Peter Nachtwey
Member
NO,NO,NO!!!!!
Ethernet encoders should not be used for any demanding motion control. This application seems to be like a flying shear where one actuator must accelerate to match the speed of a master actuator/encoder. The master encoder should have very fine resolution and be deterministic. Ethernet is not deterministic nor does this encoder seem to have enough encoder counts per revolution. This makes it difficult to computer an accurate master speed. The master speed must be accurate because it is used as the target speed for the slave axis. If the master speed is not accurate then it is impossible to use the derivative gain on the slave axis. It is also impossible to compute a good feed forward for the slave or acceleration feed forward.
Ethernet encoders should not be used for any demanding motion control. This application seems to be like a flying shear where one actuator must accelerate to match the speed of a master actuator/encoder. The master encoder should have very fine resolution and be deterministic. Ethernet is not deterministic nor does this encoder seem to have enough encoder counts per revolution. This makes it difficult to computer an accurate master speed. The master speed must be accurate because it is used as the target speed for the slave axis. If the master speed is not accurate then it is impossible to use the derivative gain on the slave axis. It is also impossible to compute a good feed forward for the slave or acceleration feed forward.
The velocity feed forward should be about 100% on Rockwell motion controllers. I know, remember I wrote the code for the HYD02 and M02AS. The problem is what I described above. The master velocity is for the velocity feed forward for the slave. If the master velocity resolution is as coarse as shown above then the slave velocity feed forward will be just as coarse.After a few days I finally figured out that the feed forwards in the axis setup were at 100%