Date: Fri, 16 Dec 2005 04:40:42 -0600 (CST)
Subject: serial interfacing
X-UID: 112

No pictures this morning. In fact, I'm just about to go to bed and catch 
90 minutes of sleep before work. And I think I'll do nothing but sleep 
this next night. I'm so tired that I'm dangerous. I'll break something.

The reason is that I just thought I burned out more I/O pins on one of my 
computer boards. Not the end of the world but a sick feeling. I 
mirror-image flipped two wires and wasn't seeing good signal on the scope. 
Instead of strong 3.3 volt pulses, I was seeing very noisy pulses of 
several hundred millivolts.

The I/O pins on the computer board for the TTL serial are arranged this 
when viewed from above.

                     GND    RXD <-- CONNECT DATA OUT

CONNECT DATA IN --> TXD    GND

What happens when you mirror-image flip the wires around?

                     RXD    GND <-- CONNECT DATA OUT

CONNECT DATA IN --> GND    TXD

So instead of connecting to TXD and RXD, you connect to GND. This is 
typically safe and won't cause damage. I had wondered why the pins were 
arranged this way. There is a good reason.

The interesting thing is the bad signal on the oscilloscope. When I 
mirror-image flipped the wires, the scope's probe was connected with 
reversed polarity like the following.

           PROBE SIGNAL --> COMPUTER GROUND

           PROBE GROUND --> COMPUTER SIGNAL

Naively, you'd think that this would just invert the waveform. It doesn't.

I think there are two things going on.

One is that there are reverse biased diodes, actual or effective, between 
COMPUTER SIGNAL and COMPUTER GROUND. Polarity does matter.

Two is that ground is fundamentally different from other levels in a 
system. It is assumed to be stable. If ground changes, then everything 
else is affected, both signals and power. But there are no such 
requirements on signaling levels. They can be noisy. So I think that the 
noisy voltage on COMPUTER SIGNAL when taken as ground confused the 
oscilloscope.

The last thing is that I realized yesterday that the Garmin 18 LVC GPS 
receivers are not TTL serial devices. The datasheet never says they are. I 
remember reading the datasheet with a slight sense of confusion. I should 
have paid attention.

The GPS receivers are RS-232 devices with voltages limited to TTL levels. 
But they do not use 5 volt TTL serial. The Garmin 18 LVC can receive full 
serial voltages. But it limits output to positive and negative 5 volts. 
The advantage is that these receivers interface directly to a standard 
RS-232 serial port (typical for industrial equipment) and yet with two 
inverters will also interface to 5 volt TTL serial (typical for embedded 
microcontrollers and computers).

This is a very useful web page.

http://www.lammertbies.nl/comm/info/RS-232_specs.html

After reading this, it made sense.