What's new
What's new

Gear reduction for NEMA motor accuracy

teachme

Aluminum
Joined
Apr 13, 2020
Im making a round rotary table, about 300 mm in diameter -- looking to get a tolerance within ±0.01 mm around the perimeter

With step angles of 1.8° Im looking at tolerance of ~ 5 mm.

Question: What gear reduction would I have to use, to get ±0.01 mm?
 
Im making a round rotary table, about 300 mm in diameter -- looking to get a tolerance within ±0.01 mm around the perimeter

With step angles of 1.8° Im looking at tolerance of ~ 5 mm.

Question: What gear reduction would I have to use, to get ±0.01 mm?

When you say tolerance of .01mm, do you mean you want to be able to position any point on the table perimeter within .01mm? In any position? Or just a few?
 
How much load? Must it turn continuously? If loading is light any reasonably consistent gear reduction can be used, such as a worm gear, with a preload mechanism. Preload mechanisms are easier if it doesn't need continuous rotation.

Gear cogging can be programmed out if the gears are cut well.
 
How much load? Must it turn continuously? If loading is light any reasonably consistent gear reduction can be used, such as a worm gear, with a preload mechanism. Preload mechanisms are easier if it doesn't need continuous rotation.

Gear cogging can be programmed out if the gears are cut well.

Load max = 2-3 lbs (0.9 - 1.3 kg)

It will pause for 10 seconds every 25.7° (where the tight tolerance is necessary)
 
Load max = 2-3 lbs (0.9 - 1.3 kg)

It will pause for 10 seconds every 25.7° (where the tight tolerance is necessary)

So you want to index 14 positions? There are better ways to do this.

Just roughly position the table, than have a wedge shaped pneumatic clamp align it and lock it in place.
 
If you're indexing the same index for every station, a shot pin solution is far and away the best.

If you need arbitrary positioning:

For that size, I've gotten good results machining a timing belt profile directly into the perimeter of my plate. HTD belts have a fully round tooth profile that's straightforward to machine. 20:1 with a 300mm plate means a 15mm drive sprocket. With a 3mm belt, you'd be looking at about a 16 tooth pulley on the motor.

20:1 is about the most you can hope to do with a single stage of reduction before things get silly.

For .01mm resolution at 300mm diameter, you'd need a servo with a 5000 PPR encoder. That's not hard to come by. If you need to position against a dynamic load, everything gets harder, of course.

If you're wedded to a stepper and can't use a shot pin, you're either looking at precision worm drive gearboxes or a harmonic gearbox, and about 200:1 gearing. Worm drive is okay if you don't need to reverse and you don't need to go particularly fast.
 
If you're indexing the same index for every station, a shot pin solution is far and away the best.

If you need arbitrary positioning:

For that size, I've gotten good results machining a timing belt profile directly into the perimeter of my plate. HTD belts have a fully round tooth profile that's straightforward to machine. 20:1 with a 300mm plate means a 15mm drive sprocket. With a 3mm belt, you'd be looking at about a 16 tooth pulley on the motor.

20:1 is about the most you can hope to do with a single stage of reduction before things get silly.

For .01mm resolution at 300mm diameter, you'd need a servo with a 5000 PPR encoder. That's not hard to come by. If you need to position against a dynamic load, everything gets harder, of course.

If you're wedded to a stepper and can't use a shot pin, you're either looking at precision worm drive gearboxes or a harmonic gearbox, and about 200:1 gearing. Worm drive is okay if you don't need to reverse and you don't need to go particularly fast.

Very helpful I appreciate it.

I am limited to a stepper, and a worm drive gear box sounds like the way to go.
Could I ask what math you used to get a 200:1 gear ratio?
 
Sure!

You want a positional accuracy of .01mm at the rim of the disc.

With a diameter of 300mm, the circumference is 300mm*pi. So about 942mm. Divide that by your accuracy requirement, and you get 94,248 positions.

A standard stepper is 200 positions per rotation. Microstepping that will reliably get you within half to a quarter of a step. So a 200:1 gearbox will get you about 80,000 to 160,000 reliable positions.

It's not an even multiple, so you're stuck microstepping either way.

Alternatively, size your table so that the circumference is an even multiple of the step angle and gearbox ratio.

For example, with a 200:1 gearbox, half-stepping with a 200 step motor, you'd go backwards:

(200 steps (motor) *2 (half step) *200 (gearbox ratio) *.01 (accuracy))/pi (circumference to diameter) = 254.65mm diameter

Or with a 250:1 gearbox
(200 steps (motor) *2 (half step) *250 (gearbox ratio) *.01 (accuracy))/pi (circumference to diameter) = 318.31mm diameter

But, realize with a worm gear box you'll have to either always approach from the same direction, or add a preload mechanism, because worm boxes have backlash.

Hope that helps. But again, if you don't need arbitrary positions, just a fixed 14, then just drive the wheel directly off the stepper, add a tapered shot pin on a preloaded linear rail driven by a pneumatic cylinder and call it a day.

Also, realize that this is just the resolution of the drive system, with no contribution from other things. For example, it's assuming your machining is perfect. If you're mounted out of round, there's nothing the drive system can do to fix that, assuming you have work being done at each station. Mounting things +- 10 microns isn't that easy. Thermal growth on your 300mm plate will take up your entire 10 micron budget with just a 3 degree C change in ambient temperature, unless your whole mechanism grows and shrinks in unison. But that's in the radial direction, so it doesn't necessarily affect the drive system.
 
Sure!
You want a positional accuracy of .01mm at the rim of the disc.
With a diameter of 300mm, the circumference is 300mm*pi. So about 942mm. Divide that by your accuracy requirement, and you get 94,248 positions.
A standard stepper is 200 positions per rotation. Microstepping that will reliably get you within half to a quarter of a step. So a 200:1 gearbox will get you about 80,000 to 160,000 reliable positions.
It's not an even multiple, so you're stuck microstepping either way.
Alternatively, size your table so that the circumference is an even multiple of the step angle and gearbox ratio.
For example, with a 200:1 gearbox, half-stepping with a 200 step motor, you'd go backwards:
(200 steps (motor) *2 (half step) *200 (gearbox ratio) *.01 (accuracy))/pi (circumference to diameter) = 254.65mm diameter
Or with a 250:1 gearbox
(200 steps (motor) *2 (half step) *250 (gearbox ratio) *.01 (accuracy))/pi (circumference to diameter) = 318.31mm diameter
But, realize with a worm gear box you'll have to either always approach from the same direction, or add a preload mechanism, because worm boxes have backlash.
Hope that helps. But again, if you don't need arbitrary positions, just a fixed 14, then just drive the wheel directly off the stepper, add a tapered shot pin on a preloaded linear rail driven by a pneumatic cylinder and call it a day.

Plus one for shot pin.
One should be able to microstep to within 1/10 step with a decent drive easily but remember that the poles in a stepper motor are not perfectly spaced so a correction table mkay be needed for counts to position.
Worm gears have cyclical errors per turn of the worm and errors in the wheel also needing correction at the level your are speaking not to mention that they have to have some lash or do not move.

It is the added errors of everything in the system that counts so this is finer than the basic math works out to when building rotaries.
HTD belt, errors, worm drive other types of errors, spur gearbox or head it's own errors, DDR it's own set of problems. Then we have to talk about the bearings the rotary is using. Who would have thought this matters

All these gear down mean lower index speeds if that of any concern.
Cost and application is always the key. I've been down this rabbit hole.
I'd say low cost commodity rotary table belt driven with a servo or stepper but not sure such will hit your number.
Can you program out the errors or is this go to where I want?
Stiffness of a shot pin is so way above just a motor but those notches and pin have to be on and slop in the pin assembly. Oh my, shot pins have a error also since they need clearance and the ring.
Everyone wants to not tell what I am doing to the world but without it all are all farting in the wind.

The math is easy and straightforward. Why the math not working out is chips in your shoes and most certainly I have made mistakes here.
I see a spec of xxx, no idea of the what or why or axis loading.

What am I doing with this axis and why the number?
Bob
 
a sherline rotary table is 70::1. If you put an encoder running the perimeter of your 300mm circle you can run a check count after a 'rough positioning'. Moving the encoder out to edge lets you check real position and should compensate for most drive train errors.
you could also run the motor on the perimeter (pinon gear design) of shop made turntable. If the motor mount is spring loaded you can keep backlash down.
 








 
Back
Top