Re: [CAD_CAM_EDM_DRO] Re: A Modest Proposal!

KalleP wrote:

> I assumed the basis for cheap servo operation in DIY CNC was handed over to the CAM software totally. If this had been the case then the CAM would have to handle the encoder sampling which could be quite high with large line counts.

HUH? How can this be? Normally, the term "CAM" is used to mean a
program that converts
a design from CAD (drafting or drawing) to G-code, a file that describes
tool path movement
to make the part described in the drawing. The CAM is run BEFORE the
part is cut, not
WHILE the part is cut. The program that would be reading encoder
position and controlling
the machine is usually termed "CNC" for Computerized Numeric Control.

Sampling encoders count by count with software is a big mistake in most
cases, as
the software can miss counts unless the sampling rate is very high. it
is much better
to use dedicated hardware to do this, so that counts are never missed.

> If the encoder feedback and velocity management is synchronised somehow with hardware to let you do actual point to point multi axis translation then servos drives certainly have a huge point in their favour with any platform.

Yes, LinuxCNC does this, and can use software encoder counting or a
variety of external
hardware to do it much better.

> I have some nice low inertial motors left over from 9-track tape streamer capstan drives that were made for fast moves (they even have a forced air cooling fitting) that are crying out for servo use.
>

Yup, tape drive reel motors are great for servo machines, low speed and
high torque.

> I was also under the impression that with reasonable micro stepping some small timing jitter would be mostly averaged out on the moves (even multi axis interpolation) due to the inertia of the system. Obviously it will start to generate some weird noises if the steps were to have an uneven clocking, I would guess that accuracy would not suffer much but may be wrong here as well.
>

Micro stepping does reduce the harm that rough step timing causes, since
each step is
a smaller increment of movement. But, it also raises the demand for
number of step
pulses per second to get the same speed. LinuxCNC also supports some
external
step pulse generators that relieve the CPU of this step timing overhead.

Jon