Randy,Also keep in mind that the accuracy is dependent on the number of satellites -and the position of the satellites relative to your ground position. It would be interesting to see if you both had the same set of satellites through the test process.
Is the GPS receiver for the mini different than the standard sized GPS receiver?
You've been saying that the distribution in the sky affects accuracy, but it just doesn't hold water with respect to on the ground testing. To wit, there is a free software program called Visual GPS View. You start the program, plug in your eggtimer serial cable. Go to start GPS, look for the serial port that is active, set to 9600 baud. Next, connect the black pin to ground on your TX/Mini and the white pin to the TX pin. Then, plug in the battery to the mini or TX.
What you will get is a real time read out of the satellites, their position in sky, relative strength and the PDOP, HDOP and VDOP values. I did it a bit ago for my TX and came up with the following screen shots. I took one each time I picked up a new satellite. I let it run for about 30 minutes, but never got better than 7 satellites:
In looking at the above plots, you'll notice a few things:
Satellites are not all in one quadrant, but rather randomly distributed. Anytime I've done this, which is typically when I build a new device, I get the same random distribution
You'll notice that the satellites on the extreme edges of the horizon while seen, typically don't establish a lock. I assume that is because the signal is weak and satellites on the horizon have to cut through the maximum amount of atmosphere
You'll also notice how the HDOP gets better with each additional satellite
To Mike's earlier point, if you run this experiment where you are, wherever that is, I'm certain you will see something similar. There are currently 31 active GPS satellites, and the system is designed for accuracy considering smart bombs use the GPS network to guide their payload to target.
Fun fact, USA-132 is the longest serving satellite in GPS history, still operational after being launched on 23 July 1997 aboard a Delta II rocket. Still getting it done over 25 years later.
One more analytical tool you can use to understand your 500 foot variance.