We are finalizing the design (for the umpteenth time) for an 8' x 4' CNC router. All of the previous designs used a 5/8 or 3/4 inch diameter lead screw to drive the "X" & "Y" axis'. Now we are considering replacing that with a Rack & Pinion system. We were afraid of the amount of "whip" there would be in a screw of such a small diameter that is over 9 feet long.

Any thoughts, pro or con?

HOW ARE YOU GOING TO CONTROL PITCH ERROR??? A COUPLE OF IDEAS= GLASS SCALES VERY EXPENSIVE!!@@**## OR HAVE THE RACK EDM'D ALONG WITH THE GEAR. YOU COULD ALSO HAVE THE MACHINE LASER CALIBRATED, AND HAVE ERROR COMPENSATION IN THE CONTROL.

Haven't got that far into it yet. Like I said, this is a new "thought" that just came up yesterday afternoon. Our concern over the the whipping of a 9' long screw led us to consider alternatives. But that is the kind of feedback I'm looking for. Thanks

I would go rack and pinon myself, it would allow for more adj. and would eliminate the need to support that lead screw. You can make a "L" shaped support for the lead screw, you make it so it can rock back and forth so it does'nt interfere with the nut.

One thing you can consider is mounting the motor on the axis and turning the nut via a timing belt/pulleys. This nut/pulley would need to be located between two thrust bearings. Then the screw will be installed on the machine in a fixed position (doesn't turn). Now you have eliminated whip, and you have much less mass to turn.

Eric

For a screw as long as you are planing I would not go smaller than 1". How will you prevent backlash with a rack and pinion design? Are you thinking of having two racks, one at each end of the gantry? What about using a long timing belt that goes over a timing gear?

You may need a gearbox for that setup.

Eric


What about using a long timing belt that goes over a timing gear?

Yes, you'd surely need a gearbox for a rack and pinion or timing belt type drive because of the mechanical disadvantage the motor would acquire by being used in this fashion. Instead of one turn of the motor causing a movement of .2", you'd get a movement of 2 or 3 inches (without a gearbox). Not good, IMO.

What do the commercially made units do to solve the screw whip problem?

There is a solution to avoid whip on long lead screws. Leave the screw stationary. Mount a belt drive nut to the carriage. Long Bobcad tables used the method.

"The Rotating Nut ball screw line can increase traverse speeds up to 200 m/min with efficiency over 90%. The RN series consists of standard diameter and standard lead screws around which a flange-type precision housing has been designed.

The housing enables a timing belt to drive the nut and contains the preloaded rotating ball nut assembly. The bearings in the housing are grease packed and sealed for life, so no additional lubrication is needed.

Available in ISO precision classes 3 and 5 in both Medium Duty (single track), or High Duty(twin track) configurations.

Thomson Industries, Port Washington, NY, "

Bet you won't find a rotary ballscrew nut like that at Walmart ;)

I also like the rotating ball nut. This is how I designed my Z axis movement in my drill-mill CNC conversion.

Backlash in rack and pinnion can be an over rated concern. I tested the backlash in my nearly 10 year old Digital Tool rack and pinnion router a few weeks ago. It was .002 in both the x and y axis. I find that to be a non-existant issue in woodworking. We keep the machine clean and tight, but I've never done anything to adjust the rack and pinnion.

Oh, and there is backlash compensation in the control softwre but I have it turned off.

BH Davis

I like the idea of the timing belt that Jeff Davis suggested. I'm going to have to check into this.

As far as a gear box on the Rack and Pinion setup, we were not going to drive the Pinion directly but rather with a belt/pulley system. We were going to use a pulley configuration that moves the gantry .100 per revolution of the motor.

Thanks for all the input and ideas. Keep 'em coming.

HAVE YOU EVER THOUGHT ABOUT USING A LINEAR MOTOR???? THEY HAVE PLENTY OF TOURQUE FOR WHAT YOUR DOING

No. A linear motor has never been considered, but I'm open to all ideas and suggestions. Tell me more.

try this out.

http://www.h2wtech.com/highdcbrushless.htm

if this doesn't fit, go to http://www.globalspec.com/
and search for linear motors.

if it's not here it not available.

This seems to be the way the future is going to go. Very flexable, and very accurate, if set-up correctly.

Chad ,

How do you controll a motor like that. ?????


Gary

Hi Chad,

I liked the two axis linear stepper on that site. So much for ball nuts and lead screws.:)
The linear motor was also impressive at 240 ips, 640 lbs and 12g's.

Chris

Very good, Chad. I really, really like the linear motor idea. I guess I need to go back to the drawing board one more time and see how this will compare with some of the other configurations. Has anyone actually built their own CNC machine using these?

THEY ARE ALREADY ON CNC'S. I HEARD THRU THE GRAPE-VINE THAT MITSUBISHI ALREADY HAS A WIRE EDM THAT USES LINEAR MOTORS, THEY ARE ALSO BEING USED BY ALL THE BIG GUY'S FOR ASSEMBLY\AUTOMATION. TO GET INTO METAL WORKING THEY NEED TO BE A BIT MORE RIGID. WE ARE GOING TO BE TRYING TO ATTEMPT TO USE THEM FOR SOME OF OUR FORMING.
I BELEIVE THEY ARE CONTROLED BY A PLC, BUT I'M SURE THEY HAVE CONTROLERS THAT ARE SOLD BY THE SAME COMPANIES. CHECK OUT THE LINK ABOVE. IT'S MY FIRST PLACE TO GO WHEN I'M LOOKING FOR SOMETHING, AND I HAVEN'T BEEN DISPOINTED YET. GOOD LUCK AND KEEP ME POSTED.

Where do you measure torque on a linear motor? :^)

I bet a linear motor would make a great cnc "slotting head" drive for a mill! Conventional slotting heads are slow and cumbersome and not much fun to set up.

As for how torque would be calculated, I suspect it might be derived from horsepower, which in the case of a linear motor, should be measurable almost directly by testing how much weight the motor can lift in a given time interval.

In the case of a rotary motor, the horsepower is more difficult to measure directly because the radius of the crank arm must be accounted for in standardizing a comparative measurement.

Hi, I have put together a large tble using the timing belt system as a sort of "rack and pinion" as suggested by Jeff. See some photos here;
http://www.axxustech.com/Axxus4x8/axxus_4x8.htm

Hi, I have put together a large tble using the timing belt system as a sort of "rack and pinion" as suggested by Jeff. See some photos here;
http://www.axxustech.com/Axxus4x8/axxus_4x8.htm

Very nice application design of the Beltdrive.
Do you have some figures on the belt tension applied?
Is this tension a percentage of the belts load rating?
How consistent will this tension be over time is there
any continous streaching going on that would lower
the tension over time?
What are reliable accuracys archivable after calibration?
Sorry to ask so many questions but I am very curious
about this design and have been considering it for
some time.
The lack of any real physical info is making me cautious.

Where do you measure torque on a linear motor? :^)

You don't. You measure force. A linear motor doesn't develope any torque.

Mechanical power of a linear motor [W] = force [N] * speed [m/s].
Mechanical power of a rotary motor [W] = torque [Nm] * angular speed [rad/s].

Arvid

Very nice application design of the Beltdrive.
Do you have some figures on the belt tension applied?
Is this tension a percentage of the belts load rating?
How consistent will this tension be over time is there
any continous streaching going on that would lower
the tension over time?
What are reliable accuracys archivable after calibration?
Sorry to ask so many questions but I am very curious
about this design and have been considering it for
some time.
The lack of any real physical info is making me cautious.

We have done several testing with belt drive system, the use of them is ok with short table or with slow table. With a plasma table with a 600 inch/min cutting speed you can't use them because they always have a little flex depending of the force applied to the axis, maybe +/- .050. So if you cut a corner at high speed, you may notice the head swing away on the turn , but as stated depending of the application....Precision issues is negligeable in wood, but if you are in metal and you want tolerance under .001 i don't recommend it. But on my wood router i have belt drive on x and y and i like it, very smooth, silent and no backlash!!

First you say it flexes .050, then you say it has no backlash???

A linear motor would cost a fortune for DIY CNC. They also need a linear encoder and the motor itself needs a magnet track as long as any one axis. In general, linear motors are better suited to high speed positioning than they are to machine tools since they have low peak forces, but are capable of very high speeds. There are some machine tools that use linear motors, but they are pretty rare and cost much more than a ballscrew version.

David B.

What sort of accuracy is a rack available in.....what is the variation per foot and per tooth?

First you say it flexes .050, then you say it has no backlash???

It's hard to explain as english is not my first language. But here my explanation: Backlash is a kind of play in balscrew or everything else that you can feel by moving foward and backward an axis. This play could transformed in vibration if too much play. In the belt you don't have backlash because you can bind it, but if you apply counter-pressure you can see a deflection of the belt, more force applied, more deflection you have, it act as a king of shock absorber, so you have what we call following error, it's similar with the error found with servo system

I think Al the Man could explain better than me in english!!

I understand what you're saying, but what good is no backlash if the belts stretch? What type and size of belts are you using? are they steel reinforced?

I understand what you're saying, but what good is no backlash if the belts stretch? What type and size of belts are you using? are they steel reinforced?

Sorry to bud in her ger21, aghobby has given a good expanation about
Backlash verses stretch or flex.
Stretch as found on a rubber string for example is found on all
materials even solid columns of massive steel.
The higher the forces the larger this stretch will become.
Works also on compression.
Obviously the percentiges vary by the material used and by the
applicable size and shape of the crosssection involved and there
length.
Timing belts have carbon fibre reenforced Layers, there loadrating
is a indicator on how much force they can comfortable deal with.
A wider belt will have higher loadratings.

Timing belts have carbon fibre reenforced Layers.


Carbon Fiber? I was aware of kevlar and nylon, but not carbon. Carbon is not very tough and I would think a bad choice for belting. Can you point me to some specs on carbon reinforced belts??

Also, belt drives are better in a dusty environment. It would also be very easy to double up on the belts if more stiffness is required. Just use a belt on each side of the gantry, and use a shaft to connect the pulleys. The link below show s a machine with a 1" belt. I think it's a very elegent and simple solution that would last almost forever. The main downside is that a belt reduction needs to be contstructed to drive gantry. This adds some complexity, but is worth it IMHO.

David B.

http://www.axxustech.com/Axxus4x8/axxus_4x8.htm

Hello Guys..

A friend of mine here has 5 CNC routers inhis shop. Theyare all Matcam(Multicam) and 2 of them are 4500mm long X travel. When he bought the machines they all had rack & Pinion drive and he found them not suitable and has since changed them to Dual 40mm Ballscrews with rotating nuts. The resolution is superior to the rack. And these babies can move fast... close to 1000 IPM cutting through 18mm +. The Ballscrews are supported in 2 places over the lenght by a spring lever with a roller on top...
When the nut assy comes near the support the support drops down (spring loaded) and returns after the nut assy goes past.. very simple and effective. I honestly don't think you could find a better system. But it may be too expensive for most people.

Horses for courses as they say.. most of us have to compromise somewhere. Personally I went for belts on each side.. 50mm belts on 25mm shafts at each end of the table... works brilliantly... even the 2 inch belts with steel reinforcing do have some stretch, when my x gantry moves there is no way I can physically stop it.

cheers

Frans

You are correct kevlar and nylon reenforcement is most common in timing belts.
Carbon fiber is used on plastic car fenders I got it mixed up sorry.

Hello Guys..

A friend of mine here has 5 CNC routers inhis shop. Theyare all Matcam(Multicam) and 2 of them are 4500mm long X travel. When he bought the machines they all had rack & Pinion drive and he found them not suitable and has since changed them to Dual 40mm Ballscrews with rotating nuts. The resolution is superior to the rack. And these babies can move fast... close to 1000 IPM cutting through 18mm +. The Ballscrews are supported in 2 places over the lenght by a spring lever with a roller on top...
When the nut assy comes near the support the support drops down (spring loaded) and returns after the nut assy goes past.. very simple and effective. I honestly don't think you could find a better system. But it may be too expensive for most people.

Horses for courses as they say.. most of us have to compromise somewhere. Personally I went for belts on each side.. 50mm belts on 25mm shafts at each end of the table... works brilliantly... even the 2 inch belts with steel reinforcing do have some stretch, when my x gantry moves there is no way I can physically stop it.

cheers

Frans


Frans,

Is there any way you could take some pictures of his setup and post on here, I'd love to see that.

My $.02, take with a grain of salt:

"Stretch" in any material can be calculated. Very easily... use Hooke's law and look up "E", Young's modulus of elasticity, for each material. Wikipedia is your friend.

Belt drive for a 9 foot X-axis router? You gotta be joking me. I'm not sure what grade of Kevlar is used in timing belts, but Kevlar 49 has an E value of 120GPa... that's 17400ksi. Compare this to steel, which has an E value of 30000 ksi, and the cross section of even a tiny 5/8" ballscrew is many, many times larger than a belt (minus all the neoprene, just the kevlar). Nylon is even worse I believe.

The problem with rack and pinion is that it only achieves maximum accuracy when the teeth mesh perfectly. It loses accuracy when the pinion rolls "in between" rack teeth. I suppose you could fix this problem by mounting a giant pinion gear such that 2-3 teeth are always touching the rack, and 1 tooth is always fully engaged. Of course, you'd need one hell of a gear box.

Companies like Komo fix this problem by making a helical rack and pinion. That way, a portion of each tooth is always fully engaged in the rack. This system probably costs justs as much as a quality ground ballscrew... maybe more.

Hi Estop, I don't want to bore you with pet theories and designs but some time ago when the problem of long ballscrews with whip on 8' X 4' tables raised their ugly head and no real solutions as to how to overcome the problem surfaced, I started to think of an alternative solution.
I expect you've already got a machine built and are now getting the spindle whip problem.
Anyway for the record, I drew,(doodled) up a design that had twin heads instead of one to cover an 8' X 4' table.
The machine is a fixed gantry type of router with moving table.
There are two heads spaced 4ft apart, on either side of the X axis on the gantry, and move independently of each other in the Y axis plane
The purpose of which is to have the table moving only 4ft and each head to cover each end of the table.
The right hand head covers from the begining to the centre and the left hand head covers from the centre to the left hand end.
This way the table only has to move 4ft with the right hand head cutting, and when it reaches the end of it's travel (table centre), the head retracts and the table returns back to the beginning.
The left hand head takes over and cuts from the centre to the left hand end.
I've made a rough model to test the feasability of the coverage and it works OK.
The machine overall length is reduced also as the table is now only moving a 4ft length on the rails.
Ian.

Hi handlewanker,

I would LOVE to see a picture or two of that contraption.

Might be kinda expensive with two spindle motors. You'd also have to deal with lead-in/lead-outs when you move from one spindle to the other.

And, of course, two Y-axis assemblies, two Z-axis assemblies. $$$ Wouldn't it be way easier and cheaper just to get a 1.5" ballscrew with 1.000 lead?

Companies like Komo fix this problem by making a helical rack and pinion. That way, a portion of each tooth is always fully engaged in the rack.

According to their website, and the few Komos I've seen in person, they all use ballscrews.

According to their website, and the few Komos I've seen in person, they all use ballscrews.

Wrong.

X axis is controlled by precision helical rack and pinion system.

http://www.komo.com/CNC%20Routers/Machines/solution_series.htm



.

When I was looking to verify that, I stopped reading after the previous paragraph that said ballscrews. They also have a "positioning systems" page that only mentioned ballscrews. Thanks for the clarification.

Hi Zumba, cost would be initially more, but as I see it the benefits of reduced table slides and shorter ballscrews is a definite bonus.
Th 8ft travel of a table would make the machine very long anyway, and reducing the travel to 4ft means the base of the machine can be shortened.
Correct me if I'm wrong but the base of a machine with an 8ft table and 8ft of travel would have to be at least 16ft.
For home construction, with limited means for allignment, this becomes a real problem.
The slides on this "revolutionary" design would only have to be 8ft long, assuming the linear bearing housings under the table are located 2ft in from each end of the table, and this would make allignment in the X axis quite easy.
As far as cutter control is concerned where the first cutter finishes the second cutter starts and as you can move a cutter to .001" 's with a ball screw, pick-up of the second cut is not a problem.
Ian.

Here is an interesting way to eliminate rack backlash, I thought of building one of these and then ran into it on this site,

http://www.atlantadrives.com/zerobacklash.htm?gclid=CL2XsamT4JACFQsvYAod_Bsu7A

dang! too slow again!

Anyone have any input on the different types of timing belts and which is the best to use for a CNC stepper motor?

Anyone have any input on the different types of timing belts and which is the best to use for a CNC stepper motor?

This should give you a lot of information.

http://www.sdp-si.com/D265/D265Cat.htm

I have unqualified statements to make:
Helical gears and racks put extreme side forces on the motor bearings.Does this mean some angular contact bearings are in order?
Belt drive is very expensive when you consider the cost of 30to one gear reduction.Almost as costly as a rolled screw.
R&P is dirt cheap and the way to go on large tables.
My $2.00 worth[inflation
you know]
Larry

Larry, belt drive is virtually the same as rack and pinion. If you need to gear one, you'd need to gear both of them. There's no difference if using a pinion with a 1" pitch diameter on a gear rack vs a pulley with a 1" pitch diameter on a belt. Belt drive would be slightly more expensive because belts cost more than gear rack, and pulleys a little more than pinion gears.

Hi
I have most of my gear to ready to convert my chinese mill to CNC. A local here has done it to his mill which is the same as mine. I am in the process of replacing the acme threads on my mill and have been searching for ball csrews but in doing that I came accross this site

http://www.nexengroup.com/rps/
its very interesting...wait for the video to load

Attached is a couple of pics of his mill

Carbon Fiber? I was aware of kevlar and nylon, but not carbon. Carbon is not very tough and I would think a bad choice for belting. Can you point me to some specs on carbon reinforced belts??

David B.

http://www.axxustech.com/Axxus4x8/axxus_4x8.htm

Carbon makes great sense for precision belt design. Carbon fiber has a modulus of elasticity of 77,000 KSI and Tensile strength of around 820 KSI and weighs next to nothing!

Gates manufactures their new Poly Chain GT belts with carbon in place of aramid fibers, check it out:

http://www.gatesprograms.com/ptsavings/rollerchain?sid=gnothingtougher

I've been looking into a belt drive solution for a filament winding design I've been working on and will most likely settle for something cheaper than this but if I had the money I'd go with carbon belts for sure!

Thanks for all the info people!

If you'd be prepared to DIY, why not consider something like this linear cam drive?

http://www.cim.mcgill.ca/~rmsl/Index/Documents/papers/slide.pdf