Update:
Sorry to bring back old thread, but the information is still valid. Just want to pass on what I have been doing and the results I have been getting.
I have been doing single axis master 2 ways.
1. I have an end of arm tool that I attached 3 dial indicators. I also center-punched 3 spots on a large section of steel on a nearby wall.
I set my indicators to all read zero when the robot is in position (mastering to a known point). When I need to check zero, I can move it to this position and check the position of each axis and compare it to my master numbers. (Useful for confirming master or when master is lost and the robot is in an unknown position). If the numbers are different, I just punch in my master numbers on the axis that are off and I'm good to go. I have this master saved as a position.
This position is a stretch for the robot...it couldn't be much farther away. My MAIN concern is Axis 1-3 and 5. This distance from the robot exaggerates any errors in position.
2. When replacing harnesses, I will put the robot in the most ergonomic position I can (usually straight out horizontally), and record the position of each axis. After replacing the harness, and clearing all alarms (rotating each axis etc), I will punch in the numbers I just recorded (I just take a pic of the screen with my phone), and single axis master it right where it sits.
Once that is complete, I move it to my mastering position described above to check the accuracy of the mastering.
Results:
Method 2 will get me to within .005" on each gauge.
Touching up with method 1, I find that if I move the robot to the same position a number of times (and at different speeds), it will vary by .002-.003" So my initial master was really only off by .002" when the accuracy of the robot is factored in.
I wrote a small program that moves the robot to the main master position, and 4 other positions that I marked out in my robot cell, then I also have it stop basically at zero for each axis to show positions in relation to alignment marks (yes, very redundant) in total I check it at about 7 different positions to be sure. Both of these methods have worked exactly the same after mastering the robot over 20 times from all sorts of positions, never caring if 2 and 3 were at zero or anything else.
This proved invaluable when someone accidentally moved 1 robot to all zeros, and then did a zero axis master.....on the wrong robot! (it was not anywhere near all zeros). I just eyeball it to what looks like zeros. Run my mastering program, and touch up the first position, enter my saved numbers, and single axis master. NO touch-ups were needed.
This all might not be anything new to some of you, but for others this might help.