CONSTRUCTING THE DK2DB TERMINAL NODE CONTROLLER =============================================== By Ken Wood G3WCS ================= This article was contained in issue 4 of the North West Packet User Group newsletter. It is included here so that those non members may have access to the text and may be free to copy it for their own use, and that non-members may see the quality of articles and information provided by NWPUG. The following describes the construction of a small VHF/HF TNC from a printed circuit board designed by Eisch Electronics. Construction is relatively straightforward but should be undertaken by persons with some previous experience in building. The question that most people ask, is how the final cost compares with that of a ready made unit such as a Tiny 2? The answer to the question depends largely on what access the constructor has to relatively modern components at low or even no cost. I would estimate that to buy most of the major components new, would result in a total cost of about 70. In my case the cost was about 45, as I was able to obtain several components at trade price and had several others in the junk box. There are a few high cost items used in the TNC, these are:- a) AM7910 or 7911 Modem Chip, b) 43256 or 55257 RAM chip, c) MAX232 RS232 interface chip, d) Z80CPU and Z80SIO chips and e) extruded aluminium case to accept the standard Eurocard size PCB. The aluminium case is available from Maplin or Cirkit. CIRCUIT DESCRIPTION ==================== The audio input and output connections, together with the PTT line are taken to a 9 pin D socket at the rear of the PCB. The audio lines are taken, through level adjustment potentiometers, to the AM7910 Modem chip. As in the TNC220 this chip serves to interface the audio signals to the remainder of the TNC. In addition the transmit tonesare generated here with no need for any setting up proceedures as in the TNC200. There are links on the PCB beside the AM7910 which permit the selection of 1200/2200 Hertz tones and 1200 baud for VHF or 1600/1800 Hertz and 300 baud for HF use. It is worthwhile noting that the audio input circuit is very simple and I have incorporated a small filter/clipping circuit in my TNC to improve performance slightly. The computer interface is handled by a MAX232 interface chip. This allows two RS232 channels and is used to control the TX and RX data between the TNC and the computer. There are further PCB links which allow TNC to computer baud rates in all standard steps from 300 to 9600 baud. The computer is connected to a standard 25 pin D socket at the rear next to the 9 pin audio socket. There is no connection to pin 8 of the 25 way D socket, on most TNC's pin 8 serves as the data carrier detect pin. This is of no consequence to the majority of operators but those wishing to run a mailbox, should note that the WA7MBL, and later versions of W0RLI mailbox software require this pin to be connected to the computer, to verify that a connect has actually taken place. Having said that, it is simple enough using one extra chip to provide such a TTL signal from the "connect detect" area of the TNC circuitry. The remainder of the TNC is fairly standard using a Z80A CPU with a Z80ASIO-0 SIO. It is important to use the -0 SIO due to the fact that SYNC signals are required on both channels A and B. The clock may be run with either a 2.45 or 4.91MHz crystal, and there is a jumper on the PCB to select either. One criticism of the TNC220 is that with a clock speed of 2.45MHz, it is too slow and tends to get "confused" when called to do a lot of processing quickly. I found this to be the case with this TNC also, and fitted a 4.91MHz crystal. This works extremely well and the Z80A CPU and SIO seem to have no problem in coping with this increased clock speed. FIRMWARE ======== The TNC uses the same firmware as the TNC200 and I am currently running version 1.1.5. The CALSET command is not recognised because there is no requirement to set up the tones as with the TNC200. All the other commands function correctly and in the manner described in the TNC200 or TNC220 handbooks. In particular, with a 4Mhz crystal the inbuilt clock is accurate to thirty seconds or so per week, which is more than you can say for my KAM!! The firmware resides on a 27C256 Eprom and there should be a number of people about who can blow such a device. The default parameters are loaded into RAM on power-up. This RAM is battery backed up to preserve the settings when the main power is removed. Like the TNC220 certain commands, such as AWLEN or 8BITCONV require the RESTART command to be issued before they take effect. CONSTRUCTING THE DK2DB TERMINAL NODE CONTROLLER =============================================== By Ken Wood G3WCS ================= CONSTRUCTION & QUIRKS ===================== Construction of the board is very simple to anyone with some building experience. I would not recommend it as a first or early project. The board is double sided with plated through holes. It is designed for small components such as quarter watt resistors and miniature capacitors. When constructed it will slide into an extruded aluminium case which is obtainable from Maplin. The layout is then identical to a conventional TNC, with computer, radio and power connections on the rear panel and five LED's, (Power, TX, CON, DCD, STA), in a horizontal row on the front. The appearence of the unit is more "chunky" than a TNC200 or 220. (Note from sysop GB7SAM - Cirkit do a slimline extruded aluminium case in plain metal or black anodizing. With a very small amount of filing of the 25 pin D connector support, and a measure of care, a slimline TNC can be constructed. Two have been used satisfactorily on this BBS for over a year.) There are a couple of small points to bear in mind. The design permits the use of LS or HC or HCT type I.C.'s. Due to the slight differences, mainly in timing between the different series of chips, you should ensure that they are all of the same type. When I constructed my unit it did not function correctly. I eventually tracked down the problem to a single HC series chip. (All the others were LS). When it was changed for an LS type everything was fine. When using software flow control between the TNC and computer, you must connect together the CTS and RTS pins in the 25 way D plug. This appears to be a quirk with the MAX232 device, other TNC designs do not seem to require this. The TNC does not suffer from the "memory dump" syndrome which seems to affect the TNC220 and I believe the Tiny 2. For those who have not seen this effect, it consists of the entire contents of the memory being sent to the VDU at once. What you are seeing is the last hour or so of activity, being played back from memory. OBTAINING A PCB =============== The PCB is obtainable from Ewald Gobel DK2DB, Wingertgasse 20, 7500 Karlsruhe 41, West Germany. He trusts in the British amateur's honesty and to date, as far as I am aware, no one has let him down. If you write to him he will send a PCB by return. He asks you to pay by sending DM40 to him in German bank notes. This is about fourteen pounds at today's exchange rates. If you do order a PCB from him, please pay promptly or you will spoil it for the rest of us. (Note from sysop GB7SAM. Alternatively, MAXPAK, the Midlands AX25 Packet Users Group stock the board and provide a decent set of documentation, in English. They have lots of constructional experience with this design, and are able to offer a good standard of technical support. Contact details may be had by sending an enquiry to the sysop of the Maxpak BBS, GB7MAX.) CONCLUSIONS =========== This PCB forms the basis of a small, reliable and well designed TNC. It works very well and sometimes "hears" packets which my KAM, which is connected in parallel, does not decode. I had some slight reservations about the lack of audio input filtering but these have been proved to be groundless. My TNC has been switched on for four months now, subject only to slight interruptions, governed by MANWEB. In this time it has only crashed once, following a night of very heavy static rain and frequent power interruptions. It runs my mailbox G3WCS-2 and interfaces with my Atari 520STFM computer perfectly. I have also tried it with a Sinclair Spectrum, BBC Model B and a P.C. Clone, all of which work fine. My only slight gripe is regarding the supplied documentation. The fact that it was written in German was quite expected, indeed it is good that Ewald is prepared to supply amateurs outside Germany. In any event many of the technical terms are recognisable. There are only three sheets supplied, a technical description, a parts list and a board layout. The board layout is drawn freehand and care is needed to interpret the positioning of components correctly. The actual PCB is NOT silk screened. Having said the above, it is a nice project, works well and has been built by quite a number of us now, so that if you run into trouble there should be someone around to help. Good luck with it if you decide to build one. Ken, G3WCS @ GB7CHS PORTED AND REALIGNED FROM PC TO AMIGA FORMAT BY BOB G0LBQ - N6XDI.