To anyone interested in making machines with servos:
This fellow I know does ornamental wood turning. One way of milling
helical grooves on a wooden cylinder works like this:
A lathe has a sprocket attached to the back end of the spindle. A
chain is draped over this sproket. One end of the chain is weighted,
the other end is attached to a cable. Said cable is run over various
pulleys and is attached to the lathe carriage. A spinning cutter is
mounted to the carriage. As the carriage moves back and forth the
spindle turns and the cutter, engaging the cylinder, milling a helical
groove. The pitch of the helix is changed by changing the sproket. So
this fellow was explaining this to me and mentioned that he wanted to
do this electronically. Use an encoder on the carriage to send pulses
to a servo or stepper driver which then controls the motor that spins
the spindle. He spoke with an engineer who told him that following
error would be so great that he couldn't move back and forth like he
could with the sproket arrangement. I didn't believe it so I built a
demo model with the signal generating encoder connected through a
precision differential to the motor shaft. Spinning the encoder one
direction causes the motor to spin in the opposite direction. If there
is no following error the the center section of the differential would
remain stationary. Any movement left or right would show following
error. Well, the center section does swing left or right. About two or
three encoder counts. At high count settings the amount of following
error, in degrees of rotation, is so small that it doesn't matter, and
is probably less than the windup in the mechanical version. Using
encoders with switchable counts shows more movement of the diff with
the lower count settings. So, after that long explanation, would
anybody like me to try to post a video online demonstrating this?
Eric

On Sat, 28 Jan 2012 22:18:53 +0000, etpm wrote:

To anyone interested in making machines with servos: This fellow I know
does ornamental wood turning. One way of milling helical grooves on a
wooden cylinder works like this: A lathe has a sprocket attached to the
back end of the spindle. A chain is draped over this sproket. One end of
the chain is weighted, the other end is attached to a cable. Said cable
is run over various pulleys and is attached to the lathe carriage. A
spinning cutter is mounted to the carriage. As the carriage moves back
and forth the spindle turns and the cutter, engaging the cylinder,
milling a helical groove. The pitch of the helix is changed by changing
the sproket. So this fellow was explaining this to me and mentioned that
he wanted to do this electronically. Use an encoder on the carriage to
send pulses to a servo or stepper driver which then controls the motor
that spins the spindle. He spoke with an engineer who told him that
following error would be so great that he couldn't move back and forth
like he could with the sproket arrangement. I didn't believe it so I
built a demo model with the signal generating encoder connected through
a precision differential to the motor shaft. Spinning the encoder one
direction causes the motor to spin in the opposite direction. If there
is no following error the the center section of the differential would
remain stationary. Any movement left or right would show following
error. Well, the center section does swing left or right. About two or
three encoder counts. At high count settings the amount of following
error, in degrees of rotation, is so small that it doesn't matter, and
is probably less than the windup in the mechanical version. Using
encoders with switchable counts shows more movement of the diff with the
lower count settings. So, after that long explanation, would anybody
like me to try to post a video online demonstrating this? Eric

Congratulations. You've reinvented the CNC lathe.

--
Tim Wescott
Control system and signal processing consulting
www.wescottdesign.com

On Sat, 28 Jan 2012 16:42:12 -0600, Tim Wescott
wrote:

On Sat, 28 Jan 2012 22:18:53 +0000, etpm wrote:

To anyone interested in making machines with servos: This fellow I know
does ornamental wood turning. One way of milling helical grooves on a
wooden cylinder works like this: A lathe has a sprocket attached to the
back end of the spindle. A chain is draped over this sproket. One end of
the chain is weighted, the other end is attached to a cable. Said cable
is run over various pulleys and is attached to the lathe carriage. A
spinning cutter is mounted to the carriage. As the carriage moves back
and forth the spindle turns and the cutter, engaging the cylinder,
milling a helical groove. The pitch of the helix is changed by changing
the sproket. So this fellow was explaining this to me and mentioned that
he wanted to do this electronically. Use an encoder on the carriage to
send pulses to a servo or stepper driver which then controls the motor
that spins the spindle. He spoke with an engineer who told him that
following error would be so great that he couldn't move back and forth
like he could with the sproket arrangement. I didn't believe it so I
built a demo model with the signal generating encoder connected through
a precision differential to the motor shaft. Spinning the encoder one
direction causes the motor to spin in the opposite direction. If there
is no following error the the center section of the differential would
remain stationary. Any movement left or right would show following
error. Well, the center section does swing left or right. About two or
three encoder counts. At high count settings the amount of following
error, in degrees of rotation, is so small that it doesn't matter, and
is probably less than the windup in the mechanical version. Using
encoders with switchable counts shows more movement of the diff with the
lower count settings. So, after that long explanation, would anybody
like me to try to post a video online demonstrating this? Eric

Congratulations. You've reinvented the CNC lathe.
Well, sort of. You are of course talking about electronic gearing. I
didn't invent anything. Just figured out a really graphic way to show
someone what the following error would be and how to change it. Maybe
I should have posted the above referencing electronic gearing instead
of following error. Because when I was first told about this guy's
desire to eliminate the mechanical connection I thought about
electronic gearing. I told him this and he brought up following error
and how he believed it would be a problem.
Eric

wrote in message
...

snip

encoders with switchable counts shows more movement of the diff with
the lower count settings. So, after that long explanation, would
anybody like me to try to post a video online demonstrating this?
Eric

Not really but if you like, you can simply remove the rh way cover on your
fadal, and then take the end cover off of the drive motor--inside, you will
see 3 small screws that are used to secure the resolver onto the housing...

Slightly loosen all three of the screws which will allow you to rotate the
resolver hand--note how the table basically moves in lock step whenever the
resolver is rotated.

On Sat, 28 Jan 2012 16:19:43 -0800, "PrecisionmachinisT"
wrote:


wrote in message
.. .

snip

encoders with switchable counts shows more movement of the diff with
the lower count settings. So, after that long explanation, would
anybody like me to try to post a video online demonstrating this?
Eric

Not really but if you like, you can simply remove the rh way cover on your
fadal, and then take the end cover off of the drive motor--inside, you will
see 3 small screws that are used to secure the resolver onto the housing...

Slightly loosen all three of the screws which will allow you to rotate the
resolver hand--note how the table basically moves in lock step whenever the
resolver is rotated.

That shows the machine responding to a perceived (by the machine)
positioning error. It does not show the following error. Of interest
to me and the guy doing the wood is how much lag one axis has
following motion from another axis.
Eric

On Sat, 28 Jan 2012 22:18:53 +0000 (UTC), wrote:

To anyone interested in making machines with servos:
This fellow I know does ornamental wood turning. One way of milling
helical grooves on a wooden cylinder works like this:
A lathe has a sprocket attached to the back end of the spindle. A
chain is draped over this sproket. One end of the chain is weighted,
the other end is attached to a cable. Said cable is run over various
pulleys and is attached to the lathe carriage. A spinning cutter is
mounted to the carriage. As the carriage moves back and forth the
spindle turns and the cutter, engaging the cylinder, milling a helical
groove. The pitch of the helix is changed by changing the sproket. So
this fellow was explaining this to me and mentioned that he wanted to
do this electronically. Use an encoder on the carriage to send pulses
to a servo or stepper driver which then controls the motor that spins
the spindle. He spoke with an engineer who told him that following
error would be so great that he couldn't move back and forth like he
could with the sproket arrangement. I didn't believe it so I built a
demo model with the signal generating encoder connected through a
precision differential to the motor shaft. Spinning the encoder one
direction causes the motor to spin in the opposite direction. If there
is no following error the the center section of the differential would
remain stationary. Any movement left or right would show following
error. Well, the center section does swing left or right. About two or
three encoder counts. At high count settings the amount of following
error, in degrees of rotation, is so small that it doesn't matter, and
is probably less than the windup in the mechanical version. Using
encoders with switchable counts shows more movement of the diff with
the lower count settings. So, after that long explanation, would
anybody like me to try to post a video online demonstrating this?
Eric

Absolutely. Please do.

--
The most decisive actions of our life - I mean those that are most
likely to decide the whole course of our future - are, more often
than not, unconsidered.
-- Andre Gide

wrote in message
...
On Sat, 28 Jan 2012 16:19:43 -0800, "PrecisionmachinisT"
wrote:


wrote in message
. ..

snip

encoders with switchable counts shows more movement of the diff with
the lower count settings. So, after that long explanation, would
anybody like me to try to post a video online demonstrating this?
Eric

Not really but if you like, you can simply remove the rh way cover on your
fadal, and then take the end cover off of the drive motor--inside, you
will
see 3 small screws that are used to secure the resolver onto the
housing...

Slightly loosen all three of the screws which will allow you to rotate the
resolver hand--note how the table basically moves in lock step whenever
the
resolver is rotated.

That shows the machine responding to a perceived (by the machine)
positioning error. It does not show the following error. Of interest
to me and the guy doing the wood is how much lag one axis has
following motion from another axis.


Following error is displayed on the fadal crt--to the right of the axis
dimensional read out

On Sat, 28 Jan 2012 22:58:08 -0800, "PrecisionmachinisT"
wrote:


wrote in message
.. .
On Sat, 28 Jan 2012 16:19:43 -0800, "PrecisionmachinisT"
wrote:


wrote in message
...

snip

encoders with switchable counts shows more movement of the diff with
the lower count settings. So, after that long explanation, would
anybody like me to try to post a video online demonstrating this?
Eric

Not really but if you like, you can simply remove the rh way cover on your
fadal, and then take the end cover off of the drive motor--inside, you
will
see 3 small screws that are used to secure the resolver onto the
housing...

Slightly loosen all three of the screws which will allow you to rotate the
resolver hand--note how the table basically moves in lock step whenever
the
resolver is rotated.

That shows the machine responding to a perceived (by the machine)
positioning error. It does not show the following error. Of interest
to me and the guy doing the wood is how much lag one axis has
following motion from another axis.


Following error is displayed on the fadal crt--to the right of the axis
dimensional read out

Well that's cool to know.