Re: [CAD_CAM_EDM_DRO] Re: Combining shaft+linear encoders

Nicolas Benezan wrote:

>
> Wow, impressive. How do you detect a broken encoder cable?

I don't. If the encoders aren't working, and this has happened on occasion, it is
pretty obvious right away. Allen-Bradley had elaborate schemes for this on their
7320 control. They used differential encoder signals, and if the A and A/ or B and
B/ signals were at the same level for more than 1 uS, it would cause a fault.
Also, they derived a tach signal from the encoder, and compared it to the actual
DC tach signal. If the DC tach ever had a magnitude greater than the derived
velocity, that would also cause a fault. These faults would cause the E-stop
logic to trip the servos offline.

> Btw. do you think that velocity feed forward improves control response
> remarkably?

Yes. The original PID calculations in EMC were flawed, and I wanted to compensate
for finite amplifier gain. Adding some FF1 did quite well, and there is a graph
of actual position superimposed on commanded position, during an acceleration/
run/decelerate cycle on my web page showing the performance with this.
See http://pico-systems.com/servo.html
You can't see the commanded, the slight jitter and quantization error obscure that
trace under the actual. But, it sure was following the commanded position
tightly!

Jon