In a normal chain driven CNC machine, chain stretch and sag don’t matter, you
apply tension appropriately and this takes up the stretch (you worry about the
movement of the chain, not it’s length) and the sag is consistant because the
tension is consistant.
In our case, we are depending on the lenght of the chain and the sag under
tension variations of 10:1
This isn’t how chains are intended to be used, and this makes the problems
fairly unique
https://scribit.design/
one way to use cables is to use really thin ones (appear to be fishing line?) on a drawing robot where accuracy is probably not really needed.
Another possible way to use cables would be to have a spiral take up spool so that one rotation is always the same amount of cable discharged.
Will work on the self tensioning and anti sag chain mentioned before. Dual drum cable driven is an interesting prospect.
I notice Folks keep brainstorming ideas to make cable work instead of chain.
Could someone help me understand what the advantage would be?
Is there a problem that cable would fix?
It’s easier to deal with slack if it can be rolled up on a spool. It’d facilitate allowing motors to be installed on the sled.
Why not roll up the slack chain then?
Roller Chain does not Roll well 
Seriously - It needs to stay in one narrow stack to not pry the links to the side - this would become a very large diameter, which would be hard to work with.
If you make it “tip off” to become a wider roll, it makes it difficult to manage the reliable on-spooling, and off-spooling without applying lateral forces to the nearby sprockets.
Spooling roller-chain is doable, but not convenient at all.
You might be able to let it pass through the sled, and let it fall to the floor, but I think that defeats the objective of a tidy, self-contained maslow-in-a-sled.
@dlang came up with an idea:
However, if you mounted those to the sled, it’d look like Mickey Mouse.
And fit with the Disney characters cut in an earlier topic. Not commenting on the technical merits, but I like it 
The mooseshop needs a garage sale and bonfire to get to the overly crowded stage, swamprats are going berserk they’re so happy
you really don’t want to have chain just fall uncontrolled, it will tangle up
and when you try to go the other way, you will have trouble.
David Lang
I’m looking into using two modified winches to offer a cable driven option.
I recently replaced a cable driven power window regulator in my vehicle. If you haven’t seen these before, it uses a Bowden cable similar to a bike brake cable and sheath. The inner steel cable is wrapped a couple times around a drum. The drum is driven by a worm gear motor. The whole thing is nice and enclosed behind a steel cover and the cable does not ever slip on the drum. They have strong mounting ears molded in. Normally, this cable drive moves a window lift linearly on its track over a distance about 16". Obviously we need a longer cable and some different geometry, but that is also neatly solvable. We just need to find an auto parts OEM supplier willing to modify an existing design slightly and sell to us.
I think these drives would be a good choice for a Maslow variant. The worm drive motors and Bowden cable drives are well proven mass produced strong 12v motors at ~3amps. The cable sheath on a pivoting stop would eliminate the need for a sprocket or pulley and the attendant chain wrap error. We would have sharply defined corners for triangular kinematics, and the cable angle could be directly sensed with a pot instead of infered. The only thing missing from the commonly available units is an encoder, but that is a solvable problem.
I wonder if there is an alternating clear to black cable that an optical sensor could read through to determine length ?
That would be an interesting solution but if you used cable resistance along the length is a more consistent measurement. Looking at some 1hp 12v winches to add an encoder on the cable feed out. I have a number of possible avenues mapped out and roughly designed though I don’t have the resources currently to develop them all. Going to put the open source SLM project currently ongoing on hold to work on some of the major accuracy and speed improvements for the Maslow.
Seems like if one increased sled weight and used stronger motors then speed might be able to be increased for long straight cuts. however the down side would be intricate cuts might suffer due to acceleration/ deceleration issues
Sled weight need not increase and encoders on the output will increase the amount of data per rotation while decreasing the effect of the gear box backlash. A rigid cover over the front will add safety and allow for a spring loaded assembly with three omnidirectional wheels to allow for enough force for materials that need more plunge force. Have the flu so will have to talk later.
Since we need very precise control the pattern would need to be very small. I wounder if this could be coupled with the optical sensor out of a computer mouse. The pattern is not very precise, but it wouldn’t drift with time, while the computer mouse sensor gives very small increments of movement, but could drift over time. The combination of the two would be valuable.
Optical sensor plus an etching laser mounted on the sled would allow for a fairly good calibration.