Date: Thu, 27 Oct 2005 02:32:30 -0500 (CDT)
Subject: serial comms ok
X-UID: 97
Content-Type: IMAGE/JPEG; NAME="img2391.jpg"

Sending bytes from a GPIO (general purpose I/O) pin on the computer board 
to the microcontroller's serial RXEN (receive) pin is ok.

Sending bytes from the microcontroller's serial TXEN (transmit) pin to a 
GPIO pin on the computer board is ok.

Using C instead of a shell script, computer board jitter is reduced to 
about 25 microseconds. This is low enough that serial communications has a 
low error rate (a bad byte every few seconds) even though there is no 
hardware buffering (everything is done in software - note that the Linux 
kernel on this board has no virtual memory system so should approach a 
hard real time system in performance). An 8 bit CRC (cyclic redundancy 
check) will be enough for error detection.

The actual pulse width sent from the microcontroller is about 335 
microseconds. I had thought that it should be 416 microseconds. This a 
lesson - trust your oscilloscope more than your calculator when it comes 
to timing electrical signals.

In the picture, the yellow trace is a byte sent from the computer board to 
the microcontroller. The blue trace is the same byte (from a previous 
frame) sent back to the computer board.

Now the technical work centers around the microcontroller firmware. But 
due to the approach of winter, I'll paint the frame Friday night and let 
it dry over the weekend. I haven't decided on the mounting for batteries 
and electronics over the frame. It would be nice to know this before 
painting so I can drill any required holes. I'm thinking that I can't know 
the right thing to do so may rely on zip-ties and adjustable straps 
instead of bolts.