Tensioning system to improve performance in bottom corners

This is a new thread to carry on the discussion of the new system proposed below, rather than keeping it in the original post that was made to share/discuss a router sled project.
Thanks @blurfl, @clintloggins, @dlang, @Keith, @ScrumdyBum, @bar for your advice and encouragement thus far.


I use 3 of the 4 pulleys on the setup. I bought 2 double-pulleys just in case I needed to add another reduction to the setup, but the 1:4 ratio seems to be about right. Any more, and I would worry about the force on the cables that attach to the chains – they are already experiencing 4x the force that the bungee cable applies to the sled.

1 Like

@dlang: I thought about just one weight, but that will pull the sled to the side towards
the weight

do you think that this would be the case using a traveller as was suggested? I can see where the traveller/trolley would want to move to the weighted side, but the tendency of the weight to want to shorten the cord should work against that, no?


There is one idea that may be related, or contribute to this. There was an idea to create a circular slack chain holder. The idea is it would be a circular spool that the slack chain wrapped around. It would be tensioned using the shock-cord wrapped around a connected pulley, to provide rotational torque to pull the chain around. This is related, because the radius of the shock-cord spool can be any multiple of the radius of the slack-chain spool. This means the motion-ratio can be any number, not just a rational number, that is the fraction of slack-chain lengths to shock-cord lengths.

Additionally, one alternative is to use a longer top-beam, that is raised up slightly. By increasing these dimensions, the performance at the bottom corners can be substantially improved.


This has been discussed extensively in the frame design threads. There is a tradeoff between lengthening the top beam and accuracy at the top of the work area, and on the other hand, increasing the height of the top beam and the accuracy at the lower corners of the work area. Doing both will increase the accurate area available to you, but at the cost of a larger machine.

See @Gero’s post here:

And that thread in general.


yes, but the fact that the traveler has tension pulling it in one direction will be there.

Test it, I may be wrong, but I expect to find that the traveler will move fairly well towards a motor that is retracting the chain, but not well away from a motor that’s letting out the chain.

i.e., in the bottom left, the traveler would move right easily, but as things need to move left, I don’t expect it to move left well.

I suggested this at the tail end of the frame design timeframe. As far as I know, nobody has tried it.
Bar’s approach of putting the slack across the center of the top beam is FAR simpler, so I would just do that.

to be nitpicky, it doesn’t reduce your accuracy at the top of the workspace, but it does increase the force needed at the top center, so if you have a heavy sled or marginal power supply, you may have the motors not keep up.

raising the bar helps this, but hurts the bottom corners. I think that raising it slightly will help the top center more than it hurts the bottom corners, but test to see if you have a problem first.

Extending the top beam is by FAR the simplist way to increase the accuracy in the bottom corners. Going from a 10’ beam to a 12’ beam doubles the force that gravity can apply to swing the sled into the corner.

Edited to add:

remember that the maslow kit is an exercise is ‘just barely good enough, make it as cheap as possible’, so it only includes enough chain for the stock design. To go to a 12’ top beam will require going from the stock 11’ chains to ~12.5’ chains (either buying new chains, or getting a short length of chain and adding it to each side with the master links that were included with the kit.)


Thanks for all of the above posts :slight_smile: I’ve recently had great success with increasing weight on a middle pulley system, but couldnt figure out why the maslow seemed to cut a little off nearest the bottom left corner. Yesterday I cut on the bottom right - and had a similar issue. I assume this is related to the discussion here - not enough tension to further point away from motors. To confirm, I should either consider a 12 foot header or add tension to pull sled towards bottom corners?

EDIT: 12 foot headers require you buy new 12.5 foot chains. Otherwise you cannot use calibration step to measure distance between motors, and you cannot attach to sled.

how far off was it?

you can try re-doing the calibration again (I’m assuming you are using
triangular kinematics, not the original angle brackets) and see if you can
improve it by more careful measurements.

also, make sure you are on the latest version of ground control and firmware.

going to a 12’ header will help

adding tension to pull the sled towards the bottom corner may help or may hurt.
the right amount will help, too much will overcorrect for the error.

@dlang I created this for testing.

  1. Drill operation going through the entire project.
  2. Square in the bottom left corner going .25 depth counterclockwise
  3. Same profile operation going clockwise at .0625 depth.
    My thought on using a shallower depth was to hopefully take off as little as possible so that the bit doesn’t just follow the previous groove.

My plan is to run this on with my current configuration at 5 degrees. I don’t have an easy solution for adjusting sled weight yet so if the drill operation lifts the sled I will halt and come up with an adjustable weighted sled. If I do not detect the sled lifting from the surface, I will move on to the L operation. If the L operation comes out clean in both directions I will do a full calibration and do a benchmark test. If the L does not come out clean I will halt and come up with an adjustable weighted sled. Once I have adjustable weights, I will restart the process.

1 Like

Not sure what bit size this is for, but a shallower cutting depth (1/2 the diameter) would reduce the effect of the bit’s sharpness and rotation rate from the test. OTOH maybe that’s part of what the test is about?

I chose .25 because that is generally the depth I go on each pass with a .25 inch bit so I wanted a real world test. I understand that it can dull the bit prematurely but I justify it by saying my time is worth more than the bit. :grin:

1 Like

From the looks of it, it is a limitation in the ability to control the position of the sled. It looks like some cuts wandered off track, not like the part is uniformly skewed. I guess this is not due to calibration. A 12’ top beam (along with slightly higher motor mount) or this tension system would likely fix the problem.

the things I miss by using text-only e-mail to reply to things. I missed the
pictures that ChuckC posted

do you know what direction the router was moving in the problem spots?

I would bet that in the bottom right it was moving counter-clockwise. those
rounded edges are exactly the symptoms that you tend to see when there isn’t
enough sideways force to move the sled

I’d also guess (with less confidence) that the bottom left was also being cut
counter-clockwise. that looks like the same sort of error where on one cut there
wasn’t enough force to move the sled sideways due to the chain angles.

going to a 12’ top beam andor finding a way to reduce friction (waxing the sled,
tilting closer to vertical, etc) would help

David Lang

P.S. the notch in the bottom right corner shows that the bit was easily able to
cut through the full depth of the workpiece in a single cut.

I dont know why I come up with a plan cuz I never follow it.

This was cut at 10 degrees 20 lb original sled weight with bricks.

  • Left bottom corner
  • Deep cut is counter clockwise
  • Shallow cut is clockwise
  • Did not lift sled on drill operation
  • 30 ipm

I will note that OSB is one of the worst materials to cut because it’s got such a rough surface, the friction is high

you don’t even need a full profile, this is a simple enough thing that you could manually write the g-code for it :slight_smile: I was not imagining a separate drill option, just the drilling at the start of the cut.

so what I am seeing is that the counter-clockwise cut (moving down on the left side) is very poor because the sled is not moving smoothly, so the chain goes slack, the sled wanders to the side and then drops

I have a hard time seeing the shallow cut, but it looks like it may be a bit rounded in the bottom left corner. but given the roughness of the OSB, that doesn’t really surprise me.

give it a try at 5 degrees and see if it works any better.

1 Like

I know this isn’t a real solution, but I did always wonder if there was a way to cut profiles counter-clockwise in the bottom right corners and clockwise in the left corners. It always seems to work better like that on my cuts. Sadly, it would have to be done in CAM, so probably no real way for the maslow to do it locally.


there is no direction that’s always best.

In this example, it looks like there were different problems in each direction, both due to the sled sticking.

the ragged down cut due to the sled sticking on the counter-clockwise cut.

the rounded corner on the transition from left to up due to the sled sticking on the clockwise cut

Which cut was first? If the shallow cut was second, it benefitted by the reduced drag from the previous deeper cut.