[Dongle]

While standing doing some photocopying at the 9 to 5, I found myself needing to clear a "mysterious non existant jam" to have a better look at the innards of the copier.

It has a wire drive to control the position of the scanner head, which appears to have virtually no backlash and be very simple, a drum with a fine thread on it, piece of wire wrapped around a few turns in a loop around the scanner head, held in place on the head by similar means to timing belt clamps.

I have also seen this system used to raise and lower the elevator on a 737 so it's obviously a reasonably reliable and accurate system.

Any drawbacks to using this on a large scale machine that's only going to cut light materials? would certainly beat the price and effort of pretty much every other solution I can think of as I can make it entirely out of scrap in the garage!!

[Corvus corax]

I can't see any reason why it coulpn't be used, and I'm surprised I haven't seen anyone else mention it. This system is used in manual analog radio tuners as well, and it maintains perfect accuracy over hundreds of thousands of cycles. I would use a fairly strong, supple cable for the "string", as its important that it not have any stretch, and the system will have to be protected from chips.

Span will be limited somewhat due to the influence of gravity on the cable, and the size of the pulley will be very important- it cannot be too large, as the mechanical advantage, and resolution, drops quickly with increasing size, and the possibility of slip has to be addressed as well. The system would be similar to a rack and pinion in terms of resolution/ speed, but far less expensive. The drive pulley would need to be carefully made, though. For inexpensive plasma, it could be just the thing. Might be harder to prevent oscillation in a larger system, though.

[Dongle]

I thought so, I might just have to give that a go - I'll probably make the machine on an "and/or" approach so I can easily switch to alternative means if needed.

My only concern when I sat down with paper and pencil (well, cad) was one you have just pointed out, the diameter of the drum for the wire to wrap around. I think I'll need to gear the motor to the drum which obviously then could introduce backlash. The chap who sold me the motor and boards reckoned I should aim for about 1-2mm movement from a 360 deg rotation of the motor - some serious gearing needed for that.

[EverythingCNC]

I have a Roland MDX-15 and it uses cable, pulley, drum system. I did some testing with the idea for a larger router, but the cable sagged and stretched too much. I think, it works fine for 12-15" axis, but not for a longer distance. Since ballscrew is $1.23/inch and a matching ball nut is $23/each, I don't think the cable system can beat it. Precision drums and pulleys are expensive. I looked them up a while back, and a 1" dia. drum cost $80. I will stick with preloaded ballscrews, either with over sized ball or double nut.

[Dongle]

hmmm thats not so good news as my router is going to be large . UNfortunately the cost comparison for me in the UK is a little difference - I can turn up the drums no problem (or indeed beg some off a mate) and the cable can be had for free from my father, whereas the cheapest I've found appropriate ballscrew is £7 ($13?) per 100mm and ball nuts at £50 ($90!) each
unless anyone can point me at cheaper in the UK?

[ballscrewed]

Everything, Where is it that you are finding those kind of prices on screws? If you would be so kind could you post a link for us. That sounds like just the ticket for my new {in the works} machine.
Sorry to highjack
Aaron

[Dongle]

Hijack away Ballscrewed - I want to know that as well!!! at those prices it could almost be worth importing them, though I'm not sure about getting 5.5ft long ballthread through the post from the USA and having it arrive straight!!!

[EverythingCNC]

Aaron,

I found two sources. I think homeshopcnc can machine the end for you.
http://homeshopcnc.com/page2.html

Or, go to mcmaster.com and search for ballscrew. The 5/8" screw is $1.27 per inch and the matching square ball nut is $23.85 each.
You might want to go with 2 ball nuts per screw.
I bought other things from mcmaster before and they sell quality stuffs. Sometimes they don't advertise the brand, but I got all brand name items.

Let us know how it works out for you.

Tan

[EverythingCNC]

Dongle,

Where are you?
US has no problem with shipping. I bought 8 foot 1/4" brass, aluminum rods through mail order and they came straight. They were shipped in hard cardboard protective tube, and I could not break the tube in half to dispose it. Just ask them about packaging options. I think DHL ship world wide. Buy insurance.

[Dongle]

Unfortunately I'm in the UK Tan - our postal service is erm... 'sadly lacking' in most respects - might be worth it but it's a fair bit of cash to spend and have them arrive bananna shape. I'll send them an email and chat it through. If they are properly packaged then they should be OK!

[ballscrewed]

Thanks for the links Everything, It will come in handy for all of us. I will post results of what i find and use as well as the end results.
Aaron

[Mcgyver]

to your original question, this type of drive is called a capstan, just like it was used on the old ships. you'll see it on the x axix of manual surface grinders and T&C grinders - places where you want a smooth, sensitive and quick action but don't much care about precision (at least along the x axis). it has no backlash, but a downside is the cable will stretch and the gearing is wrong for most machine's feed rate.

[Dongle]

The gearing I agree is a problem, hadn't considered stretch of the cable too much, I pressumed it cant be much worse than the belt drive I've seen on some larger machines?

[Mcgyver]

not sure what type of belt drive you mean, if the position along the axis is determined by tension on a rubber belt, it could be substantial. if you mean from a motor to a feed screw, there'd be some, but it would a fraction of the previous example.

i can't intuitively say the stretch of the steel cable would be less than the belt, I don't know enough about it and at some point it will depend on the their relative sizes and elasticity would make the lines on graph cross.

so far as accuracy goes, it's all relative to what you need - you can only determine whether it will be accurate enough in the context of how accurate does it need to be . Maybe some of that new synthetic spider web material will be the answer to capstan feeds!

My (yet untried) idea is to make a fluid capstan – a reversible hydraulic gear pump and light duty (pneumatic) cylinders to push the table along. I like the capstan idea but for a metal cutting machine it needs fine control and the fluid avoids the stretch problem. One day I’ll get to experimentation on it

[Dongle]

belt drive as in position along axis. I suppose you are correct, it's all relative with the accuracy. I could probably get away with accuracy so poor you would scream at it for the task in hand, but knowing what I'm like I wont want to strip the machine and rebuild once I've finshed the task!

Fluid sounds interesting, how would you get feedback to the position of the table? surely with a gear pump it wouldn't be accurate enough to say 1/xth revolution = 1/xth mm travel? (maybe it would?! not done much in hydraulics)

[Mcgyver]

you're right, there'd be slippage with the gear pump. haven't worked out all the details but it would be a closed looped system relying on linear encoders for positioning. The orginal thought was that it might be a cost effective alternative with zero backlash to ground ballscrews with the added benfit of low wearing parts, but like i say its only an idea not fully developed, want to get some time to explore it one day...

[EverythingCNC]

Ballscrewed,

I am glad that I can share the information I know with you and everyones.

Back to stretching problem with the cable. I hung a 3/16" multistranded steel cable a little more than 10 feet in length and put enough tension on it to keep it straight. It stretched and sagged down about 1 foot after 2 days. I added tension to tightening it up and it sagged again by the same amount after a few days. I think the cable can keep getting thinner and thinner. That was just the weight of the cable and without additional weight hanging on it. A larger, stronger cable might stretch less, but I will have hard time getting it to go around the drums and pulleys.

Please share if you come up with new ideas.

Tan Ta

[jimini]

In my days in the RCAF (F86 sabre, CF100 jet aircraft) we had cable controls that worked very well.
They had to be pretensioned to allow for the difference in expansion and contraction between the aluminum and steel at high alltitudes, but they held the tensions very well.

[Mcgyver]

must have been quite an experience. it sounds more like an analogue situation though, you apply force until the control surface is where you want it.... would you ever notice or would it matter if there were 5 thou (or whatever) of elasticity in the cable? have 5 thou of play on the climb part of an interpolated circle will for sure matter if you're cutting metal, the cutter could/will bite in and break.

I’d guess a certain amount of the problem is the nature of the braid, but if that is worked out, any metal will still stretch an amount depending material, cross section, force and length. As a cable has a small cross section, stretch will matter for precision apps. maybe for a laser or plasma cutter, ie something without cutting force? Still can’t see it being a big win over a rolled ballscrew

[RICHARD ZASTROW]

What about a perforated stainless steel belt (.010-.015" thick) on a sprocketed flt pulley? These are commercially available in almost any width. I used one to move a laser on an engraver.
RZ