Todd Zuercher wrote: No, Linuxcnc will not account for that s-curve acceleration, it will be seen as a following error.
If you add the encoder feedback to Linuxcnc (and use it) Linuxcnc will try to correct for the following error introduced by the drives acceleration curve. End result you will probably see spikes in the following error at points where acceleration changes occur.
You did not really get the right hardware to read encoder signals. You would need something with encoder inputs like a 7i85, 7i85s, or 7i77.
All that said, you probably will be just fine using the 7i76 with no encoder feedback and setting the acceleration curve setting to a modest setting (less than 5 probably 2 or 3). You may want to experiment a little. You might be able to connect one motors encoder to the 1 encoder input on the 7i76 for testing to see how things work.
I figured it was too good to be true. The drives are closed loop to the motors via the encoders so no need to also include Linuxcnc, they just correct if there is a variance in commanded Steps to actual steps to ensure no steps are lost, they work very well like this, you can actually force the motor out of position and the drive will return it to the correct position (providing you don't fault it.)
I figured the motion planning would need accurate acceleration values or it might begin to make some undesired and out of sync movements between axis.
I guess I will play with it some and see what happens on a few parts with different acceleration settings in the drives and see how the part tolerances come out.. probably most critical on circular interpolated moves..
Thank you for the insights!
Chris
