Hello
Is there a reason that milling heads are usually mounted at 90 degrees to the end joint?
My guess was that it has to do with accessible work space, singularities and the practicality of mounting a big motor. I have seen some bots with the milling head "in line" with the 6th axis, but those tend to be bigger bots. Whats the thinking behind the mounting choice?
With all due respect I submit the controversial opinion that the machine designer is lazy, has little imagination beyond right-angle constructions, or little real-world experience with robot applications.
I fought singularities for a number of years. Once I had to "re-do" a machine cell that was installed & worked, got relocated in the factory and re-assembled, and afterwards did not work. Investigation ultimately led to the realization that all feeder tables, end effector structures, and working fixtures were at orthogonal 90-degree angles. This presented a target-rich environment for singularities because the probability of a singularity occurring was greater with those orthogonal constructions. The techs moved it and bolted it together badly and lost all the setup precision. After the machine was relocated, the angles no longer were at 90-degrees and all the calculations were hosed. All of the work cell locations were reconstructed with effective use of user frames on all pertinent locations.
I started adding a few degrees of offset on almost every angle: 7, 15, 30, and sometimes odd numbers for fun. Yep, machinists would scratch their head and utter "stupid engineer" comments. But singularities almost never occurred again because my constructions significantly reduced the probability of a singularity. All locations were taught with frames, so movements & shifts no longer were a big deal.
I have also experimented extensively through CAD modelling to develop methods of an optimal mounting scheme for EOATs that is NOT orthogonal. I don't believe a single design exists for all cases. I have seen milling spindle mounts with a bolt-able ball joint for variable orientation. I have modeled EOAT adapters with actuation to give different angles on command during challenging robot apps. Cool idea, but may be delicate and not survive hard crashes.
On the other hand...90-degree mounts will give maximum reach, will usually produce conditions of maximum reaction torque loads on the wrist, and are easy for the shop guys to fabricate.
TD is entirely correct.
However, I would point out another factor: For the milling unit shown in the photograph, any change of the mounting angle away from 90deg would move the TCP (in this case, the tip or point-of-contact of the end mill) further from Axis 6, which will increase accuracy error issues. Going to 75-80deg to reduce singularity risk, as TD said, would be a minor issue, but (in the worst case) using an angle of 0deg would place the entire length of the spindle unit between Axis 6 and the TCP. This would be a worst-case scenario for accuracy, rigidity, balance/inertia, and other factors.
So, there is a strong motive to keep the distance between the Axis 6 mounting flange and the TCP minimal, but this must be balanced carefully against the risk of singularities and other kinematic issues. Attacking this problem robustly should be done, ideally, with accurate modelling and motion-simulation software, by someone who understands what danger signs to look for that the software will not automatically catch.
