I am having issues with the waypoints in the program. I have captured the robot tcp position and the joint position and they all read fine up until 4 or 5 decimal places but the robot itself is way off of where it should be. The fixture has been secured and is not moving, the robot base is secured and has also been measured and is not moving. Is it possible that there is a brake that is slipping over time and not registering in the encoder? We are using this robot with a visumatic screwdriver. I notice that if the robot sits idle over the weekend then the z direction changes drastically when started back up put the position reads the same.

Sounds like a TCP issue.  The brakes in the robot are physical stops, there is a "spider wheel" that spins with the output shaft and a pin on a solenoid that when active pulls the pin out of the brakes.  That is why when you have the robot powered off you can still move the joints slightly, the spider wheel it turning slightly and bouncing off of the pin as two different teeth engage the pin.  Do you have more than 1 TCP setting in the robot?  If so, are you dictating on the moves which TCP to use, or are you just defaulting to the active TCP?  If you change the TCP the pose points will stay the same but the tool will appear to go to a different position even though it actually is not.

Thank you for the information on the brake. No, we are only using 1 TCP and with the brake being a physical stop is it still possible it could be an encoder issue?

Anything is possible but not something I have seen in the field.  My understanding is there are two encoders and so both would have to slip the same amount and I am not sure how that would even be possible.  Has the robot seen any significant external forces, either a hard crash or someone forcing the arm when they should not have?  Maybe the harmonic drive is damaged, there is a very thin foil in the joint that can get twisted up and would cause the arm to physically be in a different place than it should.  You should be able to test either of these theories by sending the arm to the upright arm home position where the arm is pointed straight up.  Then measure the angle between the joints, they should all be in a line and 180 degrees between them.  You could also set the arm at 90 degree angles and measure the actual arm versus what the system says it is

The robot has never seen any excessive forces. If the points get way off then the jaws of the TCP with bump into the jaws of the fixture and fault out the robot but it is moving quite slow and just requires a quick reenable. I will take arm to the home position today and get some measurements to test out the harmonic drive. Thanks again for this information.