Jeshua Lacock wrote: > > I guess the problem is I don't know all the terminology. > > Yes, I guess that is correct. I assume the 'open-loop steppers or servos with > LinuxCNC closing the loop' is the best setup? For instance, that would allow > the fastest control, and double as a DRO when the power to the drives are > off, correct? > > In which case, I don't mind rewiring the encoders. I just want whatever is > best. > > Yes, that is what my Gecko Interface does, the encoders are powered from the Gecko interface, send signals to the Universal Stepper Controller, and also makes an opto-isolated copy for the Gecko 320 drives. It also watches the fault signal on the G320s and powers them on and off when going in and out of E-stop mode.
By including LinuxCNC in the servo loop, you can get tighter position control in some cases. You also can use Halscope to know what the following error is under various conditions. Without that feedback to the computer, all you know is that following error is less than the Gecko 320's 128-count limit. That can be a lot of error, depending on encoder resolution and the drive mechanics. On Apr 9, 2012, at 3:40 PM, Andy Pugh wrote: >> If the G320(?) has an f-error output then using that achieves much the same >> result as wiring the encoders back to LinuxCNC. >> Well, no. The fixed error limit on the G320 can be unacceptable in some cases. For instance, assume a 500 cycle/rev encoder on a motor directly driving a 5 TPI leadscrew. In quadrature, you get 2000 counts from the encoder x 5 TPI gives 10000 counts/inch. 128 counts equals 12.8 mils or .0128" which is a lot of error. That's just one example, although fairly typical numbers. With a lower resolution encoder or a coarser leadscrew, it would be worse. LinuxCNC provides a sliding scale following error limit, where higher errors are allowed at higher speeds. So, you can set it to tolerate only a small error at cutting speeds, but a larger error at rapid traverse speed. The G320 can't do that. Also, you get an adjustable I term in the PID, plus FF1 and FF2 to help reduce following errors. Jon ------------------------------------------------------------------------------ Better than sec? Nothing is better than sec when it comes to monitoring Big Data applications. Try Boundary one-second resolution app monitoring today. Free. http://p.sf.net/sfu/Boundary-dev2dev _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
