Modifying the TAPR 9600 Baud Modem for
Use as a 19.2kB Digital Repeater Controller
I just finished modifying one of the TAPR 9600 modems for use in our 19.2kB, no-scrambler, repeater. It was pretty easy, though there are still a few tweaks left to do.
The mods break down into 4 or so steps. We did this for a repeater, so these mods assume that the bit regenerator and internal clock options are installed.
1. Changing clocks for 19.2kB:
a. Bend up pins 5 and 7 on U6.
b. Use a jumper from the socket for pin 5, to the bent up pin at pin 7.
c. I had to set jumper JP3 closed for x16 clocking. I'm not sure if this is correct, as the pin 7 clock is 2x the pin 5 clock. Maybe a glitch in the PAL? Or something I'm overlooking...
2. Disabling the scrambler (no, I won't repeat the arguments about why we're doing it this way):
a. Remove U8, U10, and U11.
b. Either put a jumper between U11 pins 11 and 13, or make sure you do the DCD gating mod described below.
c. You can pick up unscrambled TXD at U11 pin 1.
3. Bypassing the analog bits to provide TTL I/O:
a. We've already done some of it. TXD comes out of U11, pin 1. RXD can be inserted at U4 pin 1.
4. Gating the RXD line with DCD:
This is necessary with some drivers, notably the P1 version 1 card and the DataEngine, to prevent RXD chatter on noise from driving the CPU crazy.
A. Two options here, depending on whether you like positive or negative logic.
Either insert a 74HC00 in U11 with pin 12 bent up, and run a jumper from the bent pin 12 to U12 pin 9
-or-
Insert a 74HC08 in U11 with pin 12 bent up, and run a jumper from the bent pin 12 to U12 pin 10. I had an '08 handy, so this is what I did.
This will quiet the RXD line on the TNC header (P3, pin 17) when DCD is not asserted. This seems like a good idea, even if your TNC/card doesn't need it.
5. Shortening the watchdog timer:
a. I reduced C22 from 100uF to 47uF, which results in about a 7 second timeout.
If you're duplicating all these steps, you can leave out U4, U5, U8, U10, U11, U17, U18, U20, and U22. Note that you may replacing U11 with another chip. You can also leave out all the associated components in the analog part of the circuit.
Problems/things left to do:
At the moment, I'm bypassing all the input filtering to keep the square waves as square as possible. I don't think this is necessary, and I may build the RC filters at P1 back in.
I don't know what's going on, but the RX clock rate being fed to the TNC header (P3) is running at 9600 Hz instead of 19.2. My guess is that this is related to the need to close JP3. As a result, the attached TNC or PI card needs to run with internal clocking and can't take advantage of the clock on the modem.
If anyone has ideas on how to solve this problem, I'd sure be interested.
Finally, I wonder whether the time constant on the DCD circuit (C24/R25)
should be changed because of the doubled data rate. That might speed up
the DCD attack time to better match the bitrate. On the other hand, we
don't see
An update on my conversion of the TAPR modem...
We noticed that the RXD line loaded down the input a little bit. It appears
that the PAL at U14 has a fairly low input impedance. Things seemed to work
ok, but there's an easy way to add some buffering.
If you have the bit regen kit installed, there are three unused sections of
U3 -- a 74HC14 schmitt trigger -- available. I routed RXD to pin 1 of U3
and took the output at U3 pin 2 over to U4, pin 1. Now RXD has a high
impedance input, with the advantage of the schmitt trigger to boot. (Note:
this is only of value if you're driving the modem with a digital signal.)
And another quirk... it seems that the clock chip at U6 (the internal clocking
option) is touchy. Sometimes it doesn't oscillate, and after random hand-
waving it will take off. Anyone else experience this? The circuit looks
pretty straightforward; I'm almost wondering if I don't have a flaky crystal.
(Later)