[LUCKY13]

I understand how a stepper motors has holding torque, but a servo I don't know how its supplied voltage, I guess.



I mean you have a plus & negative wire, and you reverse the polarity to change direction. But how is the Gecko driver supply voltage to these motors. I assume its some type of pulse modulation . WHen the servo is stationary is the driver pulsing the voltage, and switching polarity back and forth to keep the motor holding ?


I really dont understand how a servo can get its steps either. I understand its being read off the encoder which tells the controller where its at, then it knows where it need to take the motor . But how does it make small accurate steps, or maybe a better question would be how does the controler, control the motor to make a accurate step ?


I guess ever how it creates holding torque it varies this action to make steps but it looks like it would over/under shoot its target very easy. I have read many docs and post and I dont seem to be getting it. Some type of modulation would seem to be in order but its just not adding up for me. I must be missing something basic. ALthough I dont know much about motors unless where talking about a Hemi Mopar engine ( I know thats not a motor ). Through my research I have come to understand the steppers quite well, but I'm still scratching my head on the servo's. I can understand lets go this way, lets go that way, but lets move a little and stop, exactly right here, I'm lost ?



Jess

[Geof]

.... I can understand lets go this way, lets go that way, but lets move a little and stop, exactly right here, I'm lost ?
Jess

No, I don't think you are lost you have it more or less correct. What you need to do is put a value on 'little' and 'exactly'. If the servo is jiggling back and forth 'a little' when little is something like +/-0.00005" then it is 'exactly' right there on a machine with a precision of +/-0.0001".

[Al_The_Man]

The difference between steppers, open loop, and servo's is the PID loop.
Here is a previous post, the divshare link is probably the most relevant.
http://www.cnczone.com/forums/showthread.php?t=68762
Search for PID and Transconductance amplifier should get you all the info you need.
Al.

[LUCKY13]

No, I don't think you are lost you have it more or less correct. What you need to do is put a value on 'little' and 'exactly'. If the servo is jiggling back and forth 'a little' when little is something like +/-0.00005" then it is 'exactly' right there on a machine with a precision of +/-0.0001".

Thanks for the replies.

And Al, I read the two papers at the bottom of that thread/link, but the main link you listed I have to sign up for to be able to down load it. It looks like it would be a pretty complete guide and I will go back latter and do so.



Geof, so your saying the servo will be going back and forth in that area to hold its position? So the controller is modulating the voltage from side to side to hold it there. I imagine this is why some guys have a bad hum in the motors when stationary, to much backlash, improper tuning (gain/dampening), wrong size motors, gearing, shafts ext,ext......

I know I have a lot of research to do, and maybe I will get to the point that the Ole light bulb will light up ( not to long ago a didn't know how the steppers work) but I just don't get how the servo systems can be accurate. I know they can, and with true closed loop they can be very accurate, but I don't understand how.

I really am no good with math, or electronics for that matter (big understatement) but I will keep reading until I get somewhere with it. With what I do understand (or at least think I understand) in one way it would seems like with servos the big the better (to a point). But looking at another way it seems they would need a load on them to stay smooth.



Yea, there is still a lot I don't understand, maybe after I sign up to the
place Al listed I will learn something from that download.

One thing I feel like I did learn from one of the links, I want to gear my belt drives as low as I can and still maintain rapids, this should help keep the motors in there effective range and even help eliminate resonance. Of course if I was going to run Mach for a controller that could be a big gotcha real fast ( I think) but I bought the CNC Brain and I should be able to run a fairly low gearing ratio. I am also going to run the selectable count encoders that have been mentioned by a few here lately.


http://search.digikey.com/scripts/Dk...me=102-1307-ND


One question I do have about the servo ratings.

They have a "max operating speed " rating (= RPM), is this a rating that says they will turn this RPM under load (with proper voltage), or that you will cause damage if you exceed this speed ?

Jess

[guru_florida]

Not sure if you found your answers or not...but to add a little real world examples of servo type closed loops:

1. In a car, you speed up to a desired cruising speed and set your cruise control. The cruise control circuit then applies gas (correction) when the needle gets below the cruise setting. The more off the mark it is the more it applies gas (the amount of error). In this situation there is a speed sensor (feedback), user setting (command) and gas pedal actuator (power house/plant). A cruise control circuit is almost identical to a servo motor controller. However, in most cars they don't include a brake actuator and therefor the car will go faster downhill without correction.

2. On a boat, you have a gps. You set the desired heading and hit lock. When the boat heading begins to go of course, the rudder actuator engages in opposition to correct the heading. Another case of a simple feedback loop.

3. Your heating/cooling panel in your house. Another feedback controller to keep your house temperature at a desired level.

So how does a motor stay in one spot resisting change? Same as the car's cruise control. Each action is opposed with an equal and opposite reaction. Even on a micro scale, a small rotational load applied to the shaft, the motor compensates in small amounts in an equal amount. And it does it between 2 and 40 thousand times per second (depending on the motor controller.) In any case, with a good controller, it does it so fast we cannot feel the effect. However, on an oscilloscope we can see.

All these systems are very much in common in their diagrams. Only some of their implementation changes. In most cases a simple P controller, or Proportional controller works. The amount of opposite correction is proportional to the error.

In more advanced systems like motor control. The PID algorithm is used. Proportional Integral Derivative. It sounds complicated but it's actually very simple. Sometimes the proportional value isn't quite large enough to correct for small errors. For example, motors require some power just to get rolling. Add an integrator into your design and it will accumulate these small errors over time and provider a larger signal to the motor to kick the motor into correcting the minor error and reach the exact position.

The Derivative component essentially takes the last position and the current position and calculates the velocity of the error. This is also equivalent to the slope of the curve if you graphed the error. It can be used to dampen the correction signal so the system doesn't overshoot it's target because the motor became saturated.

The P, I and D components are calculated independently each iteration of the control loop based on the measured error value. The outputs of each of these components are then scaled using coefficient constants and then these values are summed to produce a single value that feeds the amplifiers to the motor or whatever the corrective actuator/plant happens to be.


The coefficients are the parameters to the PID system and will depend on the motor/plant and mechanics and load in and on the system. These coefficients are determined manually by a human by tuning or in advanced systems automatically (very advanced adaptable systems!). Tuning of the coefficients is very important in non-linear systems like machining. Even gravity can throw a wrench into the tuning if the mass is on an extended arm and thus the power to move one way is much different the the power requirement for the opposite direction. (Imagine this, you would require two sets of coefficients depending on the direction of compensation!) Most commercial motor controllers use much more complicated algorithms than *just* PID. There are also other algorithms as well.