precision, gear reduction, stepper

fred_b

Member
For my next project, I plan to build a camera panning head that is driven by a stepper motor. I would like to incorporate gear reduction in order to be able to turn the camera in very small precise increments, (around .01 degrees or smaller).

When reading about gear motors and the like, I see that backlash of around 2 degrees is common. Is this on the output shaft or on the input?

Can anyone recommend a source for the hardware? I was thinking maybe a stepper with a screw drive gear to turn a larger gear? Can something like this be bought as a unit?

Thank you
 

westaust55

Moderator
have a read at this in Wiki on backlash in gears:
http://en.wikipedia.org/wiki/Backlash_(engineering)

I would have thought one could do better then 2 degrees of back lash.



The more gears in the gearbox the more backlash you may encounter.

I picked up a 3 volt (2.3Amp / winding) 200 step per rev (1.8deg steps) Bipolar stepper motor second hand (ex a printer) for a few dollars which I am about to play with.

Half stepping can effectively give 400 steps per rev or 0.9 deg/step.

Do some research on micro-stepping - this does need a more complex motor driver circuit.

Getting down to 0.01 deg steps with just a stepper motor ?

But a stepper motor only approach would overcome gearbox backlash

What torque do you require for you mechanisms?
More information may give folks a better idea on what will achieve you goals
 

premelec

Senior Member
Spring load

You might try a simple spring load on the gear so that there is always pressure against one side of the gear.... There is also a method of using two identical gears next to each other with a spring loading them to engage both sides of the driving gear - see anti backlash methods.. you don't say what angle you want to turn or if 360+ - Telescope mounts deal with the problem as have many devices so we know it's do-able! :)
 

dennis

Member
If you need a smooth motion then you may need to incorporate some damping as steppers are hardly smooth. Maybe a dc motor and encoder would be better.
 

westaust55

Moderator
If you need a smooth motion then you may need to incorporate some damping as steppers are hardly smooth. Maybe a dc motor and encoder would be better.
This website suggests that even with micro stepping at very small increments that likely best accuracy is 1/10th normal step and that resonance/fluctuations can be expected at low velocities.

http://www.euclidres.com/apps/stepper_motor/stepper.html


Even AC motors with VVVF type speed controllers will typically "cog"/step at very low speeds.

The DC motor with a gearbox and maybe no speed controller to avoid introducing cogging/pulsing would give the smoothest motion if smooth travel is critical.
 

Andrew Cowan

Senior Member
Using a torsion spring on the output shaft can (reduce) backlash as it keeps the gears tightly meshed in one direction.

A
 

BeanieBots

Moderator
Precision robotics frequently use belts and/or steal wire on drums instead of gears. When higer powers need to be transfered, then large linear motors are used with precision single stage gears. These ARE very expensive as you have already discovered. As with electronics, with mechanics, the rule 'you get what you pay for' is even more relevant.

For lower torques, which may well apply to your application, have a look how it's done in mechanical radio tuners. These use the methods already mentioned such as dual cogs, cord & pulleys and sprung loaded gear trains.
 

MFB

Senior Member
Adding dither

A technique used by the designers of German W2 guided missiles was to introduce a small amount of constant dither to overcome the slack in mass-produced control surfaces. Rather than just letting the output from the gearbox sit in one of two states, you continually drive between them in order to achieve a mean output position. I have found this to work well on robots but it can get a bit noisy on larger models.
 

premelec

Senior Member
Dither can help if you have an accurate angle sense unit on the final shaft output - it helps with any dead band situation and can be used with A/D converters to get another bit of accuracy for instance... the thought of poking at things to see if they are stuck is kind of cool... :)
 

Jeremy Leach

Senior Member
Don't want to dampen enthusiasm, but IMHO ...

.01 degrees = 36,000 steps per rev - I'd say that's a VERY difficult thing to achieve accurately - so any commercial unit will be expensive because of the exceptional engineering tolerances etc.

Plus you've got to think about the motion you require and how you're going to achieve it:

Work out the maximum required turn rate in degrees per second then this will give you the maximum steps per second. Will the microcontroller/ stepper be able to step this quickly, especially if your code is going to handle acceleration and deceleration.

Or perhaps the absolute positional accuracy isn't so important as a 'smooth' pan? And maybe this could be achieved in a different way without steppers?
 
Last edited:

Rickharris

Senior Member
If you can go to a linear drive srom the stepper then a threaded rod with 2 nuts on it separated by a spring (can be a lump of rubber) will provide a backlash free output to a light load.

Other than that - although I have no idea how they work - I have some old tuner extensions from old radios they go between the tuner knob and the variable capacitor and provide a low speed backlash free drive.
 

fred_b

Member
Thanks for all of the replies.

It does look to be an expensive project but I think it could work:

1.8 degree Stepper with a gear reduction of at least 60, using something like this:

http://www.automationdirect.com/adc/Shopping/Catalog/Power_Transmission_%28Mechanical%29/Worm_Gearboxes_%28Speed_Reducers%29_by_frame_size/1.75_inch_frame/WG-175-060-R

That would give a rotation increment of :360 / 60 / 1.8 equals .03 degrees.

The camera is a still camera that would take photos when the motor is not turning.

I will look into this further but it looks like it has been done with steppers before:

http://www.bmumford.com/photo/rotary/index.html
 
Top