Re: [CAD_CAM_EDM_DRO] Bridgeport Boss 8 RS-232 question

Alan Hantke wrote:

> I'm trying to help a guy I know get a Bridgeport Boss 8 CNC working through the serial port. Because this does not pertain to "building" a CNC I am soliciting responses OFF the list. If I end up getting this machine then I'll convert it and will get feedback ON the list!!! Incidentally, that machine "sings" up a storm when the servo amps are on. I noticed that it is using encoders only, no tachometer. Made me wonder.... if you use a tach is the system "quieter"?

It can be. With a linear amp, the system can be continuous-time with a DC
tach, but is forced to be a sampled system without the tach. Worse, it is a
non-constant interval sampled system, which leads to serious problems
when you want both fast response and high gain. That is your singing,
I think. It could also be a switching amp that has too low a PWM frequency.

With a PWM (switching) amp, it is always a sampled system, but if the
switching bandwidth is well above the highest response of the system,
it can behave the same way. I don't LIKE singing servos, for several
reasons. When I designed my own servo amps, I went to some lengths
to get away from that instability. One of my servo amps will occasionaly sing at
about 500 Hz, but it is actually a mechanical resonance in the system, between
the motor mount and the encoder mount. I can damp it by placing a finger on
the right spot on the encoder coupling or a few other places.

> If not, what is the advantage of using a tach? Seems like a lot of added work. Now those are topics that do belong on the list. Therefore, if someone would like to comment on those (not the BPort problem) please feel free to do so.

This is not the only reason for the tach. As I mentioned above, the encoder
gives a sampled reading of position, therefore, velocity is also sampled.
The problem is that when moving very slowly, the encoder counts come in
very slowly, therefore velocity information is very slow to update, and when
a single count comes in, you can't tell if it represents a sudden external
influence on the system (like cutting tool biting into the work) or just the
next tick of the motion. With a DC tach, you get continuous-time velocity
updates, down to rediculously low velocities. At .01 IPM, which the machine
may be needed to move at doing circular interpolations or cutting angles
nearly parallel to an axis, I have encoder counts coming in at 3.33/second.
The servo bandwidth is about 600 Hz on my machine. So, the servo bandwidth
is 180 times the position update rate. No matter that the servo loop on the
computer is running at 1000 updates/second, it still only sees a new position
reading 3.33 times a second! The DC tach allows the servo amp to continue
smooth motion between these position updates. And, this really works. You
can turn off the position loop, and the servo amp will STILL hold the machine
motionless, except for the slow drift of the servo amp. I do this demonstration
with my machine, and people are amazed that it is still stable with no position
feedback! This system remains stable down to about .001 IPM, at which point
stick-slip friction of the ways takes over, and the motion becomes lumpy.

There is a way to get around this - but it requires the use of encoders with
VERY high resolution. I'm using 1000 cycle/rev encoders, giving 4000 counts
per rev x 5 TPI screws = 20,000 counts/inch, on the X and Y axes. (I have
a 10 TPI screw for the Z, so there I get 40,000 counts/inch.)
To get the same bandwidth, you'd need an encoder with at least 180 times
higher resolution (or should this by 360 to satisfy Nyquist?) or 1000 x 360 =
360000 cycles/rev. This is obviously requiring a laser encoder, and may
require a high-speed encoder counter to keep up with rapid feeds.

Jon