1987Revision E testing
- Aug 2, 2007I got the first batch of Proto 'E' boards in, and built up six of them
for testing. Like Proto 'D', it's the same form factor as the previous
revisions but the components are all surface mount, with the exception
of the connectors, crystal, and LEDs.
Changes from previous revisions include:
- No tuning pot. In my testing, I found that there wasn't any real
benefit to being able to adjust the RX tuning, compared to having it set
with a precision resistor to the calculated optimum value. It seems to
perform well across the whole temperature range, and it eliminates one
of the more failure-prone parts.
- Active RX filter. This takes the place of the passive filter network,
and seems to squeeze a bit more performance out of the demodulator.
Proto 'D' had a quad op amp, but Proto 'E' just has a single op amp in
an SOT-23 package.
- Ferrites on all off-board signals to reduce RFI.
- Red and green LEDs in the TX and RX holes instead of pots. A solid
green LED on receive, separate from the activity LED, makes it easier to
see what's going on and frees up the ACT LED for other signals.
- Jumper arrangement changed to match the OT1+, with the exception of an
EQ jumper in place of A1/A2. This eliminates a few assembly steps and
takes up less space.
So far, performance is looking good. I did discover something
interesting - if you remove the 510k resistor, it actually seems to
improve RX performance a bit. I haven't confirmed if this works on the
previous versions, and testing is inconclusive so far on the T2-135.
I was just looking over my test result archives, and it occurs to me
that the decode performance is better than any other '2211-based TNC
I've seen, so maybe it's time to finally declare it 'good enough' and
get this thing in production!
- Next post in topic >>