Duemilanove Atmega328 Communications Issue

I recently installed the board (About a week ago) on vista and was able to install a few sketches with no problem. Today, I tried to load a new sketch, and noticed that I would plug in my usb cable (The one that came with the board) and I would often get the driver to show up in the device manager under USB drivers rather than "COM Ports" as it should.

It had a little error yellow symbol over it and when I checked under properties, it showed error message:

This device cannot start. (Code 10)

I tried plugging it into different USB ports as well as re-plugging it in several times and restarting the computer several times. Sometimes, it came up under COM Ports as COM5 and sometimes it would show up under USB drivers, and sometimes it would be yellow and device cannot start, and sometimes it would show up with no errors. When I got it to show up with no errors and tried to upload a new sketch to the board, I got the message:

avrdude: ... ser_send(): write error: sorry no info avail

and then when I tried to use Tera-Term (Like old hyperterminal on vista) to view data, it says the COM Port is In use by another device! What the heck is going on here?

I even installed all the drivers on my XP computer to rule out OS issues in the other room, and have the same exact issue!

One thing of note is that the sketch that is currently rtunning on the board reads analog data on ports 0-2 and sends them out serially through the USB cable every 500ms forever (I verified that it works last time I used Tera-Term when I first downloaded the sketch into the board, so there are no code bugs - It did what I wanted it to). Immediately on powerup this is running - Could this interfere with the board detection process and be the cause of my inconsistent board identification in windows? I wonder...

Maybe there is a way to clear the sketch from program memory without the usb cable to fix it?

By the way I have the newest Duemilanove board with the Atmega328 chip in it. Just got it last week.

Any help is appreciated!

Could this interfere with the board detection process and be the cause of my inconsistent board identification in windows? I wonder...

Not sure it has anything to do with your problem, but one way to eliminate the serial data stream being outputted from the existing program in your Arduino is to press and hold down the reset switch on the board before you plug it into the PC USB port. This will hold the AVR processor in reset mode as long as you hold down the reset switch, so no code will be running even though there will be power to the board and the FTDI USB serial converter chip should still be able to initialize with the PC.

Lefty

I was messing around with cycling the board (Plugging the USB cable in and back out repeatedly) to see if I could establish a pattern, and after about 5 tries I got it to load up without errors and successfully downloaded the "blink" sample project into the board (This was on my XP machine). I also disconnected the power pin from board Vcc to the external accelerometer I was using as well as the 3 analog outputs from the accelerometer to the arduino analog input pins (ports 0-2) prior to my last attempt.

Once that was done, it became stable and I was able to reload my communications project successfully without any further issues. I have been creating programs all afternoon now without a further incident...

I wish I had any idea what the underlying issue was because if it has happened once, it will probably happen again in the future... Thank you for the help - I will use the reset button if this ever happens again.