作者 bob ball 就我對這個 tnc 的建造, 給我的電郵: building tips from wb8wga directly:
on the modem chip performance....
i have had very good luck with the modemless design. i have set up tests
between 2 of the units passing a packet back and forth for many days and
had good success. i also have used it with very weak signals (less than
1 s unit) and it works well. i have some done some very imperical
comparisons with mx614p unit i built and couldn't see a lot of difference.
(意譯: 經多日反覆的與有 mx614p 的設計比較下, 看不出與那個tnc有什麼分別, 而且就算訊號強度少於 s1, 也能成功操作。)
i am however, patiently waiting to get others experience. as a result of
the article, there are a lot of these being built right now and i will
be very interested to get feedback on this. one user has already
suggested a 2 resistor addition to add some hysterisis to the comparitor
which should eliminate any high freq jitter and improve performance.
(意譯: 有人建議在 comparitor 加多兩個電阻, 以減少高頻音噪, 改善效能。)
some things i have noted that can cause bad performance..
1. ground loop problems. depending on how the radio, tnc, and terminal
program are powered you can get big 60 hz currents(or is it 50hz
there?) on the ground circuit. with these simple designs, there is no
isolation on the audio circuits. i had one unit that worked poorly that
i finally traced to a ground problem between the radio speaker output
and the mic input. wall wort power supplies also seem to sometimes cause
problems.
2. high freq oscillations in the comparitor. as i mentioned, i think
this can be corrected by adding some hysterisis in the comparitor circuit.
3. too slow of a pic clock speed. if noise is present in the signal, the
interupt routine must be fast enough to recognize the glitch, throw it
out and resync on the packet frame. this is especially critical when it
is detecting flag words. my 20 mhz pic16f88 units appear to work better
than the 628 units i was running at 10 mhz.
(意譯: 已知的會影響效能的事 -
1. 接地問題, 因為設計上沒有分隔電路, 有可能因為接地有做成訊號干擾, 如 switching power 的 gnd 或是 mic input/spk output 的 gnd 雜噪問題.
2. comparitor 的高頻振盪問題, 之前有所提及, 可加上些電阻解決.
3. 太慢的 pic 時速, 以 16f628 的 10mhz 比較, 16f88 20mhz 做的會更好)
i believe, that if the design is not optimum yet, it can and will be fixed.
it is amazing what can happen if you get a bunch of hams working
together on a problem like this! the digital filtering in the pic is
very crude at this point and could be strengthened. there may be cases
where aliasing is occuring. what is for sure though is the mx614 has to
go if we want to get something cost effective.
(意譯: 設計再有改良空間, 而為了減輕成本, 不用mx614是勢在必行, 但也是加強pic 對數碼 filter 的處理)
