karn@allegra.UUCP (Phil Karn) (04/08/84)
All, !!! IMPORTANT !!! TAPR has recently received reports of field failures of the 7.3728 MHz oscillator on the TNC. The symptoms are improper reset of failed operation. Careful analysis of the problem indicated that the oscillator U1A, U1F, X1 and assocaited components on the kit board, would free-run at 22 to 49 MHz. The problem has been duplicated and confirmed here in Tucson. There is a fix, however! It appears that the fairchild 74LS14 ICs shipped in kits starting about January may exhibit this symptom, sometimes when cold and starting up, other times when warm. The cure is usually just a power off/on cycle or two. This is not a fix, however... We used several Fairchild Hitachi anf National parts as well as the Signetics 74LS14s that are scheduled for the next 1200 or so kits. The Fairchild parts are the most susceptible to this fault, but the others may do it as well. The fix is as follows: 1) Replace R12 (1.5 k) with a 1k 5% carbon film resistor. 2) Replace R13 (1.5k) with a 1k 5% carbon film resistor. 3) Add a 20 or 22 pf ceramic disc capacitor at location Cy (previously unused). With the above mods, all Signetics, hitachi and National parts tested performed flawlessly at hi and low Vcc, and at superhot (can't touch it) through very cold (lots of ice condensed on the part in the dry Arizona climate) temperatures. All but one Fairchild part worked under the same conditions, and the one that didn't simply had excessive frequency drift (about 50 kHz - waveform still clean, but may have drifted out of crystal-controlled lock). If the unit is still malfunctioning in this area (unlikely), replace the 74LS14 with a non-Fairchild part. All kits shipped after this date (6 April 1984) will have the proper parts along with an errata sheet describing the fix. PLEASE SPREAD THE WORD OF THE PROBLEM AND THE FIX AS WIDELY AS POSSIBLE IN YOUR LOCAL PACKET AREAS! A secondary benefit may be a reduction in the rate of NOVRAM "forgetting" symptoms -- this hasn't been confirmed, however. Feedback in this area would be appreciated. Some oscillators before the mod had the symptom of starting at a very high frequency, then locking to the crystal after a few hundred milliseconds, often requiring a toggle on the RESET switch to bring the TNC up. Thank you for your assistance in spreading the word... Lyle