There are a few things to check. From the simple to the complex.
Do you lock your knee? The series 2 has a bit more play in the knee and as the mass of the table changes center of gravity, it can rock.
You have said you set your backlash. Usually, I do this in a dynamic state. I make the backlash bigger by a few thou because I can measure this a lot easier at the points of the compass (sweep the circle with a dial indicator). If you do not have enough backlash, you actually get a flat at the compass points which is hard to measure. By seeing how big the jump is, you can deduct that value from the backlash you have dialed in. Since you changed the ball screws, the pitch error comp must be set to zero unless you know how the screws were ground.
Are you DOS literate? Do you know how to attach a full size keyboard? The next step would be to get to a rott directory program called PFM.exe. You can edit parameters here, run a program and watch position error AND watch the DAC command (digital to analog). Typically, you run an axis back and forth watching the DAC command and position error and adjust the drive card so both directions are balanced. I am assuming that the motors are in good shape, belts have the correct tension, gibs have the same play the entire length of the circle you are trying to mill. I installed 8249E in March of 1997 so your machine is probably 8 years old. Most machines are used in the center of travel and the way surfaces get "dished." As you get ouside of the usual travel area and it gets tighter, it may cause drag and a misposition. If it is overshooting, it is typically a drive tuning issue. This usually occurs closer to the 45 degree on a circle. Start looking at some of this and give us an update.