Tensioning system to improve performance in bottom corners

@aluminumwelder The Green Lean CNC machine has some similarities to the Maslow and uses a constant force spring.

  • It is vertically orientated with 4x8 work area,
  • Uses roller chain to drive the X and Y axises.
  • Frame is made mostly from wood, that comes in a kit

But the similarities end there as it is more of a traditional gantry CNC X, Y table and uses a constant force spring to counteract the force of gravity on the Z axis and spindle. It’s also over 10x the cost of the Maslow kit.:moneybag:

1 Like

Don’t immediately dismiss this idea because it’s impractical. I’m trying to take the conversation in a different direction. I’ve got the start of an idea, I want to see if anyone can make something of this concept. Use the idea but move the pieces around to make it practical.

What we need is something that acts like a magnet or gravity on both right and left side of the Maslow. Something that pulls more as the sled gets closer and pulls less as the sled moves away. The opposite of a spring. I’m thinking about using gravity itself. Using a board that rotates about pinned joint and connects to the router sled with some rollers (or similar). As the sled moves away from the center line, the board creates a moment about the pivot pin and applies force to the sled.

On the center line the board is balanced. The resultant force on the sled is 0 because it’s all resting on the pin.

At the top left, the resultant force is 45deg.

At the bottom left, the resultant is almost all in the Y but with some X.

This design obviously wouldn’t be practical as-is. If it could be relocated to hang off the chains somehow on the backside, and we could flip X and Y components at the bottom left, then we might have something to work with. Any thoughts?


I think a more practical solution would be a lower beam with additional 2 motors. Chains would then pull on 4 sides of the sled.
However, this would mean an additional board too and adapting the software too.


I too have been thinking about that. It seems to me that we have all the software that we need from the top view, now we just need to modify it from the bottom view. I suppose there will be some additional requirements to make them all work in harmony, but that surely can be done. This has the potential to fix most all of the issues.

I think a more practical solution would be a lower beam with additional 2 motors. Chains would then pull on 4 sides of the sled.
However, this would mean an additional board too and adapting the software too.

I actually ordered parts to do this, although to allow for a bit of inaccuracy and undue tension on the motors, I was going to do a single motor on the bottom driving a chain. A single attachment point with a tensioner to the sled bottom. This could then be pulled either straight down, or the carrier on the chain could be moved left or right of the sled location. There is a pin that could be monitored on the stock board for sled location I was told and a second board could then adjust the bottom carrier location.

I think that the Z-height-display gets the z axis location from that same data source…

1 Like

@eppic: I’m not sure if I understand correctly, but if you want to have a tensioner to give the needed pull in the inaccurate zones at the bottom, it should be quite short. However, won’t this give too much tension if the sled is cutting at the top?

@EdFeeney: of course it can be done. But it is not a small development. Getting them to operate in sync is one thing. Another thing is the calibration process which will be a bit harder and should be reprogrammed too. Last point: it will take a bit more processing power ofcourse. To be checked if everything still works smoothly.

I am always surprised at the creative solutions people come up with. I give it a thumbs up.


I don’t know how perfect the lower motors would have to control. If the top two motors control 99% of the positioning, and the lower two motors could perhaps be not a chain, but a bungee cord, or something like that, which simply keeps the top chains tight. That means that the programming doesn’t have to be that perfect either. Just a thought.


To really “fix” the Maslow, you would need to have three extra motors, one in each lower corner to pull outward in the lower corners and another to pull up in the top center.


That’s true. But you need tension on the bottom of the board, where the bungee will be the shortest. And the tension on the top of the board should not be too much. Unfortunately, that’s again opposite to Hooke’s law. You could try to use constant force springs, however “constant” is relative. Certainly when they are short, the tension is still lineare not constant…
I will test this one, since we only need limited hardware :slight_smile: It might be a quick and dirty fix.

I’m only getting the location as a feedback from the controller so there might be a delay from the actual command to the output of the system. I don’t know if that delay is too big for this purpose?

The data stream contains the present location of the sled, updated 5 times a second. As the sled can’t move very far between updates, I figure that this should be ‘close enough’ for a tensioning controller.


I like the creativity of you approach for what it’s worth.

1 Like

Thanks for that info! I wondered why I had so many printouts all the time :slight_smile:

I think if the ring was 360, and you were pulling on it instead of fixed sled locations, that you would have great results.

For your original design up top I mean


I was looking at my Maslow the other day and was thinking about just adding a simple spring tension on the y-axis. The thing is that the tension have to be opposite the sled position. Think of a rail on the back side of the frame. The back sled would move opposite the front sled and be connected using the same springy rope as the chain tensioner. Having a small arm wrap around from back to front will not increase the machine envelope but the tension should create a force the will ensure tension on the longest chain.
Yes, this will require a small MCU to listen for the position from the Maslow… I might have a bias toward that :wink:


Please don’t jump down my throat if this has already been answered but I have two questions.

  1. If you turned the whole thing by 90 degrees and had 4X8 cutting area instead of the 8X4 that you have now would that help with having a larger area of usable cutting space?

  2. What if you the motors had a higher torque and you you increased the weight of the sled. Would that not help take the slack out of the chain?

Please excuse my ignorance…


  1. That would allow you to make a longer piece technically, but you’d have quite a narrow usable space and potentially no usable space towards the bottom since the chain angles would be so slight.

  2. You’ll need a stronger motor driver as well. The chain sag isn’t really a problem as much as a symptom of the basic operating mode of the machine. More weight does increase the vector in the direction of desired movement, but too much and the motors will either stall, fail, or you will burn up your driver board.

No worries! We all started fairly recently on this same machine, so no one has room for too much criticism here :slight_smile: Do you have one built to test your ideas?


depending on how far the motors are from each other. You are correct if you
tried to use a 6’ top beam on a 4’ wide workpiece.

But if you go to an 8’ or 10’ wide beam, things look much better.

David Lang

1 Like