View Full Version : Timing belt speed reducer with changeable pulleys

01-29-2012, 12:39 PM
Greetings! I'm trying to design a speed reducer for a CNC router using steppers. Depending on the material being cut, the speed reduction needed varies from about 4:1 to none (1:1). Here's the problem: how to make the pulleys changeable without introducing flex or backlash?

I'm thinking of a fully-enclosed box, with the input and output shafts supported on both ends with bearings in the side plates, and the pulleys mounted on the shafts with keys and setscrews bearing on the keys. That should be pretty solid. However, to change a belt or pulley will require removing one side plate. That means that either the bearings have to be slip fits in the side plate, or the shafts have to be slip fits in the bearings. Slip fit = slop = bad.

The only thing I can think of is to make the shaft a slip fit in the bearing inner race, but slit the end of the shaft 4 ways (like a collet) and fit a tapered threaded plug in the end of the shaft. That way, with the plug removed, the shaft will slide into/out of the bearing, and when the plug is tightened, the shaft will expand and be locked into the bearing without slop. Would that work? Other ideas?

01-29-2012, 01:00 PM
Why not use tapered bearings for the end that will be against the removable plate? The race stays with the plate and the bearing stays with the shaft. That will let you replace the belt, but you would have to pull the bearing to replace the pulleys. I don't imagine that you will have frequent pulley changes. :)


01-29-2012, 01:12 PM
Look and see how a variable speed drill press works. Both pulleys slide on their shafts. Not a very complicated set up and the change in speed is smooth.

01-29-2012, 01:52 PM
Tapered bearings will overheat at router speeds. Maybe you could make one shaft of pulleys on a hinge so it could swing closer and further from the other set of pulleys and tighten with a hand screw.

01-29-2012, 03:24 PM
If this is a Universal motor router, it may be easier to get the SuperPID controller from the guy in Australia that is selling them for around $150.00.
It has simple tach feedback so is accurate speed control.
Google SuperPID.

01-29-2012, 04:30 PM
Looks as if my first post wasn't clear. We're talking about a CNC router here. The speed reduction is not for the spindle, but for the gantry drive. The gantry will be driven by stepper motors that turn at less than 500 rpm. The feed required to cut plastic or pine can be as much as 15 times higher than the feed for cutting aluminum, and the precision needed for metal machine parts is much higher than that needed for, say, wood furniture. Hence the need for a speed reduction box that has multiple, fixed ratios. The other requirement is very little flex or backlash. Otherwise, round holes turn out oval, etc.

I like the idea of a taper roller bearing. There would have to be some kind of a nut that draws the inner race firmly into the outer race, so that it doesn't wobble.

Now that I think about it more (reading your replies always stimulates the grey cells), I wonder if at these low rpms and loads (and virtually no axial load) I could just use good quality plain bushings with grease and call it a day.

01-29-2012, 04:37 PM
Won't changing the number of times the screw rotates vs the stepper motor's shaft mess with the distance the computer thinks it is traveling?


01-29-2012, 05:17 PM
Yes it will. he will need two configuration files.

I would say you have a design issue that extends further than belts. How many volts are you running your steppers at? Are you micro stepping? How much? You should be able to get way more than 500 rpm out of a stepper if everything is set up right.

Another thing is you can belt down on a router to get more resolution but you won't necessarily get any more accuracy. Routers are traditionally rather spindly compared to a mill. You may never see the gain you are expecting.

Also look at high feed milling. Instead of taking deep cuts slow it takes shallow cuts fast. In the end it takes the same amount or less time.

01-29-2012, 11:53 PM

I will probably start with a standard motor kit like those offered by Xylotex or Keling, e.g., 425 oz-in steppers on 24V. My gantry will be belt driven by two motors. The torque curves I have seen show torque dropping beyond 350 - 400 rpm, and I do plan on doing most of my cutting fast and shallow. That's why I figured about 500 rpm max. But if you say I can run those steppers faster by "setting them up right," please explain. I'm all ears.

01-29-2012, 11:58 PM
24v is too low. Kick it up to about 70.

When it comes to motors, current limits torque and voltage limits speed.

Get good drives. On all the new stuff at work we use Geckos. Specifically the 201X drives. They will handle 80v max.

01-30-2012, 12:30 AM
On the mechanical side of things, it sounds like you don't have to worry about the reduction system keeping in lock/step, but you do need it to be free from slop, so pulleys and belts could be fine. If you're willing to open the 'box' and change the belt, then you might be willing to leave the box alone and make up an idler that can be pressed between any pair of pulleys instead. It would basically be like selecting a change gear on a lathe- pull a lever back, slide along to the right hole, engage lever.

In this case you could do the same with an idler wheel, or an idler and belt combo. Imagine turning an ordinary V belt 'inside out', then putting it around two flat-grooved pulleys. Between the pulleys is another pulley which can press against the side of the belt. When this assembly is placed over one pair of the reduction pulley sets, the V belt is forced to wrap somewhat around those pulleys. Because of this it has some significant torque transfer ability. You lift the belt assembly, move it along to your needed ratio pulley set, then drop it in place. It will self-align, and you would probably spring-load the third pulley in the belt assembly to force the belt to conform around the drive and driven pulleys. You could put the top 'gearbox' cover on a hinge so you could quickly open it to make a change. Maybe the top cover would actually carry the belt assembly, and have 'detent slots' in it to position the assembly.

Another idea which comes to mind is to use two cones with a belt around them. An idler on a control arm sets the position of the belt on the cones, and thus the ratio of shaft rpms. The belt in this case would not ride in grooves, but on the surfaces of the cones. The right belt would probably work fine for this system. The belt would be bending sideways a lot as well as conforming to the diameters it's riding on, so a normal belt would probably not work well like this. In any event, this is an infinitely variable device and may not be suitable. It should have near zero backlash though, unlike gears.

You could go motorcycle transmission style, which has all gears fully meshed at all times, but there's going to be backlash. If you were to make this up in toothed belt and sprocket instead, you could virtually eliminate backlash, but your method of clutching the appropriate drive sprocket would have to be pretty solid. You could do it with electromagnetic clutches. This method has all the drive belts engaged and turning at all times, so there could be an unacceptably high power loss.

Just throwing in a few ideas. Last one here is to use the multiple sprocket set, but just one belt. You would have to move the belt to change the ratio. I can see one way to do this - have the bottom of the 'gearbox' open, and have the toothed belt hanging down with a weighted idler within it. You would simply lift on both exposed parts of the belt and move it sideways to the sprocket set you wanted to use. No removing covers, playing with bearings, etc.

Of course, anything as exposed as this will have to be well protected from debris and wayward fingers.

01-30-2012, 12:40 AM

I will probably start with a standard motor kit like those offered by Xylotex or Keling, e.g., 425 oz-in steppers on 24V. My gantry will be belt driven by two motors. The torque curves I have seen show torque dropping beyond 350 - 400 rpm, and I do plan on doing most of my cutting fast and shallow. That's why I figured about 500 rpm max. But if you say I can run those steppers faster by "setting them up right," please explain. I'm all ears.

I have a wood router running the 425 oz steppers on a Xylotex board. The gantry is made from 2 layers of 3/4" MDF and is quite heavy. I have a 10 tpi acme lead screw and I have had no end of grief if I go over 30 IPM. So if my math is right, I am running them at 300 rpm or less. Faster than that and I am losing steps and the part is ruined. It has been so frustrating that I have all but stopped using it.

I would urge you to spend a bit more and go with the Gecko drives. The max voltage the Xylotex can handle is around 27 volts. I think the peak is actually 35v but 27v is the safe working voltage due to spikes. This is from memory so don't quote me on the exact numbers.

Now that I have a mill, I have considered rebuilding the gantry in aluminum to reduce the weight. I have also considered twin or triple start leadscrews to reduce the rpms on the motors. This also means getting ball screws for the faster leadscrews. This means investing more money in a machine that has caused me great heartache.

Take this as data only. There are plenty of people who have had success with this set-up but the actual cutting speeds are significantly less than the rapid movement speeds that people talk about. I was very disappointed with the actual time it took to make parts.


01-30-2012, 12:50 AM
Belt drive or rack and pinion is more than accurate enough for a router table. I would skip the screws. Also, belt and rack are more dust friendly than screws.

01-30-2012, 09:07 AM
What about just over driving the lead screw?

I've done that with an underpowered stepper that would not let me have a faster traverse. The motor, as usual, lost to much torque trying to make it go fast. 1:2 using a timing belt was perfect.

Forrest Addy
01-30-2012, 01:47 PM
Pretty hard to beat an electro-mechaniucal drive thats been enginered and sized for the work it has to do. Step motors come in all sizes from NEMAS 8 (fits in a teaspoon) to NEMA 42 (big as a coffee can). Drivers for them abound. Overlapping and expanding the range are servo drive accepting step and direction, quadrature, what ever command input. These range drom RC airplane elevator controls to 500 HP suited for ship's rudders.

So there is no lack of available equipment. The real problem is money, doing what you can with what you already have. My suggestion ius to just plain stop, think about your task, set down on paper separate lists of what you need to do, what you wish to avoid, and what resources you have available.

I see on eBay stepper kits consisting of motors and drives some up to 1600 inlb torque and 80 volts input but they are not cheap. MSC has both rack and pinion (the very stiffest) and lead screw and nut combinations (the most positive and irreversible) and toothed belt componentass suited tor linear motion (a bit elastic but the simplest to implement.

Working from your desires - positioing accuracy - and concluding a 200 step motor can be reliably positioned to 4 steps per revolution that's 800 steps per revolution. If your rack pinion was X dia or lead screw has Y pitch you can translats positioning accuracy to any refinement you wish. If you add encoders you can add positioning certainty that microsteeping only promises.

There are rough formula predicting your motor RPM from motor data and power supply volts. This gives you axis speed.

The final detail is axis thrust. which is a function of motor size, current and mechanical advantage. I've seen some very successful home brew CNC routers driven by size 34 step motors rapid position at hunbdreds of inched per minute yet fine detail parts to 0.002" feature tolerance yet hog out pocketing cuts that theaten to stall a HF drive 1600 watt router.

I didn't pick up in my skim of the posts what your present equipment is but unless its miniscule I'm quite certain you can obtain the speed, thrust, and positioning you desire with a single speed stepper drive accepting only minor compromises. But you have to engineer the drives. If you can't do the engineering there's no shame in that. My nephew fetches my stove pellets; I have a house cleaner now all because I'm old and getting puny. Not everyone can be young and vital, Not everyone owns the talants and training of an engineer.

Offer a beer to a drives guy to look over your problem and listen to his suggestions. Post your problems and you'll get as many well meant suggestions (and bum steers) as you get responses. Many will conflict, more than a few are either impractical, unresponsivee, or plain silly, only a very few can be taken seriously. We all like to help but it's hard to beat an on-site guy who can see the whole problem, dip into his expertise, make a few suggestions (and argue you out of a few strongly held misconceptions) and in general put your feet on a simple and direct path to the success you wish.

01-30-2012, 02:03 PM
I don't want to hijack this thread. I will post a new thread to the CNC section sometime soon.