Maslow's Weak "Link"

Technical Details
1

So I’m trying to wrap my head around where the biggest weak “link” is in Maslow. It will only be as good as the weakest link so I’m trying to figure out where the most attention needs to go to dial this cnc in. If we find the perfect sled setup will that avail us little since there is something greater causing the error? I understand it all works together/against us but thought I’d break it out to discuss.

Below are some areas I’ve seen:

-Accuracy for measuring/calibrating
-Distance between motors and how rigid that set up is
-Accuracy of the motors and I’m talking about the rotational movement of the gear. A quarter turn of the gear when the sled is at the bottom equates to little movement vs when the sled is at the top. Can the motors turn slow/accurate enough for the desired sled movement at the top?
-Sled design (original vs ring vs triangular linkage, etc) and sled rotational error
-Sled friction against work piece, weight, angle
-Bit feed rate, rpm
-Software (I doubt this is contributing but thought I’d throw it out there)

There is obviously a lot that goes into making this thing more accurate - I’m just curious where the heavy hitters are. Feel free to add more if I missed any.

1 Like
2

please read the write-up I did on the wiki about the accuracy

-Accuracy for measuring/calibrating

This is far more of a problem with the stock sled than with any of the
triangular kinematics designs.

-Distance between motors and how rigid that set up is

This is one major problem.

The stock frame gives the motors/chains a LOT of leverage to bend things, and it
isn’t easy to adjust to keep the chains parallel to the workpiece and attached
to the sled near the balance point.

I believe that for many people, this is one of the biggest sources of errors.

-Accuracy of the motors and I’m talking about the rotational movement of the
gear. A quarter turn of the gear when the sled is at the bottom equates to
little movement vs when the sled is at the top. Can the motors turn
slow/accurate enough for the desired sled movement at the top?

not a significant problem, the motors can turn slowly, and we have an effective
8148 encoder pulses per rotation of the output gear, so even at the top where a
small amount of chain movement results in a more sled movement, we still have
FAR more resultion than needed

-Sled design (original vs ring vs triangular linkage, etc) and sled rotational error

this is the biggest problem, and this goes along with the measuring/calibrating
issues. With the stock kinematics there are just too many variables that are too
hard to measure (see the wiki page)

-Sled friction against work piece, weight, angle

sources of errors, but not big ones.

-Bit feed rate, rpm

These affect quality of the cut, but not a big source of errors (unless the feed
rate gets higher than what gravity can provide)

-Software (I doubt this is contributing but thought I’d throw it out there)

software is not causing the biggest issues, but there ar issues that can be
fixed in software (such as chain sag)

  • Chain Sag

This is something we can fix in software eventually, it’s a significant source
of error near the bottom corners, but right now we have bigger problems

  • Lack of Acceleration Planning

The software currently assumes that the motors start moving instantly to their
final speed, and stop instantly when they get to the end of the cut. This does
not match the real world, but is not currently the biggest problem.

2 Likes
3

Thanks for the wiki link and the response. That’s exactly what I was looking for!

1 Like
4

Depends on how you look at it,

Sometimes it’s good to add a cheap and easy to replace weak link to prevent more expensive parts from breaking.

This is slightly off topic on what you where on about though i still wanted to add that thought.

5

Agreed, as you mentioned I was talking more about the accuracy weak link as opposed to the physical weak link.

Interesting to think if under a huge bind what would break first with this CNC. Possibly a component that involves wood. I guess that’s a different topic though that seems like it’s a non issue.

2 Likes