Thinking outside the box

So I saw this on Facebook

Looks similar to a Maslow, but if you look closely on some of the shots, it looks like it runs on wires instead of chains, and the motors that control it are in the sled itself and it winds up the wire. Has anyone experimented with attaching the motors to the sled instead of the frame? For some reason I feel like that might be something that could work well because it will change the way you need thension on the slack side of the chains (but I could be totally wrong) Anyone have any thoughts on that?

1 Like

The video appears to be of a Scribit:

I was thinking about that as well, if you use a raspberry pi as a headless controller, you could put everything on the sled.
Though the problem with a cable on a spool is that one rotation on the first layer takes up a smaller length of rope then a full rotation on the second one, and the problem gets bigger the further up the wheel you get.
Also the forces of a sharpie in a sled are constant, so rope stretch isn’t really an issue. But a spinning router can give a much more dynamic and unpredictable force, which might affect accuracy.

I was also thinking about the spool issue and wondered if a drive spool (with a couple of turns for grip) could then pass the cable off to another spool, or even just allow it to hang free.

At maker faire someone suggested that we look into something called “synchromesh cable” which looks promising


Looks cool, I found this data sheet, and one thing that sticks out to me is what is listed as “. Cumulative Pitch Error ± .16 in over 100 pitches” That seems high, comes out to about 3/4in over 11ft. What is the tolerance of roller chain? I found .25% which comes out to .3in


I like the idea, is the spiral fixed to the steel cable? I would assume that over a long distance, the spiral would unwind a bit, thereby changing the pitch and length.
If i had the time to experiment with this, I would. But if i did have the time i would have assembled my maslow by now. :smiley:


Continuing to think, can you separate the encoder and the motor? If the cable is a conductor you could still put the motors on the beam, provide power to the cable, and have a small pulley on the sled as encoder. Probably needs a picture to explain.


That’s a great idea!

Here’s another approach to that. Separate drive chains and position-sensing cables could both feed from the upper corners, with the motors and chains located as at present but with separate position-sense cables driving encoders not attached to the motors. I think having the encoders located at the upper corners near the motors would make sense, keeping them away from the electrical noise of the router.

The drive chains and sense cables could all terminate on the ring travelers as the present setup. This would also separate the motors from the encoders to make different motor/sprocket combinations more workable. Using light weight cables for position sensing with just enough tension to limit sag but avoid stretch could solve the chain stretch and sag issues.

Calibrating the position could be done by detaching the position cables From the sled travelers and letting them retract to their fully-rewound positions, setting that as a reference, then pulling them down to attach to the sled. The encoders would measure the cables as they were pulled out to attach to the sled.


I started my train of thought with the “let’s place everything on the sled” idea, that’s why i was thinking of the pulleys on the sled.

1 Like

In the spirit of separating the motor and encoder, I had wondered if keeping the chains, and using a laser measuring tape at each motor point to measure the distance to the sled might work - I had seen some claiming 1/16th of an inch accuracy.

Add a couple of Calibration Structures, like a Straight Edge, on the bottom of the frame,and on one side of the frame, so that they are at a measurable 90 degrees to each other.

Then mount a couple of standard laser/laser-detector modules. They can be controlled by the microprocessor using the standard libraries.

You’d need some way of protecting-the-modules-from-vibration/software-compensation-for-the-vibration, but there are existing technologies for this, that fell out of patent years ago.

It was SOTA twenty years ago, and it still works to those specs, so it’s ideal for an Open-Source HardWare Design.

For my part, I’d like to see much tighter accuracy than that.


In the early days (before the kickstarter kits came out) there was someone who
build a maslow-like thing using motors mounted to the sled, with chain guides
to keep the chains meshed with the gears.

This was long before we started doing any triangular kinematics

1 Like

Continuing to think, can you separate the encoder and the motor?

yes, but can you get an accurate enough encoder without the gear ratio to help.
I’ve suggested that optical mouse technology could be used.

If the cable is a conductor you could still put the motors on the beam,
provide pover to the cable, and have a small pulley on the sled as encoder.
Probably needs a picture to explain.

Yes it needs a picture, but keep in mind that you only have two cables, so only
two conductors, you already need 6 to the Z motor and another couple to the
spindle motor (more if you want Z zeroing sensors or spindle speed control)

David Lang

the current system gives us an accuracy of better than 1/1000th of an inch.

David Lang

1 Like

I was just thinking out loud, but if you use a raspberry pi as controller, you can move all the electronics to the sled, and the only external things are the motors that the cables run to. I was talking about the cables with the spiral around them that Bar posted, they look to be 2 conductors each.
You’ll still need a power cable to the sled, but you can tape that to the dust collector hose.

You got a point about the resolution of the encoder though.

I’ll make a fancy ms paint picture if i ever get in front of a computer.

But not evenly across the workspace which this might do, and this might maintain accuracy over much longer distances

Not that I am arguing that it is the answer for right now…

assuming that you can keep the lasers pointed at the same point (i.e. the sled never tilts)

The encoders do a pretty good job of being consistant across the workpiece, the errors are mostly in other dimensions (now that we added chain sag and are adding chain tolerance)