servos vs steppers: resolution

Not a simple subject! As an example, the head positioning servos in later technology hard drives use a voice-coil actuator and digital signal processing. Drives once used stepper motors. You can get wonderful resolution with either ... generally, the choice is driven by economics.

What's that Lassie? You say that Mark Haase fell down the old sci.electronics.misc mine and will die if we don't mount a rescue by Wed, 09 Jun 2004 23:56:52 -0400:

The stepper's res. is based on the steps.

The servo's res. will be based on the encoder attached to it, and the servo amplifier's error.

In the stepper system, the driver advances one step, and the stepper motor follows.

In a servo system the encoder gives the motors position to the servo amplifier and it compares this with the desired position to get the error. The amplifier then sends current to the servo motor to make the motor move into the proper position, reducing the error.

I've gotten ~426 discrete movements in the ~200 deg turn range of a Tower Hobbies RC servo. This was using the Kronos Robotics ezservo chip. With stepper motors you can use reduction gearing to get about any degree of accuracy. You may want to check the comp.robotics.misc news group for more info.

That's a servo, not a stepper. While the operating concept is similar, in that they're both able to position an object ot a given orientation, the mechanism of the two is entirely different, and has distinct limitations on the accuracy available when using each type.

Only somewhat true. No matter how you gear it, a stepper motor still moves in discrete... (wait for it...) steps. Each step covers a specific range of "swing". For example, assuming zero positioning error (which isn't "real world", but neither is this discussion), a 1.8 deg. stepper will turn a full circle in 200 steps - 0.0 deg. 1.8 deg. 3.6 deg. 5.4 deg. 7.2 deg., etc, etc, etc, around to 360.0 degrees, where it started. Which means that a device attached directly to the shaft can never be successfully pointed at "4.6 degrees". Even with gearing, there are going to be numbers of degrees which are simply impossible to turn the mechanism to. Exactly what those numbers are will vary depending on the resolution of the stepper, and how the gear-train is designed.

In a nutshell, a stepper (with or without gear-train) is a set of "preset" positions you can move to. Any positon that's not part of the "presets" is unattainable by that motor or motor-and-gear-train combination, and can only be reached as an approximation. Granted, that approximation may be "close enough" for a given purpose, but when you get right down to brass tacks, 4.6 degrees isn't the same as 5.0 degrees. if you don't believe that, try pointing a telescope at an object that's listed as being at 4.6 degrees above the horizon with a stepper that can only do 3.6 or 5.4 degrees... Let me know how well your observation of the object turns out :) On second thought, don't - I'll tell you how it will turn out: You'll never lay eyes on the object.

A servo is a motor that can be stopped anywhere you want it, with no "detents" either needed or present. You can turn it to any position youlike (within its range, of course), and assuming it's been properly "dialed in", it's reasonable to expect that when you say "turn to 4.6 degrees" and punch the "go" button, it's going to turn whatever it controls to point at a reasonable approximation of 4.6 degrees, rather than having to call 3.6 or 5.4 degrees "good enough".

So to answer th OP's question, Servos have (theoretically - there are all but certainly going to be practical limits) infinite resolution.

Steppers have distinct "steps" of resolution, and if your desired orientation doesn't coincide with one of the steps the motor or motoer/gear-train combination is set up to handle, you're never going to get the device pointed to that orientation.

Short form: I'm trashing EVERY E-mail that doesn't contain a password in the subject unless it comes from a "whitelisted" (pre-approved by me) address. See for full details.

Hi Don,

I like your "whitelist" technique. I've been considering something similar, except I've been trying to develop an automated daily password change ;-)

...Jim Thompson

When I first started playing with the concept, I was also going to go with a rotating "password", but I've found that just one is all that's really needed. Why make it harder than it has to be? :)

Should the spambots actually catch on (which I think unlikely, but don't consider to be entirely impossible) it only takes me two quick edits to change the password - total time? Counting "log into the shell and fire up an editor", probably under a minute for the whole operation, unless I've got an INCREDIBLY slow connection when I go to do it.

Of course, I'm doing my filtering at ISP level, so that the garbage never gets transferred to my machine - It's flushed LONG before my "at home" filters ever get a crack at it. Why bother to move it around when I already know it's spam?

Don't forget microstepping! You can get the stepper motor to come to rest BETWEEN two steps by turning on both coils half way and so on to

There is a rather complete discussion of this at

: > In the stepper system, the driver advances one step, and the stepper : > motor follows. : >

: : Don't forget microstepping! You can get the stepper motor to come to : rest BETWEEN two steps by turning on both coils half way and so on to : 16 or more divisions between each actual step. : : There is a rather complete discussion of this at :

It is possible to achieve micro steps in the order of 10 increments between the native increments. I have used Microsteppers with 2000 steps per revolution (200 * 10) They are incredibly smooth!

============================= So why stop at 16 microsteps per step? A pair of 8 bit dacs putting outsin and cos to the coils would give 256 usteps per step

Very true! And I have seen just such systems. Now, there are limitations: The divisions generally turn out to be less than totally linear. For example, with the two coils each half on, you are going to get a position that is just about half way between the two steps. But when you go for 1/4 and 3/4 coil drive, you find that the position isn't quite 1/4 of the way from one step to the other. And there are lots of factors which effect the error... Or at least I've never seen any way to get a consistant error correction across different motors, etc...

The linistepper

Some of the top end chopper controllers support 256 microsteps and even have error correction systems that are said to make 256 * 200 =