Re: Heck of a time tuning servo motors.

Erik Reikes wrote:

> So I've finally got my motors mounted to my machine and I'm having a
> heck of a time getting them to operate properly.
> The basic problem is the motors seem to stick as they slow down and
> pull huge amounts of current (greater than the 10amps I can see on my
> meter). Every so often they will stop in this state and just sit
> there pulling a ton of current. This of course eventually causes the
> fuses to blow.
>
> I'm getting the sneaking suspicion that my motors just don't have
> enough torque for this application. So they slow down to where they
> should be stopping and just stick there stalled unable to move the
> final last few clicks to get to their commanded position.

Hmm, this is a strange problem. You can try deflecting the motors manually
to see how high the gain is near zero. My system is a 'true' servo,
with the
CNC control computer inside the servo loop. This gives me some advantage
in that I can watch the actual position and compare to commanded. (With
EMC
I can get plots of actual vs. commanded position or velocity, which are
enormously helpful in diagnosing problems. But, you only get that with
encoder
feedback to the computer.)

My system has no deadband, and very high DC gain, so it will pull to
within .0001"
of commanded when not moving. This high gain may require either very high
resolution encoders or DC tachometer feedback (I have the latter) to
avoid nasty
dithering.

> Let me run down the specs of everything I've got so if there are any
> glaring errors someone can point out the component that is completely
> wrong.
>
> I'm retrofitting a shoptask 3-1 machine with regular acme lead
> screws. They are 10TPI leadscrews and I have a 2-1 belt reduction to
> the servo.
>
> I'm running G320 gecko servo drive units. These are powered by a CAM
> tronics 36V 25Amp supply.
>
> I've been using master5 to command the geckos.
>
> The servo drives I'm using are Leeson PMDC motors with 100 line
> USDigital encoders attached.

Ahh, here I find a possible problem. These encoders will give only 400
'ticks'
per revolution, which may be a bit coarse. I'm using 1000 cycle/rev
encoders,
that give 4000 counts/rev. With a 5 TPI screw, that is 20,000
counts/inch, or
.00005" per count.

Your unit has 800 count/rev (of the ballscrew) x 10 TPI = 8000 counts/inch.
Not really low, but without tach feedback, thing may move in small jumps.
Your belt needs to be REAL tight, to prevent it becoming a 'spring' between
the motor and the leadscrew.

> I don't have a torque spec on the motors and can't find it anywhere
> on the Leeson sight, but here is what I do know : They are labelled
> 1/8HP, 24VDC, 5.7A, 3300RPM motors. They have 3/8" shafts and are
> labelled model # cm31d33nz1b

Well, we can derive some info from this. 24 V x 5.7 A = 137 W, which is
quite consistent with the 1/8 Hp. Now, that is probably a continuous
rating,
but we don't have a peak rating. I have shunt-wound field motors, so I
don't
have to worry about wrecking the permanent magnets. My motors are rated
at 75 V 1.54 A and also 1/8 Hp, but at 1725 RPM, which is a lot better for
this application. I have my current limit set for about 6 Amps, which
gives
a very roughly calculated linear force at the table of 1000 Lbs. My screws
are ballscrews, and very low friction, and I have a 2.5:1 belt reduction.
Acme screws have vastly higher friction, of course.

> I know that I am running 24VDC motors at 36VDC, but I figured for my
> likely duty cycle it would be ok.

Yes, quite right.

> My real worry is the mystery torque amount of the motor. I had
> assumed based on the size of it and my gut feeling based on the
> available specs that I would be ok. Now I'm a bit worried. I will
> try to loosen up my gibs a bit and see if things improve, but this is
> not the ideal long term case for me.
>
> I supposed the next solution would be to figure a way to gear them
> down further . This would be quite difficult as they are allready at
> 2-1. 3-1 might just be possible with a really thin pulley wall.

Since you have a motor designed for higher speed, that sounds like a good
first thing to try. If you have the current limit set for 5.7 Amps, you
might try
turning it up a bit. Undoubtedly, the peak rating is going to be well
above the
continuous rating if this is a true servo motor.

Jon