Re: Jon Elson's comments:

"Arne Chr. Jorgensen" wrote:

> Why not post the design you had in your #1 comment,
> and tell more what is wrong with it.

Well, it was a modified stereo amp. Basically, a very boosted op amp,
with lots of gain, about 2-3 Amps output up to + or - 36 Volts.
But, it was not able to put the machine on the exact position desired,
due to friction. So, up the gain, right? But, then it overshoots, and
starts oscillating with such magnitude I thought it was going to tip the
whole Bridgeport over! Why? Because, it starts out off position, and
the command out of the CNC control keeps going up until it breaks
static friction, and then it suddenly starts moving so fast it overshoots
the mark, and has to repeat the process going the other way.
Very unsatisfactory. The reason, of course, is that there was no tach
to tell the servo amp when the static friction had broken loose.

>
> 2: ) Your second comment:
>
> I did not think of that ! But do you often have to tune them ?

Always. ALWAYS! You'll NEVER see a servo system that just installs
without tuning (at least the first model).

> 3: ) This you have to explain ! "we now know the LS7166 encoder
> counter chips
> occasionally read glitched data into the CPU " - I have never
> heard of this problem, is it serious ?

Not anymore. But, all users of EMC with the Servo-to-Go card report
'clicking', which is sudden large velocity commands out of the DAC
outputs from the card. Nobody knew what they were, or where they came
from, and most people didn't care. I had high-bandwidth servo amps,
and my earlier PWM design would trip an overcurrent fault every time
that EMC did this. After getting the real-time logging feature working
(there is essentially a digital oscilloscope built into the motion control
part of EMC) I found that the problem was coming from occasional
large errors in the reported position coming from the STG card.
It only lasts one sample, and then goes back to normal, but that causes
the next analog velocity command sample to be wild.

Jon