karn (01/11/83)
By Lyle Johnson WA7GXD, and Dan Morrison KV7B BLACK THURSDAY All was going well. The parts were in hand, PC boards delivered and volunteer (slave) labor lined up for the BIG PUSH -- would you believe Beta TNCs in time for Christmas? Neither would Murphy! On Wednesday, 8 December 1982, the first TNCs were delivered to TAPR from one of the two assembly plants. The first one worked per- fectly! Unfortunately, by 2:00 AM, Thursday, 9 December, a serious problem was very apparent. "Black Thursday" was upon us. Only 3 of the first 19 TNCs were functional. After a fitful (morning's) rest, an emergency meeting of the TAPR executives was called, production halted, and an investigation into Murphy's doings launched. It turned out that there was a problem in the plate-through job on the PC boards, meaning that the top half and the bottom half of the board were inter- mittently (dis)connected. The 119 TNCs assembled by the time of the halt in production were sent to the scrap bin, and an assessment of the damages made. The losses were a staggering $8600! TAPR, being in no position to absorb such a blow, determined to finish the project anyway. A salvage operation was mounted, and in the end there is a $3000 loss to contend with -- not insubstantial, but one that, with a judicious amount of finagling, and some voluntary $upport from those interested in seeing this thing through, can be overcome. New boards were ordered fabricated (free) and parts to populate the boards re-ordered. After a few slips in the schedule, 10 boards were received the week of 28 December 1982. Two were hand-assembled by WA7GXD, and worked the first time. Three were assembled by one of the two assembly plants, wave-soldered and delivered. All three worked! The delay time was used by both the software group, for additional debugging and refinement, and the hardware group, for additional testing and interfacing. Meanwhile, the approximately 70-page TNC Manual is being edited and clarified. The present schedule calls for board assembly to begin the week of 9 January with assemblies delivered to TAPR late in the week of 16 January. The boards will then be powered up and power supply voltages checked. Passing this test, the ICs will be loaded and calibration performed on the MODEM. Finally, the TNC will "connect" to itself via a digipeater, then be packed up, along with the manual, for shipment to the Beta coordinator. The Beta coordinator will then be responsible for distribution of the TNCs to the indi- vidual site participants. It has often been asked if the TAPR TNCs will be compatible with the Vancouver TNCs, and the answer is an emphatic YES! Harold Price, NK6K and co-author of the high-level PASCAL software, established a con- nection with Skip Hanson, WB6YMH, in Los Angeles on 8 January 1983. Harold was using a Beta TNC, while Skip used a VADCG TNC. Packets were exchanged for a half-hour or so, and the test was a complete success! The initial software released with the Beta boards is a version of the AMSAT protocol agreed upon in Washington DC the weekend of October 8th, in addition to the more-or-less standard Vancouver protocol in general use. The software is outlined in more detail below. SOFTWARE UPDATE This update would be written by Margaret, KV7D, rather than by me (KV7B), except that as these words are being composed she is putting the last bells and whistles in the TAPR/AMSAT AX.25 level 2 protocol software which will appear on the TAPR Beta board at time of distribution. The people responsible for this monumental effort are: Dave Henderson, KD4NL, who was responsible for implementing the protocol; Margaret Morrison, KV7D, who put together the entire package of real-time assembly language routines and debugger; and Harold Price, NK6K, who built the extremely flexible command parsing language. Of the 24k bytes of PROM on board, 22k are used in this release. The high-level routines were written in compiled Pascal, and occupy about 16k of the total. The entire development time was roughly from the AX.25 protocol meeting in October until today (mid-January). The simplest description of the software is that it's everything any of us dared to hope for and more. Not only does it implement AX.25 level 2 protocol "by the book," (it follows the October 10 document produced at the AMSAT meeting, based on the AMRAD recommenda- tions; and where ambiguities exist, follows Ma Bell's BX.25, issue 2 document), but it can also operate in Vancouver compatibility mode with no problems. Present at this time are some 66 user settable parameters for controlling the behavior of the board, many of them for determining the details of terminal communication. In this short space only a few high- lights may be mentioned, but they include such items as hardware and software flow control, separately con- trollable in either direction, with the software flow control characters being user definable. In order to permit total transparency a single command defeats all special character traps and packetizes according to timer and packet length criteria. At the other extreme, input editing at the character and line level are available, including a separate "cancel packet" command. Perhaps the most important features are the ela- borate timer operated CSMA procedures, including (all user settable) separate key-up delays for digipeating and originating packets, separate delays for voice repeater key-up and tail-drop times, and random wait times after collisions. Flow control on the RF link also involves time delays. Altogether, there are ten distinct timer functions associated with the RF link, requiring 4 separate clocks to be maintained by the interrupt driven routines. All in all, the functions presently on board should be more than adequate for the "shakedown cruise" coming up during Beta Test, and future releases will have even more features (we've still got 2k PROM to go before adding the next deck to the board!).