Index Home About Blog
From: dnichols@d-and-d.com (DoN. Nichols)
Newsgroups: rec.crafts.metalworking
Subject: Re: servos or not servos
Date: 21 Apr 1998 18:37:43 -0400

In article <353D041B.3EE253E8@sympatico.ca>,
Darby  <ddarby@sympatico.ca> wrote:
>Thank you to all you who responded on this thread
>
>            It's nice to see so many people  eager to help.

	[ ... ]


>	I have written a bit of code to do intercepts and as a result of that,
>never dreamed that a servo motor would need a tach to calculate velocity.  I
>just assumed time and position would or could provide it.  I just assumed
>	you told
>the servo how fast to go and if it was not doing the correct speed
>	[according to
>difference time and location] then it would be fed more or less power.

	This *can* be done, but it is a lot more load on the computer which
is already trying to compute the various points and the speeds needed to
maintain the SFM correctly.  Asking it to also monitor the motor speed, and
adjust the drive level according to the load being presented at the moment
is a bit much, unless you have a *really* fast CPU>

	Instead, what happens is that the CPU plus interface produce a
signal ranging between -5V and +5V (or between 0V and 10V) proportional to
the speed desired from the motor.  That voltage is taken as one input to an
operational amplifier, with another input being the voltage generated by the
tach saying how fast the motor is *really* going.  These voltages are added
(with one being the popsite polarity from the other) to determine the
momentary voltage level to apply to the motor to either keep it at the right
speed, or to get it there *quickly*.  Usually, there will be another input
which is monitoring the current through the motor to make sure that it
doesn't get high enough to demagnetize the permanent magnet pole pieces.

> Ah!  Maybe understanding what makes them tic is a little like knowing
>what goes into a hot dog.  :>)

	Yep!  There's a lot more complication that appears at first -- but
if you are buying the motor, the tach generator, and the amplifier as a
package, you don't have to worry about much.  It's when you are combining
parts for cheap that you have to worry about the internal details.

	Good Luck,
		DoN.

--
 NOTE:     spamblocking on against servers which harbor spammers.
 Email:   <dnichols@d-and-d.com> | Donald Nichols (DoN.)|Voice (703) 938-4564
 My Concertina web page:         | http://www.d-and-d.com/dnichols/DoN.html
	--- Black Holes are where God is dividing by zero ---


From: dnichols@d-and-d.com (DoN. Nichols)
Newsgroups: rec.crafts.metalworking
Subject: Re: servos or not servos
Date: 22 Apr 1998 17:34:04 -0400

In article <1998042205091400.BAA28838@ladder01.news.aol.com>,
SLEYKIN <sleykin@aol.com> wrote:
>I know nothing about CNC but a bit about electronics so this may not apply well
>but, it seems to me that an optical device similar to (or you might be able to
>gut one and use it) a mouse position sensor would work really well.  It would
>be speed independent and all you would have to do is measure the period of the
>pulses for feedback.  You could even go with a D/A converter to get you
>error/correction voltage.   Just a thought FWIW

	This is fine at higher speeds -- but when you're cutting something
which is *almost* parallel to an axis, you wind up with the motor jittering
back and forth a bit more to get enough signal from integrating an encoder
than you really want.  The output from a motor style tach generator can be
enough to control *very* slow motion.  I've got a servo amp and fairly hefty
motor which can be run so slow that you need to watch the shaft for a while
to observe that it is running.  And it is *very* smooth at that speed.  I
don't see any way that you could get that kind of smooth slow-speed
operation with the optical encoder serving as a tach.

	However -- you *do* need the optical encoder somewhere in the system
-- either in the motor assembly, or (better) monitoring the actual machine
tool's linear motion.  This needs to go to the controller so the CPU can
monitor position, while the tach feedback goes into the servo amp, to allow
the voltage from the controller to translate into a speed irrespective of
load (within the motor's capablilties, of course.)

	Enjoy,
		DoN.

--
 NOTE:     spamblocking on against servers which harbor spammers.
 Email:   <dnichols@d-and-d.com> | Donald Nichols (DoN.)|Voice (703) 938-4564
 My Concertina web page:         | http://www.d-and-d.com/dnichols/DoN.html
	--- Black Holes are where God is dividing by zero ---



From: erich10983@aol.com (ERich10983)
Newsgroups: rec.crafts.metalworking
Subject: Re: servos or not servos
Date: 23 Apr 1998 11:34:51 GMT

> The output from a motor style tach generator can be
>enough to control *very* slow motion.  I've got a servo amp and fairly hefty
>motor which can be run so slow that you need to watch the shaft for a while
>to observe that it is running.  And it is *very* smooth at that speed.

Don:  Very good summary of the advantages of servo motors.  I have tried to use
microstepped motors for our crystal pullers because of the supposed smoothness
at low speeds, but also had a conflicting requirement of very high speeds to
return the carriage to the beginning.  I needed a slow to fast speed ratio of
about 6000:1.  0.1 inches/hour to 600 inches/hour.  The steppers were fine at
the low end, but couldn't come anywhere near the top speed needed.  A 1/4 hp
servo with tachometer and encoder (for monitoring speed) works just fine.

I still use steppers for some projects, but you really need to look at the
requirements and make an intellegient choice.

Even microsteppers have problems at low speeds.  There is some variation in
speed because the windings aren't exactly at 90 degrees.  You can feel the
variations as a type of "pulsing".  I have used controls that have an
adjustable offset of the two drives to compensate, but even this doesn't seem
to take it all out.  I have gone from microsteppers to servos in our small
pullers because of this.  For optical fiber, the smoothness of pull rate is
extremely critical.  This is kind of a sidebar issue, but may be of interest to
some folks.

Earle Rich
Mont Vernon, NH



From: "Don Foreman" <foreman_don@htc.honeywell.com>
Newsgroups: rec.crafts.metalworking
Subject: Re: servos or not servos
Date: 21 Apr 1998 04:10:15 GMT

Totally unsuitable.  But I'll give you three bucks each and pay shipping --
maybe I can find a use for them....  :<)

Near as I can tell (after talking with an engineer at a servomotor company)
the attributes of a "servomotor" (as differentiated from just a DC motor)
are probably irrelevent for use as CNC drive -- I expect that's what you
had in mind.  Servomotors are designed to have high  acceleration and low
inertia. Since we're interested in driving  big masses of cast iron via a
leadscrew, the inertia of the motor is trivial.

These probably realy are servomotors if they came out of tape drives.   But
if they can handle 11 amps at 60 volts, they're some husky motors -- should
work great.


From: erich10983@aol.com (ERich10983)
Newsgroups: rec.crafts.metalworking
Subject: Re: servos or not servos
Date: 23 Apr 1998 22:17:31 GMT

>I understood that a 200 step per rev stepper had 200 distinct stopping
>locations as it made it's way around the full 360 degrees. Probably when
>you slow down it's these stops that you feel.

No, I was using microstepping.  Even at the highest resolution of 50,000 steps
per rev, I could distinctly feel the change in speed that would occur at the
basic 1.8 degree resolution of the motor.  Microstepping is great, but it
doesn't buy you any more accuracy than the fundamental motor construction.  If
the windings (or rotor magnet) is off, then so will the rotational accuracy.

Earle Rich
Mont Vernon, NH




From: Jon Elson <jmelson@artsci.wustl.edu>
Newsgroups: rec.crafts.metalworking
Subject: Re: servos or not servos
Date: Wed, 06 May 1998 14:41:26 -0500

DoN. Nichols wrote:

> In article <1998042205091400.BAA28838@ladder01.news.aol.com>,
> SLEYKIN <sleykin@aol.com> wrote:
>
>>I know nothing about CNC but a bit about electronics so this may not
>>apply well but, it seems to me that an optical device similar to (or you
>>might be able to gut one and use it) a mouse position sensor would work
>>really well. It would be speed independent and all you would have to do
>>is measure the period of the pulses for feedback. You could even go with
>>a D/A converter to get you error/correction voltage. Just a thought FWIW
>
>         This is fine at higher speeds -- but when you're cutting something
> which is *almost* parallel to an axis, you wind up with the motor jittering
> back and forth a bit more to get enough signal from integrating an encoder
> than you really want.  The output from a motor style tach generator can be
> enough to control *very* slow motion.  I've got a servo amp and fairly hefty
> motor which can be run so slow that you need to watch the shaft for a while
> to observe that it is running.  And it is *very* smooth at that speed.

Yes, I've built the same thing.  I'm using a 7V / 1000 RPM tach, driven
directly by the leadscrew.  It is absolutely amazing how slow it can go and
still be smooth.  Even if I halt the CNC control, so there is NO position
feedback, you can't move the table by twisting the leadscrew.  The
slight deflection that you cause by twisting the leadscrew is picked up
as a velocity change by the tach, and the amp resists it fiercely!

>  I
> don't see any way that you could get that kind of smooth slow-speed
> operation with the optical encoder serving as a tach.

Well, you actually can, and some CNC's do, but it is much morecomplex than a
freq-voltage converter.  But, the new DSP servo
systems have enough CPU horsepower to do all the nth-order
integration to make this work.

>         However -- you *do* need the optical encoder somewhere in the system
> -- either in the motor assembly, or (better) monitoring the actual machine
> tool's linear motion.  This needs to go to the controller so the CPU can
> monitor position, while the tach feedback goes into the servo amp, to allow
> the voltage from the controller to translate into a speed irrespective of
> load (within the motor's capablilties, of course.)

If your encoder has fine enough resolution (like .000025" ; 25 uInches -
standard CNC resolution) AND the CNC has a high enough sampling
rate (like 10 KHz) then you can skip the tach, with the right code in
the CNC.  But, to do it the simple way, a tach really works well.

Jon



From: dnichols@d-and-d.com (DoN. Nichols)
Newsgroups: rec.crafts.metalworking
Subject: Re: CNC Conversion question
Date: 18 Feb 1999 22:28:44 -0500

In article <36d6cab7.158197937@news.rdc1.bc.wave.home.com>,
Jens <psh@home.com> wrote:
>dnichols@d-and-d.com (DoN. Nichols) wrote:
>
>>>Are you driving the leadscrew through the 150:1 gear box? what about
>>>gear backlash?
>>
>>150:1 gearbox?  I doubt that he used a gearbox.  Good servo motors will go
>>that slow directly at the shaft -- no gears needed.
>
>What is the difference between a good servo and the surplus treadmill
>motors I have coming (I can't see running those without a gear box)

	Arrggghhh!!!

	Another one e-mailed and copied to the newsgroup.  Well ... I just
got through answering the first, so I still have the answer here.

 ======================================================================
1)	Many poles and segments on the commutator, so the field from the
	rotor is always at the optimum position relative to the Permanent
	Magnet field.

2)	No iron in the rotor -- just windings so the inertia is at a
	minimumm.

3)	Good ball bearing races at both ends.

4)	A similar, but smaller "motor" at the back end, which acts as a
	tachometer generator.  The signal from that is summed with the input
	(speed command) voltage to the servo Amplifier to adjust the power
	to the motor to maintain the proper speed (including slow enough
	with a 1mV input (full-scale is 10V) so you have to look carefully
	at a tape "flag" on the shaft to be sure that it *is* moving.

5)	*Many* dollars difference in new cost. :-)

	Enjoy,
		DoN.

--
 NOTE:     spamblocking on against servers which harbor spammers.
 Email:   <dnichols@d-and-d.com> | Donald Nichols (DoN.)|Voice (703) 938-4564
 My Concertina web page:         | http://www.d-and-d.com/dnichols/DoN.html
	--- Black Holes are where God is dividing by zero ---

Index Home About Blog