.75 * 4,800 bs = 2 bp = 2 * 7,200 Hz = 3,600 Hz. (A-2) = 14.4 kHz. (A-4) Application note AN-03^ suggests the use of a three pole-pair Chebyshev filter with a bandpass ripple of 1 dB. The jus- tification is that the filter can be con- structed with just three operational am- plifiers (one IC), six capacitors, and o -"Figure numbers without an A- prefix refer to figures in the main text. '*Work cited in footnote 2. One further consideration before the frequencies can be determined is that of harmonic interference between the two groups. The function generators used in the modulator have a sine wave output with low distortion. When the frequen- cies are selected, the third harmonic of frequencies in the first group must not lie in the spectrum of the second group. More precisely, the third harmonic of the R1A o-wvMHh^ R2A V <7 V V FIGURE A-1. • Chebyshev bandpass filter. 15 lowest frequency in the bandpass of the first group is greater than the highest frequency in the bands top of the second group. Translated into equations, it is 3(fcl2 - bp/2) > fc34 + bs/2, (A-5) where fcl2 = carrier frequency of group 1, and fc34 = carrier frequency of group 2. Substituting equations A-3 and A-4 into A-5 and solving for the carrier of the first group yields fcl2 = fc34/3 + 6,000 Hz. (A-6) The closest the two groups could be is the point where the highest frequency in the bandstop of the first group is equal to the lowest frequency in the bandpass of the second group, or fcl2 + bs/2 = fc34 - bp/2. (A-7) Then, from A-3 and A-4, fc34 = fcl2 + 10.8 kHz. (A-8) The simultaneous solutions to A-6 and A-8 yield fcl2 = 14.4 kHz (A-9) and fc34 = 25.2 kHz. (A-10) Then, from A-2, f , = fcl2 - df/2 = 14.4 kHz - 3,600 Hz/2 = 12.6 kHz, (A-U) f2 = fcl2 + df/2 = 14.4 kHz + 3,600 Hz/2 = 16.2 kHz, (A-12) fj = fc34 - df/2 = 25.2 kHz - 3,600 Hz/2 = 23.4 kHz, (A-13) and f4 = fc34 + df/2 = 25.2 kHz + 3,600 Hz/2 = 27.0 kHz. (A-14) The application note AN-025 contains a worked-out example for the design of the filter shown in figure A-1. An addition- al reference may be necessary for better understanding of filters as used in oper- ational amplifiers. In particular, an- other reference^ was consulted for better definition of the individual component calculations. All of the capacitors of figure A-1 can be set equal, and the re- sistor values become Rlx Qx Ho 2Trfx C R2x = R3x = Qx (2Qx - Ho) 2Trfx C 2Qx 2Trfx C where Qx = the Q (ratio of reactance to resistance) of the individ- ual stages A, B, and C, fx = the center frequency of the stage. Ho = the gain of the stage, and C = the value of capacitance chosen. Either of the references (cited in footnotes 2 and 6) can be consulted for the procedures to determine the Q and center frequencies of the stages. The gain is a free parameter and can be cho- sen for the desired amplitude response. In practice, Rl is much greater than R2; -"Work cited in footnote 2. ^Graeme, J. G. , G. E. Tobey, and L. P. Huelsman. Active Filters. Ch. 8 in Op- erational Amplifiers, Design and Applica- tions. McGraw-Hill, 1971, p. 293. 16 so R2 can be used to trim the Q. Then, to adjust the center frequency, R2 and R3 can be simultaneously adjusted by the same percentage with negligible effect on the Q. Figure A-2 is a set of reproduced pic- tures from a spectrum analyzer, showing the response of the filter used in the LSM. This filter is centered at 25.2 kHz, the carrier of the RCCM, and has an overall Q of 3.5 with unity gain in the bandpass. Amplitude is given in decibels with a reference level of 1 V root mean square (RMS). Panel 1 , of figure A-2 shows a noise source used as an input to test the indi- vidual stages of the filter. Panel 2 shows the output of stage A with the above noise at the input. This stage has a Q of 14 and a gain of 12 dB (Ho = 4) at its peak frequency. Panel 3 shows the output of stage B with noise at the in- put. It is a reflection of stage A, about the geometric center of the filter, with the same characteristics. Panel 4 shows the output of B with noise at the input of A and the output of A into B. Panel 5 shows the output of stage C with noise at the input. This stage has a Q of 7 and a gain of 12 dB at 24.94 kHz, the geometric center of the 3-dB bandpass points. Panel 6 shows the final output of stage C with noise at the input of A, and Panel 7 shows a comparison of input to output. Figure A-3 is a reproduced picture of the spectrum response from an RCCM. The -40 -10 -20 -30 -40 -30 -40 -50 -60 . .y^ ^'''''''''^. . - 28.75 25.25 -10 +2 +4 +6 -t-8 +10 -6 -4 -2 +2 +4 +6 +8 +10 -|0 -i FREQUENCY, kHz FIGURE A-2. - Filter response. ], Noise source input; ~, stage A; 3, stage B; J, stages A combined; .5, stage C; 6, stages A, B, and C combined; 7, input-to-output comparison. (Number each panel is the center frequency of the panel, in kilohertz.) and B below 17 top waveform was measured at TP4 of the LSM (fig. 6) and shows the unfiltered re- sponse. The two peaks correspond to mark and space frequencies transmitted from the RCCM. The bottom trace was measured at TP5 and shows the filtered response. -70 TP4 raw 15 17 19 21 23 25 27 FREQUENCY, kHz FIGURE A-3. - RCCM filtered response. 29 33 35 18 APPENDIX B. — MODULATION DESIGN AND CALIBRATION The modulator and demodulator used for the LSM and RCCM IG's are the XR2206 and XR2211, respectively, from Exar Integrat- ed Systems, Inc. The application notes^ on these circuits cover the theory of op- eration. The justifications for the equations as well as worked-out examples are included in those notes. The design solutions for the RCCM and LSM demodulators are presented first. Refer to figure B-1. Choose the value of the timing resistor (RO) between 10 and 100 kQ or -50 kf2. Then CO = 1/(R0 * fo) = 0.00139 pF; let CO = 0.001 uF, 5% for convenience, RO = 1/(C0 * fo) = 69.4 kQ = 64.9 k$7, 1% + 10 kO, potentiometer. RCCM DEMODULATOR Given that f, = 12,600 Hz and £2 = 16,200 Hz, calculate PLL center frequency as fo = (f , + f2)/2 = 14,400 Hz. Exar Integrated Systems, Inc. (Sunny- vale, CA). Function Generator Data Book. May 1981, 53 pp.; Phase-Locked Loop Data Book. Feb. 1981, 85 pp. Calculate Rl to give a lock range equal to the mark-space deviation as Rl = (RO * fo)/(f2 - fl) = 277.7 kfl -270 kfl. Calculate CI to set loop damping equal to 0.5 as FSK input iCD VCO fine tune FIGURE B-1. - Demodular circuit. 19 CI = CO/4 = (0.001 uF)/4 = 250 pF -270 pF. Calculate CF for data filer time con- stant as CF(uF) = 3/baud rate = 3/4,800 = 625 pF -620 pF. The capture range, dfc, is recommended to be 80% to 95% of the lock range. Picking 90% yields dfc = 3,240 Hz. For the full-tracking bandwidth, the lock- detect filter capacitance is given as 16/2 dfc (in microfarads). CD (yF) = 16/(2 * 3,240) = 0.0025 pF. LSM DEMODULATOR Given that fj = 23,400 Hz and f^ = 27,000 Hz, calculate PLL center frequency as fo = (f3 + f4)/2 = 25,200 Hz. The calculated values follow, as for the RCCM demodulator: CO = 1/(R0 * fg) = 0.00079 yF. Let CO = 0.001 uF, 5% for convenience. Then RO = 1/(C0 * fg) = 39.7 kJ2 = 34.8 k^, 1% + 10 ka potentiometer, Rl = (RO * fo)/(f2 - fl) = 277.7 kD. -270 kn, CI = CO/4 = (0.001 yF)/4 = 250 pF -270 pF, CF = 3/baud rate = 3/4,800 = 625 pF -620 pF, CD = 16/(2 * 3,240) = 0.0025 yF. Next, the design solutions for the RCCM and LSM modulators are presented. Refer to figure B-2. 0>2V p-f| ^y . Q > > J1 -2 J1-4 ^si, 1 ii J1-1 Ji -3 . ^ 25 ^ B, '^^ cw C6 I/.F S2 t — v^AA- RI3 ii 4.7 kil B' I 15-V power supply i- C7 lO/xF ac power FIGURE B-3. - RCCM calibrator schematic. 22 8. Adjust level for dB on voltmeter. 9. Adjust £4 for 27.00 kHz. 10. Move voltmeter-counter probe to TP6. There should be no output (less than -50 dB). 11. Switch SI to fj position. The gate indicator should go out. 12. Readjust level for dB on volt- meter (you should have to increase it by about 5 dB). 13. Adjust fj for 23.40 kHz. 14. Calibration is complete. Remove test equipment and seal RCCM adjustments. <^^ FIGURE B-4, - LSM printed circuit layout. 23 LSM CALIBRATION The modulator transmit frequencies of the LSM are adjusted while controlling the digital input at J2 pin 2 (see figure 6). 2 The output level, as measured at TP3, is adjusted for dB with a 600-i^ resistor connected across TBl-4 and TBl- 5 The reference oscillator is adjusted while monitoring at TP9 and the level set to -50 dB. . . To calibrate the demodulator, it is best to have a freshly calibrated RCCM. Modulate the RCCM, at TP4, with a 1-kHz square wave and connect its output to the input of the LSM. Adjust the demodulator for symmetry as measured at TP6. Figure B-4 is the printed circuit lay- out of the LSM shown in figure 6. Figure B-5 is the layout for the LSM line split- ter of figure 7. And lastly, figure B-6 is the layout of the RCCM of figure 8. 2Figure numbers without a B- prefix re- fer to figures in the main text. FIGURE B-5. - Line splitterprintedcircuit layout. FIGURE B-6. - RCCM printed circuit layout. INT.-BU.OF MINES, PGH., PA. 27962 ..„ -nn- !-»,- n-nnn n/iT 'V 1 40 1 D DD > ^ Q_ O 9- 3 30;^ D O O i -* -.5-3-0 (t n 3 ™ -< -n -n J] > ■n r- 3J m < > c en 03 c ■f' 3DgcD O o c -oO> 2S = < m z > z m z cc O m > H Tl (/) -^ m H m O 03 > z m O c > o ■0 ■o O JO -\ c z •^R > <« 2 > t -1 Z en '1 -n H m X rn m « 5i m . w o ^ 'Cp vv '^0^ .^^ pr.c^%-"- \^ ». "^ o 4^ ' ^°-^^. ^^Z*^-^.^^' *■ J" CS '\/^ ft ^*. -^^0^ c" . -> -^^^^^^ ^' ^°1 o „ . « ^/xj v^^ • .<-' <> V V' - ■ * 'o . » » A x/ ^ *.,.♦ ^^ t^ A^ » ^ •» o * o o " " » '^ <> "'T.%* .0 ^-.^9^' ^V^. ••*•»'% •.^^•- /\ •: ■vO- .CO, ^^ 0^ %> ^.,,.- «,V I • » « rt 'I "^ ^ V S • • B . .« A ° A^^ - \ 1" ^ "SJ^W^" "5!. « ' ^U -^ *.,!•■ ^ V^ .»'^L,'* «^ r:^' ,0 ,* «? V. 'yj%^,^ ^v -^^ /.'^:^^'^°o *^* o o " » « <. *'Trs^\.0^ r "^^. %^^^^/ >^ * aV -S^ • o V i-J^" . :. -^--o^ .„ o, 'o . . * A /^\/.. V"^-/ \/^\/ v^V V^\/.. v^-/ •^ A^ » ^9 ' SsJ>\S^Ik<2 I '<^^ ♦ aV "Su <■» c » " » * .5.^"-^ ^0 ^ "Vu ".101?^ ^°-^^. TT^TPH^ '^_ V'^ * o » o ■ %/«S&\V^^"' /..i^>>o .,/y^;^A. ./..i^vA ..^"\c:^%\ ^o^ K . '^bv^ ^0-^^^ . " o , <'. ^oV^ 3r^ V.^"^ •\/\ '.^•' /\ itt- ^*^\ «^ "•-•' aO' .. \"*!^*V^^ V'^^^V ^.'^^'V^"^ %'^'^^V^ '^\ -o;-o' ^°-^. '■^o' .V %.^^ ,' 4'°^ '^^ *'"•'' I- \/ *' ^.- /Rials'- ^^% '>0^ v%.^^'»&&\X/ /ife'vX^^\'^& '^- ^- "^ >°-^^ .V \^ o_ •• jP-n,. -n^o^ o^ % ' W^»\/^^^ V^ «^ X/- ••». ■ %..^ ^v -o^ ^.-^.T*- ^ >^/'»%?'' ^0^' '^^.'^l^l^V.'V "■' "V '%5-.^;^ .>" ''-^. .'J^" ..t'., "^^ ^* ^ ^' .4q 0'" .cOjL"'', "^o ^^-n^. •^» ^ V^O^ *bv^ ^°-^^. °", '*' ./ V '.S^^° ^^'"'^-. °^^^" ^'" .^^°- 'br ^"'^*-. c- •:/ : ...... v-^;< . . .. v^'-y. . . . , v-^-;. %.