Beyond the green

[Gijs]

Hello here is an update of my experiment to go outside the lines…
My setup was like this:

It is a big table in the middle of my shed and the anchors are on the walls.
Since my table is longer on the picture left side I was gonna test outside the calibrated area there, but I would not be able to go to the corner very far…
So I decided to extend the table for this experiment and shifted my waste boards towards one corner. From previous test I already found out that while maybe not completely accurate mm wise, the Maslow does seem to be very symetrical in its deviations from the expected. So I thought this time I’de only test on corner see how the Maslow behaves outside the lines(see here for the previous measurements: Having trouble after calibration - #56 by Gijs )

So here are the results, Is is all done in the top left corner from the Maslows point of view.
The diagonal cross or X is the center of the normal calibration area. I calibrated at 1600x1000 so this one quadrant (top left) is 800 by 500 in the pictures from the X (0,0) to the white tape (about -800,500)
Within the calibrated area the lines seem to go as planned. Although it does start to get a bit wobbly on the side close to the white (but still in the green).
Outside the lines it starts to get really wobbly but the overall dimensions still seem to stay pretty good.
The wobbly is directional. What I mean is if for example i move from -900,0 to -900,500 the sled wobbles side to side and also wants to lift up on the bottom side.
Doing the same move at x=-1150 the sled actually lifts up several inches and slams down again. This might have to do with the board i am using which is pretty rough, but mostly it is from the weird forces so close to the corner.
Going the same path back down (from y=500 to y= 0 at any X value outside the green) the sled has no problem and makes a straight line. You can see in the pictures some lines are wobbly and straight, the wobble was from going up, the straight line from going down on the same line. Here also the dimensions and straightness are not bad when the wobblies are not included…
Anyway, I hope it helps Oh yeah, the red tape is where the red area starts the long white tape is where the white area starts, inside the white tape is the green area but I do not have green tape.
Overview (picture taken from right middle from the Maslows perspective):

Calibrated area 1/4 top left:

Beyond the calibrated area:

Detail wobblies:

A video of the worst wobblies:

[dlang]

that wobbling is not the type of error I was mapping the different zones for.

the wobbling is the sled not sliding, too much friction against the workpiece (waxing the bottom of the sled, smoother workpiece could help

one thing that could be related is that as you get outside the green area, the belts are not reaching as far as the maslow expects, some of this will be seen in the lines warping (pincushion like effect, I would expect a slight distortion towards the center in the white area and the edge in the red area) but some of it will be seen as the tension on the belts being higher (@bar @ronlawrence3 it would be interesting if we could setup a debug mode that reported the motor currents so we could see how much harder the motors are pulling in these areas), and with the belts anchored at the level of the wasteboard, higher tension will result in more down-force on the sled, which will result in higher friction.

the other thing that could be done is to add weights around the edge of the sled. the added weight will add more friction, but it will also resist tipping

Thanks again for doing these tests.

[Gijs]

I thought of using weights too but I think it will make it worse. I need a better waste board, but since it is a waste board I used crap I had lying around… The sled is nicely scratched on the bottom so the wax should hold
I have not done measurements yet, I did the cutting last week but I had a hard week because of a jaw operation, so it took a while to even take the pictures of the results.
The side to side wobbles/ curvy lines I think are also due to too much sled friction and then getting in a resonance wave pattern. The sled friction is due too the waste board material, yes, but in combination with the belts pulling down, because in the middle there is no friction problem. I think going too far out of the calibrated zone can build up belt tension too high.
Anyway I did it this way, with the anchors still low (at the level of a standard frame).
My next test will be with the anchors at the individual arms level and the software adapted as we spoke of before, to see what the difference will be. I think I will leave the waxing for now to make the comparison fair. And I should measure the results to see how the distortion is compared to the next test with horizontal belts.

[dlang]

It would be fantastic if you could end up testing all 4 combos

waxed vs unwaxed
anchors low vs anchors level

If it would help to throw a few bucks your way to pay for wasteboard material,
let us know and I know I will, and probably others will as well.

weights would be an interesting thing to try, they will add friction, but as it
starts to tip, they will tend to keep the high end down, it’s hard to say if it
would help or hurt

David Lang

[Gijs]

I know, but once waxed I cannot unwax it… I might try though. And I might use Silbergleit instead of wax.

[dlang]

Gijs wrote:

I know, but once waxed I cannot unwax it… I might try though. And I might use Silbergleit instead of wax.

I was thinking add a 2nd anchor (either all around or on the top left location
that you are pulling towards) and trying that. then wax and try both levels of
anchor.

David Lang

[Gijs]

@dlang

add a 2nd anchor (either all around

Yes, I think I will do that.
I will also have to save all the yaml files and such to get back to this setting, because I will have to re-calibrate when the anchor points change…
Will be quite an operation… Maybe this weekend…

[dlang]

no rush, again thanks for doing this research.

David Lang

[WeldingRod]

One of your challenges is that with the very shallow angle between two belts, you get very little force that direction, and poor control.
Rotating your table 45 degrees may get you a happier test pattern due to better belt angles.

[Gijs]

Yes but that would defeat the point. I am purposely going where the machine should get into difficulties to see what actually happens.

[James_Presnail]

I’m not good at enough at designing a 3d object… But I wonder if some ‘plugs’ into the nut/bolt hexagon holes on under side of the sled… Could be ‘plugged’ with hexagon shaped plugs that have a roller ball in them… To get rid of some friction for certain jobs.

Fidget Roller Ball by KingOfTheTramps

Something like that?

[Gijs]

Love the idea. Wonder if the holes are big enough for a “system” like that but it is a really smart idea. If they are not big enough one can always make more holes They might leave more traces on the work piece if using soft wood.

[dlang]

James Presnail wrote:

I’m not good at enough at designing a 3d object… But I wonder if some
‘plugs’ into the nut/bolt hexagon holes on under side of the sled… Could be
‘plugged’ with hexagon shaped plugs that have a roller ball in them… To get
rid of some friction for certain jobs.

my guess is that the balls would be so small that they would not roll. But I
also don’t think you need balls, just a smooth/waxed surface is worlds better
than the stock sled

or slippery tape (which will need to be replaced over time)

David Lang

[geo]

@WeldingRod Can you please explain this and/or point me to a thread that does?
I am i the process of designing my frame. I have plenty of room. And will do vertically with top mount t wall or ceiling via hinges and the bottom will be pulled out for an ideal angle (15 degrees r more).

My issue is I do not know what the perfect frame measurement from the 4 angle points should be for cutting a 8’ x 4’ sheet edge to edge.

I have read the threads and checked the various calculators. I assume you want to strive for the entire 8x4 sheet in the green when using the above referenced. But some mentioned the belt angle but I do not under this 100% yet.
Which belts are compared and what is the happy range angle (or the extremes to avoid)?

Thank you!
Geo

[dlang]

Geo wrote:

I have read the threads and checked the various calculators. I assume you want to strive for the entire 8x4 sheet in the green when using the above referenced. But some mentioned the belt angle but I do not under this 100% yet.
Which belts are compared and what is the happy range angle (or the extremes to avoid)?

So what the green/white/red areas in my calculator are all about is the belt
angles.

background:

The arms can pivot on the router, but at some point they hit the uprights. The
entire sled can then rotate until a 2nd arm also hits an upright, at that point
you don’t stop being able to move, but the belt no longer makes a straight line
between the anchor and the center of the bit. This makes the distance from the
anchor to the center of the bit be a little less than the maslow thinks it is.
This increases the tension on all the belts, and makes the sled position be just
a little off.

From the earlier theoretical tests (see
Maslow 4 frame size checker - #77 by arjenschoneveld ) the error
starts out small ( 0.5mm at 130mm beyond the green) but grows non-linearly
(2.5mm 400m beyond the green). This is a theoretical calculation that does not
take into account the fact that the belt doesn’t translate directly into error,
it translates into increased tension and belt stretch (and longer belts can
stretch more than shorter ones)

according the Bar’s design, the arms get blocked when they get to within 20
degrees of the verticals (i.e. no two arms can get closer than 40 degrees to
each other, or furhter than 140 degrees from each other), my view of cad gives
it just a little more, and I have a version showing what I think would happen if
you removed the verticals around the leadscrews and clipped the top clamp to
avoid problems. That is the ‘angles’ pulldown on my calculator. As you can see,
the wider the angles, the bigger the green area.

How this applies:

When you get close to the edge, the belts to the two closest anchors get far
apart. The red area is where those belts will both hit a vertical.

When you get near a corner, the two belts to the corners of the opposite
diagonal both run into the verticals (so as you are nearing the top left corner,
the bottom left belt and top right belt get to their limit). That defines the
edge of the white area of the graph.

So in the green area, the angles all work and you should have no problems
(although if you get to really small frames, the Z angle and diameter of the
sled start to be a problem, the theoretical green area on a 20" square frame is
a lot larger than the 4" square that you can cut and still have the 16" sled
clear the side of the frame, and that close to the anchors the Z distance to the
anchors needs to be small)

Now, ALL OF THE ABOVE IS THEORETICAL, the default frame that @bar built does not
work according to theory, but he was able to cut out the world map without
noticing big problems. It’s all going to depend on what your accuracy
requirements are, and that can vary from project to project.

The testing being done in this thread is an attempt to see how much of this
theoretical error translates into practical distortion.

Does this (long) explination help? At this point all we can say is you want to
keep as close to the green area as you can, and we have seen a number of people
propose frame layouts that are longer than is useful, resulting in the top and
bottom red areas reaching deeper in, while the green area is longer than the
workpiece they are planning for. In those cases, moving anchors in from the ends
can actually create a more usable machine.

For example, compare a 10x8 (default size) from with a 12x8 frame, and then look
at a 11x8 frame which seems to be better than both

David Lang

[WeldingRod]

Geo,
I’m really not an expert (yet)
This is a slippery subject, as there isn’t a “perfect” number out there. The wooden frame anchor dimensions of 95" x 116" seem like a really safe bet for 4’x8’ work area.

When you head toward one of the sides of the box defined by the anchor points, the two long belts can do a good job of controlling their directions. However, as the angle between the short belts gets near 180 degrees, they have very little ability to pull the sled toward their side anymore. At 180 you have zero force heading for the edge; all of it ends up just pulling between the two short belts!

[geo]

@dlang Perfect. Thank you so much taking the time to explain this. Very helpful.

It was clear Green=Good. Just wasn’t sure what made it green. lol So than you!

Yes I did quickly realize 11x8 looks better then 10x8, 12x8 (I made that exact comment at the end of the Thread you tagged).

I’m leaning towards something like 17ft x 10.5ft or 14ft x 11.6ft.
Theoretically which of the above do you think will be better?
My guess is the 14 x 11.6 as 1ft shorter belts.
IF cable stretch comes into play?

Wonder if anyone made a frame that big and their results?

Thanks again!
George

[dlang]

Geo wrote:

I’m leaning towards something like 17ft x 10.5ft or 14ft x 11.6ft.
Theoretically which of the above do you think will be better?

I suspect that the more square one will be slightly better than the long, thin
one, but there are no numbers to back that up.

David Lang

[jwolter]

@geo

I’m leaning towards something like 17ft x 10.5ft or 14ft x 11.6ft.
Theoretically which of the above do you think will be better?

I made mine 14’2" x 12’4" (4320 x 3770 mm) for a few reasons:

1. I wanted to make it big enough that if I had a work piece that was a little larger than 4’x8’ I could still cut it.
2. I wanted to be sure to be able to reach the corners of the workpiece,
3. but I also wanted to leave a little belt on the spools even at the corners because there has been a report of the M4 pulling the belt from the spool if you expose the belt/spool connection to direct tension.

I haven’t done any calculations, but I think I would lean toward the squarer anchor pattern as well.

[geo]

@dlang

Is there more information on anyone whom “clipped the ears” as you mentioned?
I did some searching and couldn’t find it.

@dlang

I have a version showing what I think would happen if
you removed the verticals around the leadscrews and clipped the top clamp to
avoid problems.

Thank you sir!
Geo