Maslow Home Maslow Community Garden

Using a Maslow for edge cuts on a full sheet? If so, which frame to use for accurate cutting?

frame

#1

I’m interested in using a MaslowCNC to cut pieces for a Shelter2.0.
The design includes cuts that go right to the edge of a sheet on the flooring.

Is there anything to be aware of in using a MaslowCNC for a project like this?
Is there a Maslow frame design that has good accuracy for edge cuts across a full sheet?

Thanks.


#2

Shelter 2.0 is actually designed to be made using templates! The idea is that you would CNC a few copies of the templates and then you can arm an army of folks with hand held routers to duplicate the templates and build a lot of shelters quickly.


#3

Bar,
Thanks for sharing MaslowCNC, it is a shining example of what open source is all about. THANK YOU!
Yes, we are on the same page on Shelter 2.0. Cutting out the templates requires being able to cut accurately over a full sheet. I’d like to make a template and am thinking about building a Maslow to do it.

In analyzing the design of the standard frame and reading the forums it seems like there is some risk. I’m hoping to have a discussion about what, if any, modifications are needed to the standard frame to get accurate edge cuts over a whole panel.

Thanks.


#4

We are still trying to figure that out :slight_smile:

That said, there are two basic problems

  1. When you get to the edge, the sled is unsupported and can tip. This is a
    smallish problem on the sides, but a big problem along the bottom.

Some people are adding 9" frames around the edge to support the sled.

I think this is overkill and just a couple inches on the side is good enough,
none on the top and quite a bit on the bottom (how much depends on the balance
of your sled)

  1. the machine is suffering accuracy issues in the bottom corners, this is due
    to the fact that the long chain has very little tension on it. This manifests in
    multiple ways

a. chain sag is large (and we don’t know how to compensate for it properly yet)

b. the tension from the bungees can be higher than the tension from the sled,
causing the motor to think it let out a little bit of chain without the chain
actually moving

c. chain is not actually 6.3500 mm per link, it’s a smidge longer, and we don’t
have a way of figuring this out and accounting for it.

Added to these two problems, the fact that there are these variables that we
don’t understand and can’t tweak means that the calibration routine ends up
‘correcting’ the variables that it does know about to incorrect values to try
and make things work.

That said, it depends on how accurate you need it to be. There are a number of
people reporting 1mm accuracy everywhere.

Go with a 12’ top beam instead of a 10’ top beam (requires 13’ chains instead of
11’ chains)

go with counterweights instead of bungee cords

test near the edges and add skirts as needed.

Or, offset the material towards the center, drill a hole at a known place, cut
the part that fits in the ‘good area’, shift the work to the other side, find
you location with that hole in a ‘known place’ and cut the rest of it.

or some combination of the above.

David Lang


#5

There might be ways to make templates that don’t have that requirement. Take a look at this short video where @bar experimented with templates that avoided it (if I remember right…)


#6

@dlang Thanks for the details. They make sense.

On 2b, compensating for chain sag. Is the problem having an accurate, real-time estimate of the length of the catenary caused by the weight of the chain?

Two follow on questions.
Q1: Is there benefit to going even wider and higher on the crossbeam?
Q2: Is rotational alignment of the sheet an issue? I can imagine Maslow having good at relative accuracy but only fair absolute positional accuracy. This seems closely related to a current merge being considered in the code for motors at different Y heights.

Cheers.


#7

@blurfl I love the idea of a story stick. It would require a large investment of time to adapt the design in this case.


#8

@dlang Thanks for the details. They make sense.

On 2b, compensating for chain sag. Is the problem having an accurate,
real-time estimate of the length of the
catenary caused by the weight of
the chain?

yes, we have the case where the low point of the curve is outside the two anchor
points.

Two follow on questions.
Q1: Is there benefit to going even wider and higher on the crossbeam?

there is a point of diminishing returns, play around with the spreadsheet at

it lets you compare two configurations

the key things to look at are the minimum and maximum tension, you want the
minimum tension to be as high as possible, but you need the maximum tension to
not be too high.

The stock machine is pretty much at both of these limits.

Q2: Is rotational alignment of the sheet an issue? I can imagine Maslow
having good at relative accuracy but only fair absolute positional accuracy.
This seems closely related to a current merge being considered in the code for
motors at different Y heights.

The sheets are not all uniform size, and don’t all have perfectly parallel
sizes.

My answer to rotational alignment of the sheet is to make a pass with the router
to cut the ‘bottom’ so that it’s perfectly aligned with the motors (it’s not
uncommon for other CNC machines to make a pass to cut the wasteboard and
reference surfaces so that they are perfectly square with the mechanism)

David Lang


#9

at the risk of getting on another soap box (a pile of them gets pretty rickety
:wink: )

watching lots of woodworking and contruction shows, and tinkering for years, the
story stick approach is FAR more reliable than measuring. As much as possible
you want to start from a known point, make things flush, and then either
re-use the same mark, or ideally push the next piece against the same surface on
your reference. Then you move your reference stick to the next point and repeat.

We are good at making things the same when they are against each other, and good
when referencing the same thing again and again.

But we are bad at measuring, especially measuring from point A to point B then
from point B to point C, then C to D… By the time you get through a few of
these D in one place is not going to line up with D in the next.

but do it with spacers, no problem

measure from A to B, then A to C, then A to D… is FAR better than a-B B-C C-D
but not nearly as good as spacers


#10

In that case, You can make a difference for accuracy in the problematic sections ofd the workarea by considering the effects of travel direction when laying out the lines to be cut.
There was a good series of posts documenting this earlier this year, but I haven’t been able to find them to provide a link. I hope someone from the community can remember…

If I remember, though, the ideas were generally to arrange horizontal cuts so they would be pulling toward the most distant motor, and to generally favor upward direction for vertical cuts. For example a rectangle in the lower left area would be made starting from the lower left corner of the figure and cut the bottom and right sides (further optimization would be to arrive at the starting point from farther left) and then return to the starting point and cut the left side and finally the top edge. There was also some consideration of the right-ward pull created by the router bit causing the sled to climb, I’m not remembering the details of that.

I guess that represents a large investment of time as well. and longer cutting times to boot. There are community members that have found the accuracy adequate to cut the sheet-sized parts for ‘stitch-and-glue’ boat projects, and “…perfect is the enemy of (good/done)…” might be the best answer.


#11

How close to the edges can you–in general–reasonably go before inaccuracy becomes a problem?


#12

There are two separate problems

  1. when you get too close to the sides or bottom, the sled will start to tip.

We’ve had people report that they can get within 2 in of the side before this is
a problem, but at the bottom it’s worse because of where the bricks are (and
need to be)

The solution to this is the same solution you would use if you were trying to
use the router hand-held close to the edge, put some material of the same
thickness as your workpiece adjacent to it to provide support for the router.

This additional support has gained the name ‘skirts’ in the formum here, and
there are many people building frames that have 9" wide skirts all around the
workpiece. I think that’s overkill, you need just a few inches on the side and
6-8 inches at the bottom.

On the bottom, I would make a L from a couple pieces of wood ( plywood or
lumber, 6+" wide, one ~4’ long to fit between the legs, the other 8’ long to go
across the entire workpiece), put blocks on the inside of this, and cut slots in
the short piece so that you can bolt it to the bottom of the lower crossmember
and slide it in and out to match the thickness of your workpiece + wasteboard.

for the sides, it’s a bit more work, but a similar idea, you want something you
can adjust to the thickness of your workpiece.

note that you are someday going to cut into these, so they will need to be
replaced at some point, just like your wasteboard.

David Lang


#13

sorry, I didn’t get into the second problem.

The second problem is that as the tension gets low in the bottom corners, the
sled doesn’t move as well. This isn’t a matter of ‘past this point it gets bad’
but rather that is just gets gradually wose, so the cutoff point depends on how
accurate you need it to be and how well your sled slides on your workpiece (it
slides much better on Oak plywood than on OSB as a couple of extremes)

making your top beam (and chains) longer helps with this, but you have to test
and see what works for you and your conditions.

David Lang