The unit also delivers only very rough approximation as to azimuth or bearing from the user, and even with the screen displaying a dot for the position of the intruder...the refresh rate is slow enough that traffic is often not where the unit says it is because the angular difference has changed in a turn.
I still reckon that the errors you see are either because of your location or the location of the PCAS.
I use mine in the southern UK where there is pretty much 100% radar coverage, and in my experience in straight and level flight the azimuth info in a 270 degree forward arc is very accurate. Update is as per the radar sweep, as per the other day I was flying parallel to the air ambulance chopper and it's position was exactly as indicated on the 496 screen, when they started to descend into their landing site, altitude changed immediately as they went down.
I was taking to a military controller with a traffic service and he warned me of some military aircraft performing instrument training. It was 2000' over cast and the PCAS picked up the two recovering aircraft before they broke through the base of the cloud. The PCAS had told me exactly where they were, I'd been following they yellow spots on the gps screen, and low and behold they appeared as expected.
The only inaccuracies I have experienced tend to occur when the traffic is right behind, and I guess this is due to the "blind spot" of the rear fuselage. The Zaon can determine azimuth to 45 degrees, I guess based upon received signal strength on two of the 4 quadrants of the antenna. It would be better with a hull mounted antenna of course, but what would be even better is if we all fitted FLARM and started squirting our GPS coordinates out.