>Via UO-14 >From : WB9MJN >To : ALL >Title : 430 Mhz, 56 KB FDX Station. >Keywords : CELLNET, FDX, 56KB, TCP/IP >Uploader : WB9MJN >Uploaded : Fri Aug 02 19:18:13 1991 __________________________ 430 Mhz Band Full Duplex 56 KBaud CELLNET Prototype Notes Donald V. Lemke, WB9MJN 25 July, 1991. These notes cover the technical details of our two operational 56 KB 430 band full duplex CELLNET prototype stations, using the WA4DSY 56 KBaud RF modem. These stations are operating on a 10 mile test link, with 10 watts, on 425 Mhz and 433 Mhz. Full details of the CELLNET concept are available in the "7 th ARRL Networking Conference Notes", in the article "Celluar Area Coverage Transport Networks" authored by myself. The RF prototypes have been operated at the ILNAP:K9VXW-1 and the N4PCR-1 (Gracilus) sites for nearly 2 years, and have been in active network use since the installation of a Gracilus PacketTEN at the ILNAP site for nearly a year. For the first year of operation, the ILNAP site station was in regenerative mode, simply relaying the transmitted data from the Gracilus site, back to that site, for testing purposes. This test link has been very reliable. During one memorable trip to the ILNAP site, to work on equipment, all links to the site including the telephone were down, except for the 56 KB full duplex link due to iceing conditions, and t/r switching wair and tear. Utilisation of a single 430 Mhz antenna, and feedline, results in a very economical 3 link radio system, as a result of operating in full duplex. Duplexor Cavities are usually required when operating a station from a high performance RF site, for protection from IMD and nearby lightning strikes. Consequently, split band technigues, such as 220 and 440, are more expensive, since these technigues still require Cavities to be reliable, as well as the added cost of a second feedline and antenna. Multi-banded systems are also more dificult to obtain permission to instal on high performance RF sites. Besides the extra tower loading of the additional antenna(s), the determin- ation of the interference potentional with the sites primary services is much more complicated. Diplexing 220 and 440 on a single feedline with a compact antenna mounted filter network is tuff to do. I don't believe a 220/440 Mhz nor a 145/220 Mhz diplexers are available commercially. 145/430 diplexors are very common, however. Thus, with diplexing filters, a 2 meter network access station can also easily be accomadated with the same feed-line as the CELLNET transport radio system. If the Japanese, or European versions of the popular dual band antennas were availbale in the U.S., cost would be even less, as a diplexor filter on the antenna end of the feedline would not be neccassary. Allas, the 2 meter / 440 antennas that are available here, do not work very well below 433 Mhz, and pleas to both COMET and DIAMOND antenna companies to supply their home market versions of their popular antennas in the U.S. have either not been understood or rebuffed. Configuration of 430 band, 56KB CELNET Prototype Stations: _____ \ ! / Antenna \!/ ! ! ------------- ! ------------- ! /\ ! ! ! /\ ! --! \ !-!-! / !------------ ! ! \/ ! ! \/ ! ! RG-142 ! ------------- ------------- ! RG-142 ! BPBR Duplexor ! ! (Phelps Dodge 4 Can, >10 watts) ! 10 Watts out ------------- (Motorola 2 Can, 10 watts) /\ Pauldon PD440 Filter A ! /\ ! ft / \ (uses S-AU4) ! / \ ! +12v(A)---/ \ ! / \ ! /______\ ------------- ! .25 Watt in - - - - - -!- - - - - - - - - ! : ! 425 (433) Mhz : ! 433 (425) Mhz : ------------- : ------------- : ! Hamtronics! :ft ! /\ ! Filter B : ! CA432-2 !-------------- +12v(B) ! / \ ! : ! RX Conv. ! : ! / \ ! : ------------- : ------------- : ! 21 (29) Mhz : - - - - -!- 1 Watt in- - - - : ! : : ! 433 (425) Mhz : : ------------- :ft : ------------- : : ! WA4DSY !-------------- +5v ft: ! Hamtronics! : : ! RX'er ! : +12v(A)-----! XV4 ! : : ! ! : : ! TX Conv. ! : : ------------- : : ------------- : : ! RXA : : ! 29 (21) MHz : - - - - - -!- - - - - - - - - : ! ~1 miliWatt : - - - ------------- - - - -ft ft: ------------- : : !WA4DSY ! :---- +5v +5v-----! WA4DSY ! : : !Demodulator! :ft ft: ! TX'er ! : : !Decoder ! :---- -5v -5v-----! ! : : ------------- : : ------------- : : ! RXD,RXC,DCD - - - - - - - - - - - - -!- - - - - - - - - - - - - - - ! - - - - - - - - - - - - - - : ! TXAI,TXAQ,TXEN -------------------------------- : - - - -------------- - - - ! ! TXD,RTS ! WA4DSY ! : ! PacketTEN 5port Stand-Alone !-----------! modulator ! : ! "NosInABox" uses 68302 chip ! : ,TXC ! encoder ! : ! ! : -------------- : -------------------------------- : : ! ! ! ! - - - - - - - - - - - - - - - 4 Other ports to other radios, and ft - Pi Section DC Feed Thru filter local console. Murata-Erie # NFT403-806D1H403 --------------- --- +12v(A) -------------- ! ASTRON ! / ! Switching !------ +12v(B) ! RS-7a !-------------------! Power !------ +5v ! ! ! Supply !------ -5v --------------- -------------- Enclosures - The Hamtronics CA432-2 and WA4DSY Reciever(s) are enclosed in a 9 by 4 by 4 inch Die Cast Aluminum Box, Hammond # The CA432-2 was mounted on the bottom of the box, along with one WA4DSY reciever. There is room for a second reciever to be mounted on the removeable top of the box over the first WA4DSY reciever. The Hamtronics XV4 and WA4DSY Transmitter are also enclosed in a 9 by 4 by 4 inch Die Cast Aluminum Box, Hammond # seperate from the Reciever Enclosure. XV4 was mounted on the bottom of the box, with the WA4DSY reciever mounted to the removeable top. The WA4DSY Demodulator/Decoder and Modulator/Encoder are encolsed in a 9 by 7 by 2 inch aluminum chassis, Bud # AC406 with # BPA-1593 bottom plate. A second such chasis, was in- cluded to allow for 2 additional WA4DSY Demodulator/Decoder Circuit Boards. The chasis were butted together, and a hole drilled to allow for wiring to pass between them. Shielded DB9 connectors were mounted on the back of the first chasis for the signals to and from the WA4DSY transmitter and re- ciever. Switching Power Supply - Radio Shack # 277-1016 (AKA Kogyo # 10053214-2), modified for 12 Volt D.C. input, per RMPRA notes. I.E. change R7 (a 240 Ohm, 1 watt) to 120 Ohm, 1 watt, and R19 (a 910 Ohm) to 470 Ohm. Remove the input Full Wave Bridge rectifier, short the "+" terminal to the terminal that leads to L2 and the "-" terminal to the terminal that leads to L1. This sup- ply provides plus and minus 5 volts for the WA4DSY modem, and a filtered +12 volts for the Hamtronics CA432-2, with 12 volts D.C. input. Filter A - 6 Helical Resonator Band Pass Filter. Approximately 4 dB loss. We used the MIXER assembly from a Motorola MOTRAC- Mobile. Similar filters can be found in MOCOM70-Mobiles, and MICOR-Mobiles. The MOTRAC-Mobile has a tube transmitter, making it cheaper to junk for parts. The basic casting is what is of value. We stripped out the mixer diode C.B., and soldered on a BNC connector and lead tapped at the same point as the original input was. The original input was used as the output to the Hamtronics Recieve converter, and the BNC was used to connect to the Duplexor, with double shielded 1/4 inch size coax. Filter B - 6 Helical Resonator Band Pass Filter. This filter is from the Motorola MICOR-Mobile. It is smaller than Filter A, and has more loss, 6 dB, and a tighter Bandpass response. Its is needed to reduce the linear modulator, and mixer noise floor, which is only about -70 dB out of the XV4. Measurements also indicate that the great majority of the noise is from the WA4DSY Modulator C.B.. THIS FILTER IS AN ABSOLUTE MUST, for full duplex to work. To get it down to 433 and 425 MHz, add about 1 turn to each helical resonator coil of a 450-470 MHz version filter, or buy the 400 to 420 version filter coils. Sorry, don't have the Motorola Part numbers for the coils. Anybody out there know these? Here is a talley of the RF levels to convince you to USE a filter in the transmitter. TX power 40 dBm Linear Modulator Noise Floor -70 dB Duplexor Isolation (2 Can) -50 dB --------------------------------------------- Noise Power ON-RX-CHANNEL W/O Filter -80 dBm <-Desense! Filter loss at 8 Mhz Split -55 dB --------------------------------------------- Noise Power ON-RX-CHANNEL W/ Filter -135 dBm <-Negligable Desense. Why is the WA4DSY Modem Linear Modulation anyway? The WA4DSY modem uses a linear I and Q modulator to achieve true coherently modulated MSK. This modulation is narrower in modulation than non-coherent FSK, with about the same BER rate for Eb/N0 . Also, it meets the FCC Bandwidth rules for data emisions on the 220 and 440 MHz Bands, while doing the legal signalling rate limit of 56 KBaud. This is probably the pri- mary reason for linear I and Q modulation, to meet the bandwidth limitations and still be able to do the legal signalling rate. Thus, we have only a -70 dB modulator noise floor and the the requirement for the TX'er filter when doing Full Duplex with this modulation. More On Filters: The TX'er filter isn't a big thing, its easy to modify and tune up, alto it is probably something that those familiar with full duplex radio construction for FM voice are not expecting. Filters such as these are an everyday affair for RF engineers. In recent years exciting improvements have been made in these kind of filters. They made possible the Cellular Telephone revolution. Each Cellular phone has a small ceramic duplexor, which is the equivalent of the Motorola 2 Can BPBR duplexor, in performance. At Dayton in 1991, the Motorola booth had demonstrations of similar 400 Mhz filters that were approximately 1 by 3 by .7 inches! Two of these could do the 433/425 duplexor , and be smaller than the 10 watt RF power amp! One more in the TX strip, and another as the RX preselector filter, and your done. Looking forward, its not hard to see that FULL DUPLEX book sized 56 KB USER ACCESS stations are a possibility. When Microwave links replace the CELLNET links, way down the road, the 433/425 CELLNET hardware could easily be converted to support users of book sized data stations. How fast do u want your 100 K file, 2 seconds good enuf?!?! Its only RF. Modification of the WA4DSY RF modem to 21 Mhz band. The CELLNET prototype stations use Hamtronics recieve and transmit converters to translate the 29 and 21 Mhz WA4DSY modem outputs to 433 and 425 Mhz respectively. Thus, no modifications, or even recrystalling of the Hamtronics transverters was neccassary. Since the WA4DSY RF circuits are somewhat less involved than the Hamtronics Transverter, it was easier to modify the WA4DSY RF circuitry. Since crystals for the WA4DSY modem are not supplied with the kit, and had to be ordered anyway, we ordered the neccassary 21 Mhz crystals straight away. Note: in the CELLNET application, only A transmitter, or THE recievers at a single site are modified, but not BOTH the transmitters, and the recievers. Modification of the WA4DSY Data Radio for 21 Mhz Operation: Reciever Modifications: Change Componants as follows: C1 - 300 pF C2 - 47 pF C4 - 10 pF C5 - 51 pF L1 - 1.33 uH (TOKO # BTKANS-9449HM, Digikey stock # TK1412) Transmitter Modifications: Change Componants as follows: C28 - 100 pF C29 - 200 pF (Parallell 150 with 50) C35,39 - 43 pF C48 - 130 pF L13,L14 - .29 uH (TOKO # BTKXNS-T1047Z , Digikey stock # TK1406) C41 - 30 pF C42 - 300 pF L10 - 1.33 uH (TOKO # BTKANS-9449HM , Digikey stock # TK1412) Retune per WA4DSY instructions. Its easier if both a Transmiter and a Reciever are modified at the same time. Observed Performance: The two test radios have consistantly resulted in good BER performance. The PacketTEN NOS has a monitoring facilities which indicates the number of frames recieved, and the number of those frames in error. With this fa- cility we have monitored the the 56 KB Full Duplex links psuedo BER. As the BER's of the link are so good, always observed to be in better than 5e-5, its a safe assumption that the psuedo BER is very close to the actual BER. On average the link puts in a 1e-5 BER. As stated above its never been observed to have worse than 5e-5 BER and at times its as good as 1e-6. The 10 mile path is a slightly sub-grazing path, with a 100 foot high 6 dBd antenna at one end, and a 50 foot high 10 dBd antenna at the other. Significant foil- iage blokage is present on the 50 foot high end. The path is not flat terrain, but includes a river valley, and ridge line at least 50 foot higher than ground level at the 50 foot high antenna end. Overall, this is a good path to test over, with many real world path variations. The Gracilus end of the link is close to downtown Aurora,IL, which seems to be a very intense RF envior- ment. The police station is nearby, with a communications tower, and many antennas. There are several FM broadcast stations in downtown Aurora, and apparently allot of paging transmitters. 2 meter hts can not be operated, even in the basement, and on a rubber ducky antenna without the familiar paging IMD breaking squelch. Significant reduction in BER was observed when the above mentioned ICE storm caused the 10 dBd antenna at the Gracilus site to sag at a 30 degree down-angle. Previous to that time, the link was regurlarily doing 1e-6, and afterwards only occaisionally. The reported average perfor- mance, tho, includes the performance with the damaged antenna, since to date it has not been repaired. The reported performance is in line with prediction equations in the "7th Computer Networking Conference Notes". First let me mention that the equation on page 125 for Reciever Noise Power begins 198.6, when in actuality it should have begun -198.6. The tables calculated used the correct equation. 4 dB Noise Figure was not atainable with this hardware prototype, due to additional reciever preselector loss, and larger CA432-2 Noise Figure than assumed for the estimation of 4 dB. The actual Noise Figure of the prototype recievers was approximately 6 to 8 dB, depending on individual preselector variation, and the larger CA432-2 Noise Figure. Additionally, the test path used has an unknown path loss, due to its variability. The combined impact of these effects, easily accounts for the observed BER's. I approximate a 30 mile truely grazing path link using this same hardware would have a 7.5 dB worse performance, which would be about 4 e-5 average BER, or about a 14 dB margin over 10e-3. In otherwords, the link would successfully send 1000 byte packets between 90 and 95 % of the time.