Optimizing Precision in Stepper Motor Driven Gear Systems

[Justin Boga]

I am driving 2 different gears with a belt off of a stepper motor that has a step angle of 0.018 degrees. For the purpose of my project, I need to have high precision between the stepper motor and the gears that it's driving. Is it necessary to have 20,000 teeth (360/0.018) on the gear that's being driven in order to keep the same size steps from the motor to the gear? I am an electrical engineering student so I have no experience with gears so I'm not sure if a step will be lost if there aren't enough teeth to catch the small movements. I went with a belt instead of a gear train because I feel like that would preserve the precision better and to keep the mechanical portion relatively simple but if not, feel free to offer any suggestions.

 

[DaveC426913]

I know little about gearing and stepper motors, but it seems to me, you'd get better precision if you used a gearbox to step down from a higher rev stepper motor. Your steps on your motor could be large, more easily-controlled fractions of a turn, resulting in better control of small outputs. I think that's how precision timepieces work.

 

[Baluncore]

The shape of the gear teeth is selected to give rolling contact without any harmonic evidence of the number of teeth on the gear wheels in the box. For correctly cut gears, a steady input shaft speed will result in a steady output shaft speed. If that was not the case, vehicle gearboxes would shake the vehicle apart.

The problem you may have to face is backlash in the gearbox. A flat belt or Vee belt may slip but will absorb some of the step noise from the motor. A stepped belt will have a problem with elasticity and backlash, depending on the shape of the belt teeth.

You have not identified your application so we do not know what shaft ratio you require, or what backlash is tolerable. There are ways to reduce backlash.

 

[Nidum]

Bit of reality needed .

A stepper motor is not that precise in positioning in the first place . Plus or minus half a step positioning is all you can be reasonably certain of without using feedback .
Stepper motors can also quite easily go completely out of sync with the incoming step pulses .

Common stepper motors only have small numbers of steps/rev - typically 200 . That 200 steps/rev gives 1.8 degrees/step . Even with exotic stepper motors and microstepping drives it is difficult to get more than 2000 steps/rev or 0.18 deg/step .

In any case microstepping drives do not improve fundamental positioning accuracy .

Tell us what you actually want to do .

 

[Justin Boga]

To be more specific, the stepper motor actually has a planetary gearbox that drops the motor down to 0.018 degree step angles from 1.8 degrees. I am designing a novel optical spectrum analyzer that relates the angle to a specific wavelength so my resolution would essentially come down to the precision and repeatability of the stepper motor. Since the reflection gives us 2 theta, I need to rotate one object at half the speed of the other. A full explanation of how the project works would be too complicated for this thread but I hope I gave enough information for at least the mechanical aspect. Thank you for the responses and I'm sorry for the ambiguity of the original post, this is my first thread for an engineering topic.

 

[Baluncore]

The obvious way to rotate components at ω and ω/2 is to use two gears with a 2:1 gear ratio. The stepper motor then drives the outer edge of the larger gear through a small pinion which gives the best reduction ratio. Without an idler gear, the distance between the two objects will be fixed by the gear radii.

Backlash between the main drive gears may be a problem. By using two smaller gears on the same shaft, one spring loaded against the other, that can be eliminated.

To eliminate hysteresis and get repeatability from the stepper and planetary gears you may need to approach the destination angle or always sweep in the same direction.

 

[Nidum]

Gear systems don't magically conjure up accuracy where there is none to begin with . Generally they make things worse .

A better way to achieve the level of accuracy needed would be to use a servo drive and to monitor the actual position of the object being moved .

 

[Baluncore]

A better way to achieve the level of accuracy needed would be to use a servo drive and to monitor the actual position of the object being moved .

Optical encoders can give you 2000 cycles per turn, with quadrature signals that works out at about 0.05° resolution.

At zero current, the servo torque is zero and the noise of a single servo motor is greatest. You may therefore need two servo motors in torque opposition on the one axis. The motors will have a small fixed current difference so there is no hysteresis in the drive. The motors will still work together when accelerating or braking. A component such as a mirror or grating can be mounted between two such motors, working together in slight opposition.

 

[Nidum]

http://www.renishaw.com/en/resolute-absolute-encoder-system-with-rexa-rotary-angle-ring--10852

 

[Rx7man]

Baluncore has a good point.. spring load the mechanism to remove backlash, perhaps use a worm drive gear at 50:1, that would give you a pretty accurate position. The hardest part will be to zero it consistently.