[sinha_nsit]

I am trying to understand how interpolation works on a CNC machine (I am not from electrical/electronics side). Let me first say what I think I understand correctly.

Let us say a stepper/servo motor turns by 1.8 degree by one DC pulse. Let us assume that the desired motion in the XY-plane is tapered in the ratio 3:1. This means that for every three pulses supplied to the X-motor, the Y-motor would be given one pulse. This is achieved by regulating the frequency of pulses. The resulting motion, however, is not jerky because of inertia of mechanical elements. Therefore, even with discrete pulses, both the motors rotate with constant rpm's (during mid-path motion in linear interpolation).

If the above statements are correct, how can one explain the shown toolpath in the attached figure (taken from a Fanuc manual). It looks as if the X-motor continuously moves, but the Y-motor starts and stops in the time ratio 1:2, indicating jerky movement in the Y-direction. Is the servo output indeed jerky (at micron/nm level), unaffected by mechanical inertias? In fact, this appears to be the case.

[txcncman]

Yes, there is "jerky motion" in machine movements. The jerky motion is at a frequency equal to the steps per second of the servos. The result is beyond human senses to detect, thus deemed acceptable during inspection and use.

[sinha_nsit]

Thanks for information.
Mechanical inertia plays no role possibly because of micro/nano-level steps.
Therefore, there is no smoothening effect in the resulting motion.

[tc429]

sometimes cosmetcs will show this...we run aluminum wheels, mounting flanges are often called out slightly concave, but when turning stress relieves the part it moves opposite- often extremely slight tapers get programmed, if its like .001 in 2" of actual x motion, there will be 10 little lines where the servo moves in its .0001 command increment- note, there are servos, not steppers, but still, same thing.

on some of our servos, we changed resolution to 1/10- a tech had 'special' software that allowed us to see we were clipping the servo gain due to stickslip on .0001 moves, and even though the control resolution was only .0001, the servo software could be configured to x10, and it would take the .0001 move at .00001 counts to ramp into it. hocus pocus like this was why I still love analog servos over this serial/simulated stuff...we'd fought that problem 2 weeks before hiding it in invisible detect units that were only identifdiable with secret tools...uuugh.

inertia does damp the movements- face it- it has to. the servo can feed at high pulse rates, its certainly not stopping/starting at every pulse...at slower speeds on steppers, it will start/stop, but you will hear/feel the hum as it accelerates into smooth rotation. some pulse/direction controlled servos can be tuned for motor inertia/response so it gives quiet/servo smooth operation even with stepper input pulsetrains.