No belts, twice the accuracy and speed (CABLE!!!)

So this is a concept I've been going over in my head for quite a while now. I figure I'd post this to see if someone else had already tried this approach. This idea came to me while working my previous job at a machine shop. One device that we produced seriously caught my attention, it was a test tube pick and place device.

Now I will confess I was slightly turned on at the sight of hundreds of brand new NEMA 17 motors we would purchase
It was the manner and approach that they were used that had a profound effect on me. The pick and place machines we produced were a perfect 3 axis cartesian coordinate system. Now granted, they lacked the support necessary to support the extruder/heater assembly but the proof of concept was amazing.

The basic idea is that instead of toothed belts, wire rope was used instead. This wire rope was wound onto a large spool with grooves to hold the cable in place. The other key feature of this machine was a double pulley system on the carriage itself. There was another RepRapper who created a wonderful similar double pulley design that used belts. I have seen this system working in a production machine on a small scale. The only thing necessary now is an adaptation into a RepRap machine.

The basic concept behind RepRap is a self replicating machine. Obviously however, there are many components that are very difficult to produce with a 3D printer. I feel a more practical goal is to produce a machine with a minimum of SPECIALIZED parts . Basically my goal is to help design a machine in such a way that any non-printable parts can be purchased at any common hardware store. It also seems to me that the RepRap Prusa is becoming a wonderful development platform for all the new design innovations, as such the cable double pulley system will be built with that in mind. I am already in development of this system for my next Prusa, I only ask if any of you know of anyone who has attempted this method to provide a link

It has been suggested in the past by many, myself included [www.thingiverse.com] and the reaction always seems to be it was tried and failed but no one has produced any links or documentation on said failures so I still believe it is viable.

---

FFF Settings Calculator	Gcode post processors	Geometric Object Deposition Tool Blog
Tantillus.org	Mini Printable Lathe	How NOT to install a Pololu driver

Interesting concept.

Main thoughts:
The cable should stretch very little under tension. Otherwise you would be required to wind it from 1 spool to the other quite a number of times to obtain an equal stress throughout the cable.

How does the solution you saw solve the implicit non-linearity of the system? (e.g. when the spool unwinds, the distance to
the guidance pulley changes, this distance would be different with each revolution of the spool. This would imply the need for a wire tension-er and a translation step in software to achieve linear movement of the platform. The error can be reduced by moving the spool a good distance away from the pulley but this would mean a lot of wire to be used.

Gut feeling:
It would be possible but probably not really practical to overcome these issues to replace a €10 belt.

evert

Forrest Higgs has done a Darwin style machine with Wire drive - See Here

---

[richrap.blogspot.com]

If i recall correctly he did use a screw drive to drive a single axis and transferred the motion that way to the other vertical support points.

Why do you think this has twice the accuracy or speed?

---

www.Fablicator.com

In the system that I observed, the ends of the wire were secured with a spring-tensioner type system. This method keeps the wire constantly taught, and any slack will be worked out. The double pulley system would improve accuracy by removing any backlash present in the machine. The wire drum could be easily printed and could lend itself to another possibility.

Another idea I had came out of the sheer panic I felt when calibrating my machine. Like almost every other reprapper out there, we have seen our machines attempt suicide when the opto flags miss. While working at my previous job, I also used a special torque screwdriver. The foot pounds of torque was set with a dial and the screws were tightened until the handle spun freely.

So my idea would be to print a hollow cylinder the wire would be wrapped around. This would allow the torque system to be placed inside the drum. In the event the opto flags miss, the cylinder could spin freely and allow the machine to spin without destroying any components. Its just an idea but it might be very useful.

I think you should look at the old stratasys FDM1600,1650 or the 2000.

They all had a cable based positioning system, from the info I have the conclusion is as the machine and cables heat and cool accuracy will change because the cables will change length.

I remember reading a service manual for one of those machines and they had a special note in the calibration procedure was to let the machine fully warm up otherwise the results could be affected.

Stepper motors are essentially torque limiting devices. If you crash the carriage they simply skip and make noise. No need for an additional torque limiting mechanism in the (would need to be large) drum.

Belts have essentially no backlash and don't require a spring tensioner system. They are cheap, and widely available. Also, nylon/fiberglass/Kevlar belts will last longer than metal cable (even the metal spine belts have some issues) because of metal fatigue from constantly bending/unbending.

Wire drive also has some problems because the theta of the wire changes as the spools wind/unwind which can lead to small positioning errors unless accounted for.

I've heard one of the reasons wire drive was originally used in the stratasys machines was because the plastics originally used in belts could not handle the high chamber temperature they were printing at without stretching. Modern belts survive much better at 70C, and the high temp/expensive fortus machines use ball screws.

These are the reasons I use belts instead of wire. Your mileage may vary.

---

www.Fablicator.com

Wait a minute...stepper motors are torque limiting devices? So are you trying to say that I shouldn't have a massive heart attack if the flags miss the opto sensors? When I was calibrating my Prusa, at first I thought I was a rockstar. The hot end was working, (before I fried my thermistor) and was extruding little string balls. I got the pronterface software and skeined, a whistle. I then hit the magic print button and then quickly had a heart attack as the build area did not fit the travel area. Off topic question, but under manual control I can heat up my hot end, extrude plastic knots, but under print settings, nothing happens. I'm pretty sure I need to adjust my configure.h settings in my fireware? The machine goes through all the motions of travel, just no extrusion. Yet, when I have manual control I can make string balls all day long.

So yea, my whole idea behind the wire cable method was ease of accessibility. The toothed rubber belts I see as a specialized component, an obstacle in the self-replicating goal. I have been working at a big box home improvement store since high school, so I try to carefully memorize common components. While wearing my engineering hat, I try to design new inventions with this concept in mind. The next best thing to completely self replicating IMO is being able to get it at any hardware store in 5 minutes

Accessible is one thing, but it was nowhere in the thread title

Yes, steppers are torque limited. Unlike servos, they will happily skip steps all day long when your carriage misses the opto and rams into the frame. They will vibrate and be noisy, but shouldn't hurt anything unless you have something fragile where they collide.

My first repstrap homed the X and Y axis by over travelling the axis limits until you were sure the carriages were at their physical limits, and backing out the zero point from that known location.

The z axis is another matter. There is enough gear down from the threads that by the time your motor hit's it's torque limit the hot end is well on it's way through the bed.

As a side note, I have used ball chain from a hardware store in place of belts. It's not pretty, but does work well enough for a proof of concept machine.

---

www.Fablicator.com

I'm still building a repstrap with cable drives, and I posted these concept drawings a couple of years ago. With several wraps around the drive cylinder, there's no slippage and the amount of stretch means that the return pulley can be set up to accomodate the occasional adjustment, constant (spring) loading isn't needed. I started on this because the z drive only needs one screw because of the parallellogram cable, in addition to the cable I used, the stranded leader wire used for fishing, being available everywhere, in a variety of strains/thicknesses. The extruder traverse drive is simpler too. Turnbuckle-type tensioners aren't needed, and the ends of the cables can be fed through holes in the drive drum and tied off if the drum's hollow, or glued, screwed or wedged to retain.

Edited 2 time(s). Last edit at 12/21/2011 09:57AM by murrayd.

As CdnReprap notes, the old Stratasys FDM series used a cable driven X-Y stage. Here's a photo:


More details are here: [3d4u.org]

---

[haveblue.org]

How would fishermen wire work for this? Would it be any better?

... the Nylon type is much to soft/springy, best use fishing lines with Kevlar or steel core ...

---

Viktor
--------
Aufruf zum Projekt "Müll-freie Meere" - [reprap.org] -- Deutsche Facebook-Gruppe - [www.facebook.com]

Call for the project "garbage-free seas" - [reprap.org]

I cant personally see the benefit if cable. I did consider it when designing my mod with the idea of being able to get a smaller diameter drive off the motor. But I just didn't think it was worth it and would be hard to get good grip on a small pulley. The belt pulley is small as is.

One of the ideas behind the double (And if you designed a machine from scratch triple pulley.) is it could give you the torque and accuracy to do light milling or routing. There, speed is much less a requirement but torque and accuracy are. It would be cheep and simple compared to a screw system where it needs to be fairly fancy to prevent backlash. Fancy by DIY standards anyway. By gearing it down you could still use small motors and the pulley system makes the weight distributed over multiple belt strands. That way it can carry a lot of load compared to one belt without stretching or causing unwanted movement so much when milling.

A Kevlar cored belt is a must if it bends a lot. Though the way it bends can make a huge difference to the life and for no obvious reason. But Kevlar stops the need to worry so much so I am sticking with that now.

---

Make your Mendel twice as accurate.
[www.thingiverse.com]

Well cables are way more cheaper than belts, you can get cables from any art stores. To deal with slips there is perhaps two ways to deal with that, for the stepper motor you would just wind the cable multiple times (and perhaps cover the motor with rubberbands for better grip). Second method perhaps is to add high tension springs to the end of each cable as it is attached to the moving platform, this will ensure it remains tight.

e.g. wrapping cables around a pully [www.cnczone.com] for better grip

2.e.g. Tantillus: [reprap.org] "Uses an interesting drive train: low cost wire rope (fishing wire) or High test Braided fishing line instead of costly belts (It wraps 5 times around the rod and then goes through a hole and wraps an additional 5 times resulting in no slip). This is a roll on roll off system with a fixed anchor in the middle."

Edited 2 time(s). Last edit at 05/30/2012 01:12AM by mofosyne.

Hmm, I'll just throw the concept of a differential windlass (aka Chinese Windlass) in here, [en.wikipedia.org]

I can see it giving mechanical advantage at the expense of greater complexity of attachment to the moving part of the axis and having to have some form of return mechanism.

UHMWPE Fibers seem to be well suited for this purpose. Kevlar flexes 2-4% before it breaks. Dyneema for example is available with smooth surfaces ( non coated , these coats tend to wear out ) and has the lowest flex i know off (1-3%<).

I am currently running my 500x750 H-Bot with these. For designs the require a lot of belt/cable its a viable alternative.At lenghts of 3+ meters these timing belts get expensiv. A meter of Dyneema can be sourced for 0.33€ per Meter ( 0.2 mm @ 33kp strenght). In (some) Hbot designs u may also simply tension the cable via motors. You can theoretically indirectly measure the tension by scopeing the current draw of the motor in idle mode.

How 'bout bead-chain?

If you need a continuous loop, gets harder to find, more expensive, and probably doesn't compete with flat belt prices.

But otherwise, you can get it cut to length at a well-stocked hardware store, and I've seen light-duty drive systems that use it.

I have no idea about accuracy, backlash, or whether an effective bead-chain pulley could be printed.

..just food for thought.

mkeveney Wrote:
-------------------------------------------------------
> I have no idea about accuracy, backlash, or
> whether an effective bead-chain pulley could be
> printed.
>
> ..just food for thought.

I've looked into this. Problem is that when you get the tension high enough, the cheap metal that ball-chain is made of starts to stretch and deform.

Hi i want to build a big 3d printer capable of printing cf ,peek , ultem if possible i do try find out what is the best set up from axis movement to frame etc. Here some photos i found about wire instead of belts. I wonder how easy is to calibrate the wire set up and how you keep the wire in tension. The wire has his advantages like is heat resistant etc .I also would if possible can you name me the term they use for that wire set up in the photos i attached ?

Edited 1 time(s). Last edit at 07/19/2016 08:29PM by The Mighty Gorgonion.