The Haas DWO function relies on a very accurate measurement of the relationship between the B-axis and the C-axis. This relationship is heavily dependent on how precisely the machine is leveled. The UMC is configured with probing as a standard feature and the CNC control has a conversational probing template that walks the user through the process of recalibrating MRZP offsets that drive the DWO calculations. Bear in mind that the process is only as accurate as the results of the probing cycle. The WIPS work probe is extraordinarily accurate, but it is not a laboratory-grade inspection device – nor is the inside of a machine enclosure as clean as a laboratory. It is reasonable to expect a couple tenthousandths of an inch of error one way or the other when probing the included calibration tooling ball. The MRZP calculations probe the ball at many different locations to minimize the effects of these tiny probing errors, but there might be small amounts of error nonetheless.
Recognize also that the B- and C-axes of the machine are remarkably accurate; but like the work probe, they aren’t perfect. No rotary table from any manufacturer is. Understand that an angular positioning error of 20 arc-seconds, while tiny, turns into a linear positioning error of 0.001” at 10” from center and 0.002” at 20” from center.
If the MRZP offsets have 0.0005” of error in the X-axis direction then a hole bored halfway through a part at B90 and then finished from the other side of the part at B-90 would have 0.001” mismatch just from the slight inaccuracies in the MRZP offsets. But B-90 is outside of the travel limits of the UMC. This kind of feature needs to be machined halfway through at B90 and C0, and then finished at B90 and C180. So now the part might have a small true position error along the X-axis from any inaccuracy in the MRZP X-axis and Z-axis offsets combined with a small mismatch from both sides in the Y-axis direction from any inaccuracy in the MRZP y-axis offset. Now add in any small deviation that could come from B axis and C-axis positioning errors and it is easy to see that true positional errors of a few thousandths of an inch aren’t unreasonable when using the machine this way.