View Full Version : Timing Belt dividing Head

03-22-2007, 04:31 PM
I was musing today as I was looking at buying some small timing belts could these be used to make a dividing head. If a stepper motor was used with a 1.8 degree step and a 10 tooth pulley and a 100 or more on the dividing head then each step would be 0.18 degree which to me seems enough.
Only problem I see is that the motor would be 90 degrees off line.
Still a good basic idea and would be quite accurate with very little backlash!


03-22-2007, 05:16 PM
My designer catalog states that for MXL, XL, L etc. a minimum of 6 tooth must be in contact with the pulley, to equal the strength of the belt. More tooth in contact reducts backlash. I think this will also apply to other types of belts.

I have also thought about this. Maybe a staged approach 1/4 * 1/4 = 1/16 would make life easier.

03-22-2007, 07:33 PM
A two stage reduction might be a good idea for that large a ratio.

On the other hand, the de-rating of the strength may not be a material problem either. Looking at an HTD3 belt (3mm pitch), the loading capacity is 400lbf/in and the HTD5, 5mm pitch belt is 742lbf. The narrowest standard HTD5 belt is 9mm so it's rated for 263 lbs. Sacrificing, say 40% would put you down in the 158 lb. range which still seems like it might be enough for a dividing head.

03-22-2007, 08:10 PM
I'm not sure what you are planning on using to drive the stepper but most stepper drives can use micro stepping. So you could probably figure on having 8 to 10 times finer resolution than single step control.


John Stevenson
03-22-2007, 08:50 PM
Sorry but it won't work.
The detent torque from a stepper running thru just a belt drive isn't enough to hold against cutting forces.

My first one was built this way using a large 1200 oz/in stepper running at 4:1 reduction and the first job, a keyway, you could see the spindle notching over as it was cutting and finished up with a slow spiral.

This is why high ratio worm and wheels are used because they cant run back.


Paul Alciatore
03-23-2007, 01:48 AM
I'm not sure what you are planning on using to drive the stepper but most stepper drives can use micro stepping. So you could probably figure on having 8 to 10 times finer resolution than single step control.


Micro stepping is not a total blessing. First, the microsteps are not linear unless you use a controller that can compensate on a micro step by microstep basis. And even then, the compensation will only be good at a single torque value. Also, you loose a lot of holding torque when trying to hold at a microstep. It can be only 10% or even less than the full step holding torque and depends on several factors. So you would need a much larger motor as opposed to using full steps.

As for the original idea of using a 10:1 reduction producing a 0.18 degree per step, how adequete that is may depend on the diameter of the work you plan to do with it (the radius to the holes or gear teeth or whatever feature you are machining from the center of rotation). At a 1" radius, an error of half a step or 0.09 degrees would be a linear error of about 0.0015". This error will increase with distance from the center of rotation in a linear manner so at 2" it will be 0.003" and at 4" it will be 0.006". This does not sound like it is good enough for gears, but it may surfice for drilling hole circles.

And it is hard to get more than a 6:1 reduction with commercially available pulleys.

As for holding against the cutting forces, you could add a lock, just like the worm gear models have. I always engage the lock before cutting. If you are cutting circular slots or spirals, you would need a larger motor and bigger belts and pulleys. Probably not practical.

But why? With some searching, a worm and worm wheel can be purchased for less than a set of pulleys and a belt.

I have also thought about such things.

03-23-2007, 02:08 AM
There are other ways to lock something than a purely mechanical lock or relying on the holding torque of the stepper. This sounds like an interesting idea and I have been thinking about it off and on today. Holding force was the first problem that occurred to me. I use a magnetic clutch on my electric cross feed drive on my South Bend to engage/disengage the feed. It has a holding torque of at least 5 foot lbs or about 1000 oz/in.

It's the round black object at top left:


If the drive belt were partly wrapped around an appropriate pulley on the clutch output shaft engaging the clutch would lock it from moving if the clutch input were permanently held from rotating. That would solve the holding problem and would be easy to implement in G code using a clamp code. The clutch has zero backlash when locked and is bidirectional.

Clutches such as these are commonly used in photocopiers of the larger variety.

03-23-2007, 04:38 AM
On the torque point surely the motor torque would be multiplied by the bearing ratio. So the only problem as I see it is that the belts do have an amount of stretch / resilience built into them as they are in the main designed for continous one way driving.
The idea was for a simple device for the production of gears or flats on shafts in which I would think the rotation stress due to milling would be reduced.


John Stevenson
03-23-2007, 05:32 AM
No the belts are fine as there is virtually no backlash stretch etc.
The problem is holding torque and if you start micro stepping it's reduced even further. Steppers are rated in oz / in of holding torque but no one ever says what torque a cutter is putting out on it's cutting edge.
Given that the spindle is being driven by a motor roughly 10 times the size of the stepper may give a hint.

Using locks / clutches etc is just an half assed way to rectify a fault that shouldn't be there.
If you are indexing then a lock will work but what happens if you want to do rotary machining? Do you have a belt driven head with lock for indexing and a worm driven head for free machining ?

Might as well just use the worm head.

As I say I went down this route on a large head built on a lathe spindle from a scrap 12" lathe, big 1200oz, motor 70 volt drive and 1" wide L section timing belts, certainly no wimpy build up but a simple 3/8" keyway proved I was going to have more problems in the future.
A pity really as it was a neat design but as non are made like this in indusry it should have put the flags up.


03-23-2007, 07:12 AM
Well I think that just about kills that idea and as you rightly say if it aint done that way in industry then it is usually a no no!!

I really must stop musing and get some work done but then again nothing gained by not experimenting!!


Paul Alciatore
03-23-2007, 11:50 PM
I can't speak for the others, but I was just discussing the problems, not trying to put an end to the idea. I do think that something COULD be done, but you do need to keep all the problems in mind while doing it or it will be a disaster.

If limited accuracy of small diameter work is all that is needed, then the stepper and a modest step down ratio could work. Just don't consider it for cutting gears. I also believe a brake could be added to overcome the holding problem. In an automation system, a solenoid operated brake could easily be added to the system with some additional software to drive it. But then, no circular or spiral milling. It just depends on what you want.

Another approach that may overcome the holding torque thing, eliminate the need for a brake, AND improve the accuracy is to use a double or even triple gear down for a higher ratio. Consider a 200 step per revolution stepper and three sets of timing pulleys, perhaps 3:1, 4:1, and 5:1. I used three different prime ratios (well the 4 is actually two twos) to give as many solutions that come out in a whole number of steps as possible. 200 X 3 X 4 X 5 = 12,000 steps. That's about 2 minutes per step. At a 1" distance from the center, one step would be about 5 tenths of movement. At 2" it is a thousanth etc. Not as good as my 10" RT, but getting better.

If you can find the pulleys, you could change the 4:1 to a 7:1, thus making the steps even finer and adding another prime number (7) to the mix. 200 X 3 X 5 X 7 = 21,000 and we are at about 1 minute per step or about twice as accurate as the first example.

And I doubt that you would be able to back turn either of these designs with a crowbar. But you may need heaver belt sizes. No, not may, but will.

03-24-2007, 12:08 AM
FWIW, has anybody else noticed the newest programmable indexing heads from Hardinge use servo motors with a hole down through the armature as an indexer. Basically just a motor.

These are suited for indexing and rotary milling.

03-24-2007, 06:42 AM
I suspect it could be made to work, but as with many things if there's a cheaper, easily available, proven and widely used solution there's usually a very good reason that this is the case.
Off the shelf worms & wheels are available at prices which make the multiple pulleys & belts option look less than attractive and overly complex.
Don't you guys in the states have somewhere like


They do pulleys & belts too!

John Stevenson
03-24-2007, 07:00 AM
Servo's are different animals to steppers.
Once a stepper is commanded to move then it does but if it gets knocked off position it doesn't know, often known as open loop.
Servo's OTOH, NEED to know where they are at all times, servo's are basically DC motors that are told to hold a position and to do this they have an encoder that relays any movement back to the drive and if it has moved, or not moved enough it's told where to go, often known as closed loop.

Many of the commercial servo's as mentioned by DR above are rated from 170 volts at 50 odd peak amps up to three phase servo's running at 600 volts per phase at 30 to 50 amps, so we are talking serious torque and not to be confused with home shop steppers.

To get back to Pauls reply of multiple drive belts going to 7:1 ratio, even these will wind back under power.
Bottom gear on a truck is anywhere from 7:1 to 10:1 and these will still roll away on hills.
There is a formulae on worm gearing to determine the point at which a ratio won't wind back.
It's dependant on the pitch and the ratio, fines pitches lock earlier than course because of the lead angle.
Not sure if MH has this info but many worm drive books do.