Hello all,
In doing some research on improving the D-Rats I have discovered a hardware error which seems to be in a number of TNCs.
I have found this documented (not as the hardware errata that it is) in two different PAC-COMM TNC manuals and in an MFJ TNC manual.
If you are generally using only the quick wiring of RX data, TX data and signal common on the packet modems then you should be OK.
If you are using the "non-standard" wiring documented in the TNC manuals they you also should be OK.
But who bothers to read the manuals in detail? And when you see a 9 pin or 23 pin serial port, you may just plug in a standard wired cable like common ribbon base cables.
If you are using a standard full wired cable you could be in danger of burning out the serial port on your computer or causing other damage to the or USB serial adapter.
Someone made a big mistake about those signals and it looks like everyone copied them instead of correcting it.
The TNCs are expecting the custom non-standard cable where the CTS signal from the computer is wired both to TNC CTS signal and the TNC DTR signal. And that the cables do not have anything connected to pins 9 and 10 of a 23 pin connector.
This is not the way that off the shelf standard full signal cables are wired, either have all signals wired straight through or the common MODEM subset of 9 wires wired straight though.
Worse is that TNCs may be incorrectly shorting CTS and DTR signals together inside the TNC, as documented in the PAC-COMM TINY TNC-2 which stated that it is based on the TAPR reference model. That is one of the things that can burn out your computer!
And also that if the cable has a 23 pin connector, that pins 9 and 10 are not connected, since the TNC may be unexpectedly providing power signals to those pins, which are defined in the specification as reserved. This should be ignored, but if the computer is doing something non-standard on those pins, that could cause a problem.
Why this is a problem, is what the standard computer interface is supposed to be doing:
The DTR pin when you are not using the serial port should be setting the DTR and the CTS signal to a value of about +3 to +15 volts.
When a program is running using the serial port, the DTR signal is typically set to about -3 to -15 volts.
The CTS signal when hardware handshaking is disabled should also be set to -3 to 15 volts, but may be left at +3 to +15 volts, which when using a standard wired cable is a short circuit of up to 30 volts.
When hardware handshaking is enabled, the DTR signal stay at the -3 to -15 volts, and the CTS signal can periodically alternate between the positive range and negative range, which is an intermittent short circuit of up to 30 volts.
It is on the expensive computer side where the damage can occur.
There are about 2 to 4 different documented ways (3 wire and 9 wire) that 9 and 23 pin connector serial connectors should be wired that will work 99.9% of all devices.
There are seems to a very small number of devices that need special wiring because the hardware designers did not implement the specification correctly. So far I have only encountered one commercial printer, and some TNC modems that require a variant from a standard cable wiring.
And also because people do not bother learning and following the specification there are hundreds of wiring diagrams on the Internet for 9 and 23 pin serial port connectors, with most of the 3 wire implementations missing jumpers for DSR/DTR and RTS/CTS signals on the 9 or 23 pin connector side.
And now I need to make and keep track of special cables for my older TNCs, so that I can test them in AX25 and KISS modes.
73,
-John
wb8tyw
