I was using 8 mm metric screw to drive the axis of my small router The speed was not so disturbing for a small distance The max jog speed was about 6 mm per sec .Then I built a new router with a 1 meter long X axis but now I have a new problem :low jogging speed. I checked the cost of ball screw , rack pinion and belt drive but all of them require usd 100 per axis. So I decided to try steel wire drive method .I have tensioned a 0.2mm steel wire and wound it on a 8 mm steel shaft at 3 turns. With this way the motion transmission is perfect and there is no backlash at all.The only problem is linear motion of the drive shaft .When the shaft is turned it is forced to move axialy too .it is because of the wound effect of the wire. I hope I could describe what I mean . May be this picture help. The shaft will be directly driven with a step motor. Do you have any idea to compansate the linear motion of the shaft?

Murat..

Yes -- the screw is moving 1 thread laterally, per rotation. If you mount the screw on a nut, and then allow the whole thing to move laterally, the entry and exit points for the wire will remain constant.

A similar idea cropped up on DIY-CNC a while back. You might check the Yahoo archives.

-- Chuck Knight

solid wire?
Has anyone determined the wire life span?
D

it works on some photocopy machines more than 10 years.

Rceebuilder,

You made me curios. I do not understand very well your system. Could you, please, get in more details? Eventually a sketch will be helpful.

Thank you,

Zoltan

Zoltan , Here is the 2 type of wire driven gantry project. Of course the wire will be plied type. The only problem is how to compansate drive shaft's lateral movement.

rceebuilder,

If you are going to directly drive the 8mm shaft with a 1.8 degree/fullstep motor then you'll get 0.1256mm (0.0049 inch) of linear motion per step. With micro stepping it gets slightly better. 8mm rod with 0.2 m wire you can drive 600mm of linear motion with only 24 full rotations.. so that's only 5mm of lateral motion. Is that fatal?

I'm looking seriously at using the looping wire system on my first machine, which will be a simple 2d plotter and foam hotwire cutter.

Of course going with a small diameter drive rod gets better resolution, but more wear and tear on the wire and even more lateral movement.

Playing around with numbers I found that using a 1/4 inch drive rod results in a ~0.1mm step (0.0997mm but who's counting). Using English parts to get metric motion, a bit funny.

Ideas for the lateral movement.. I like Mr Knight's idea of using the nut and threads. If your steel wire 'threads per inch' matched that of available nut/screws you'd be all set. Changing the gauge of the wire can help get you there.

Charlye

How about setting the drive shaft at an angle to the wire: you know the diameter of the wire and that of the shaft, so calculate the angle, and squint the beast a bit. I used to do this on a home-made machine of very limited travel. It might work for you!
Or set two small ball-races, one each side of the shaft, to force the wire to enter-and-exit the helix where you want.

How about setting the drive shaft at an angle to the wire: you know the diameter of the wire and that of the shaft, so calculate the angle, and squint the beast a bit. I used to do this on a home-made machine of very limited travel. It might work for you!
Or set two small ball-races, one each side of the shaft, to force the wire to enter-and-exit the helix where you want.


That will work :)

I applied it to my 40" x 20" router in x and y axis . The result is awesome . Particularly I liked gantry speed. Precision is enough for me. I will send some pictures soon.

Murat..

You could use allthread, no? A 3 foot piece of 1/2" dia is about $5 USD, and add 35 cents for a nut. Not the greatest setup, but I think it'd be more reliable.

It occurred to me that my other post could be construed as rude, so I'll add that MIT used wire drives in some of their experimental robots.

The idea was to get force transfer from a small shaft (eg, the motor) to a larger shaft (eg, an arm joint) with no backlash, and cheaply. Very clever design.

Here are the photos of my cnc .

Here are the photos of my cnc .
Interesting.. took me a minute to realize that te Y axis motor is mounted ON the Z axis carriage.. very nice.. any pics of your X axis system?

Eddie

100u$ per axis isnt too much, i wonder. anyway please consider that time is money too. and when testing things out like this, mitakes are common, so if you make a mistake, you are trowhing money out. (been there, done that)
i live in argentina, when a ballscrew set for a 60x60x10xms router is almost 1500u$.

Hi,

Are you satisfied with the accuracy and backlash of this solution? Did you verify what force can you transmit through this system? How did you solve the issue with moving the wire along of shaft? What do you think about using steel tape, like those from tape meter or what name they have in English - they are use for measurements? I was playing today with the idea...even I was thinking about few radial electromagnets inside of pulley of let's say 20 mm diameter, powered from 24 V at 1 A through two isolated rings to improve the adherence coefficient which is w/o around 0.18..?

Zoltan

I have tried to mill some PCB . As you know PCB milling requires high accuracy . Also I tried to drill same 10 holes 20 times . It never missed a hole . I think accuracy is about +/ - 0,1- 0,15 mm . So in my opinion with this system wood working can be done succesfully. But works like pcb milling may not be satisfy . In my trials no backlash setting applied. It worth to try

The cable drive is actually a good system. This is a close-up (the best pic I could take) of a 1.25" bronze "gear" driving a 1/16" 7x7 stainless steel cable. This drive system is the GOTO drive for a 200lb telescope. It has to be very accurate to track the sky at 1000x .

I got this scope before I moved to New York from

www.webstertelescopes.com

They use this drive system even when the scopes weigh 600lbs or more. The cable makes 2 wraps around the gear. The gear is driven by a Pitman servo motor with a 2000 tic encoder.

If you lift a spring held lever, it releases the cable just enough to let the scope freewheel. This may be another feature one could use for CNC.

If you need better pics let me know.

vid

right now i cant, but i makes me want to build some sort of trying device for this system..seems pretty good.

Basically its the same system that we used on Shrimp Boats when I was younger to haul 1000's of pounds of fish out of the Gulf and onto the boat deck with. The rope wrapped around a smooth wench head and the hoister could let it free wheel idle in the air just by pulling or easing the tension on the other end of the rope.
So with this concept, you can get as much torque and strength as would be needed.
It may be that some of the accuracy was in the settings. Perhaps some minute settings in the size of the pully or shaft could be toyed with to acheive better accuracy.

Another concept that might be fun to try is chain drive. At least with a chain (like bicycle) you would have the same backlash over the length of the chain. This could be dealt with and accounted for in the software.
I was thinking a straight run of tensioned chain. Then a sprocket attached to the motor. You would need a skid plate to keep the chain in the srocket, but thats asy enough.
Probably could be done from parts in the garage or on the curb at your neighbors house.
Anyone toyed with chain?

I will sure try this out, im wondering if im going to use this system with a 1" rod, and with 25600 steps/rev (microstepping offcourse) i will get up to 0.00311 mm of resolution, and if the stepper is running at 300RPM, i will get 23938,992mm a minute.
(thats 942.48 inches a minute...) pretty good i think. today i went to a fishing store.
i got steel wire, that can handle a weight of 120lbs , for 0.42u$ / meter. i think this can do the job...

Hi all

Zoltan asked:

"How did you solve the issue with moving the wire along of shaft?"


Any trial ?


Thanks

I also thought about the cable system. On my first machine 66" X 42" I spun the nut instead of the 1/2-10 acme threaded rod (worried about leadscrew whip). Was getting about 55-60 ipm rapids with 128oz stepers. Currently I am building a new machine 30" X 72" with two z axis (z & a) on each side of the gantry, one will be a dremel and another a router. So I will be able to have a cut area of about 30" X 48" with both the dremel and router (tool offset). Upgraded to the Hobby CNC driver and I am now getting about 85 ipm rapids on the tests so far.

I like the cable system which would increase my rapids but my concern is when you increase your speed you are decreasing your driving force. going with the cable system with a 1/4" cable drive pulley I could increase my rapids to about 665 ipm that's over 7 times faster! But I will loose my driving force, don't know if it equates to 7 times less force but it will reduce force. Could build a reduction unit for the steper but this will increase cost. In my eyes a machine that can move nice and fast isn't any good if it can't drive the router through the work piece without loosing steps this may not be an issue for a plasma machine.

I still like the cable idea, have a lathe so this method would be cheap for me.

I am also considering going from 1/2-10 acme to 3/8-8 hoping that the smaller diameter will make up for the less threads per inch and increase my rapids by 20%? I am also looking for larger steppers.


John

so if this cable system is used for telescopes then it should be extreemly accurate if used on a plasma where no force is encountered? i wanted to use gearbellts but after reading this im wondering if i should go for wire instead

About this problem, i still didnt have the time to figure it out. but im thinking on a long "pulley" so the cable can move along it. (the size depends on the travel distance i guess)
any ideas????

Would it be better,not to anchor the ends of the wire?

I was thinking maybe if you used to free wheeling axles of the same diameter of the motors at each end with the wire looped in the same direction and for tension join the wire ends with a small turnbuckle. The wire will still more lateral but it wont effect the accuracy because it will be moving lateral on the idle axles at the same time.

Using this sort of system what would be the maximum distances could you expect to travel?

BIGZ1, if you use freewheling at the ends, it wont move, it will just rotate and dont produce any linear movement .... im i right? or maybe i didnt understand you at all!
any graphic to upload?

You got me. Thats the sort of genius you get the morning after an England victory in the World Cup(no hand of God to stop us now). Plus its very hot at the moment. (chair)

But bear with me. How about using the same setup. This time anchoring the ends of the wire to a micro linear bearing with an 1" of travel(usually go for $10 on ebay). This would have to be attached to the part you wanted to move. . I will try and draw something but life is a little hectic at present as I am expecting to be a father again VERY SOON. :cheers:

with your setup, i think you will have some friction (not the necesary) loss on wire itself. i will turn and wont translate any move to the piece.

i think that the stepper should be mounted and fixed on the part itself. so on the very same axys of the stepper, one should mount a pulley with 4 to 6 turns of wire arround it.. just like the detail picture of the telescope transmision system.

we need about 79.79 mm of wire lenght for each turn in 1" diameter pulley, so in 5 turns arround it , we have 398.9832mm.
ok. supposed that the wire is 1mm diameter, we need 1.4 meters to have 1 meter displacement . and we`ll have a 10mm variation at the pulley, from end to end.
so in the pulley we should have a 12mm groove (*) to let the wire enaugh room to move side to side while turning all the way (end to end)
it doesnt need any compensation , as the wire will move itself, from turn to turn...
and then come back to its "original " position, as we reverse the turn direction.

(*) 12 mm grove or "thread alike" so the wire can we wrapped with some sort of guide.

im i clear? im i right about it ????

I'm digging this drive system. Would have to be the cheapest out there. My trials will most likely use the big spool of heavy woven kevlar thread... some 250lbs strength per line... I'm thinking the thread may be more grippy than wire line.... but I sure to have enough to through some away on trials :)

(can buy very large spools on ebay for $25... and I bought it already for composite construction, and it's very handy stuff)

Just a thought,
Could a toothed belt be used instead of wire. I am led to believee that there are belts with almost no stretch.

Just a thought,
Could a toothed belt be used instead of wire. I am led to believe that there are belts with almost no stretch.

Yup, been looking into them recently. They all don't stretch, but you can pick what they're woven with, nylon, kevlar etc. But they have a cost that is higher than cheap drive screws, and approaching what appears to be the good middle-ground of drive screws. A belt the length for a 4 foot table is $29, with $10 for a pulley, so you're up running around $50 an axis, more if you figure tensioner pulleys or whatever.