# Musings on a new design

#21

Instead of pinch rollers for the encoder, use a single wrap of the rope around a drum, like a windlass. The motor/gear/winch/storage spool could be more compact, multi-layer. The rope tension around the encoder cylinder would be sufficiently larger than the bearing friction to insure no slip.

You have an engineering challenge/compromise to deal with: As the rope gets thicker and heavier, you will have less stretch and more sag to compensate for. Thinner rope will give you less sag and more stretch. Same for tension: Lower tension means less stretch and more sag, higher tension means more stretch and less sag.
Stretch is a linear function of tension. I didn’t pursue Engin school long enough to know the equations for sag, but I suspect they involve tension as well as angle off the vertical.
Oops, that’s the end of my time allotment. The task of optimizing two simultaneous equations will have to be left for homework…

#22

Minimum bend radius is same as diameter of the rope. 1/8" rope can be spooled on a 1/8" spool.

In addition to use on fishing and tug boats, these ropes are also being used on high-spec racing sail boats to replace heavy stainless steel wire rope rigging. In the construction industry, synthetic ropes made from Spectra and Dyneema are slowly being adopted as wire rope replacements for large cranes, and have been used for years as slings and bridal legs for heavy lifting jobs. Really amazing stuff.

#23

I agree with cables vs chain. This may not be a realistic solution but what about not spooling the extra? If you had two cables that cross in the center, then you’d have less cable to spool and take care of. You would still have to deal with a little extra when the sled was in the center of the work area but this could possibly be taken up with a “dumb” spring spool or something similar. I’m going to think through these measurements live here in my post, this idea might be stupid but we’ll find out in a few minutes…

Assuming symmetry in both x and y directions we can focus on two single attachment points, ap1 and ap2 in my diagram below. And we can cover all measurement extremes if we think of the sled in three distinct positions, P1, P2, and P3 in the diagram.

Now, we are testing between two cable management ideas, for this example lets call them spooled and shared. Where spooled is 4 cables spooling at 4 places on the sled and shared is 2 cables with slack being addressed at 2 places on the sled. For purposes of illustration I will not count any material needed for cable attachment or the distance they will be separated on the sled etc.
Assuming a 4’ x 8’ work area (black rectangle) these are accurate measurements rounded to the nearest 0.5":
A1 = 21.5"
A2 = 53.5"
B1 = 115.5"
B2 = 60.5"
ap1P1 = 75"
ap1ap2 = 150"

And some definitions:
Lower left spool = sp1
Upper right spool = sp2
Take-up amount = ta

The results will be the length of cable that needs to be taken up/spooled at a given point.

Spooled
At P1: sp1 = A2 = 53.5"; sp2 = A2 = 53.5" (there will be A2 (53.5") on each spool )
At P2: sp1 = (A1 + A2 + A2) - B1 = 13"; sp2 = (A1 + A2 + A2) - B2 = 68"
At P3: sp1 = A2 + A2 = 107"; sp2 = A2 - A2 = 0"

Shared
At P1: ta = (B1+B2) - ap1ap2 = 26"
At P2: ta = 0"
At P3: ta = (B1+B2) - ap1ap2 = 26"

(disclaimer: I did this quickly and there’s a good chance I screwed up some simple math.)

It appears that the most cable that will need to be stored (spooled) on a spooled design would be 107".
And the most cable that would need to be stored (who knows how?) on a shared design would be 26". Heck, 26" means a loop only about 13" long, letting it just hang or pool in a cup might be just fine. Plus that max number would be consistent all along the line ap1ap2.

I’m not saying cable management is a deal breaker here but that’s almost 4 times as much spool space needed on each spool for a 4-cable design vs a two cable design (which only needs two take-up points).

I think my idea probably introduces bigger problems that are not worth the savings in space. Like how to attach the motors and move accurately along the cables when you don’t have access to the end of the cable.

Thank you for letting me think that through. Sorry for the mess.

Cool! Are you thinking that since you would now have four accurate measurements and a known sled size (known anchor points) that the need for a ring or linkage is gone? Or are you saying that optical calibration can alleviate the need for triangular kinematics?

#24

I theorize that if the sled can be made to stay vertical by the use of four attachment points (remains to be seen), then yes, I think optical calibration can potentially compensate for quadrilateral kinematics (likely with a higher order polynomial). If not, then if the rotation of the sled is consistent (i.e., it’s always a certain degree from vertical at a given point), then yes, again, I think it can handle it. If not and not, then no, I don’t think it will work

#25

I’m thinking the motor/spool assembly should work something like this.

The four motor assemblies would be mounted to known locations on the sled with the router bit centered between them.

Moving the sled to each corner until the limit switch is hit might be sufficient to calibrate. With string I do not think sag will matter much, stretch would be a concern tho.

Once you know the length of each string the location of the router bit can be calculated. Sled rotation should never happen, the controller should manage the distances to keep it level.

I wonder if we can make calibration this simple:

1. Turn on
2. Enter desired cutting area size
3. For each cable, Maslow feeds out a little cable then retracts till hitting limit switch to zero
4. Maslow feeds out a fixed length of cable for one corner
5. User screws that cable end to that corner, perfect accuracy not required
6. Repeat from step 4 for the other corners
7. Maslow moves to each corner until hitting the limit switch
8. Maslow moves to center of work area and is ready to cut.

#26

I’m trying to follow your idea. are you saying to take one cable from the top right motor run it to AP2 then continue it to P2 , then on to AP1 and back to the bottom left motor?

#27

Oh, I think if the sled doesn’t tilt, then math can solve for its position. I think the tilt was what made quadrilateral kinematics so complex (though I admittedly haven’t dug into it). Optical calibration probably isn’t needed, though it still might be a solution for things like sag and stretch if it can’t be solved mathematically.

#28

That Amsteel Blue stuff is amazing. It lists a 10% elongation at 46% load, but the rated loads are so high we would probably mostly operate in the 1-5% region. I’ll have get some to test.

Two linkages might be the way to go also! That is a cool idea.

I knew I would get some very smart suggestions if I posted something

#29

Hi, new here, I would like to see a 4 motor arrangement. The way I see it, you could have 4 corner brackets fixed on your workpiece (with known distances). Then all you need to do is measure the workpiece, and input it into the software, assumed the cable lengths are known, means the encoders need to know the actual position of the cable, not a Impulse encoder which is lost after power failure. The big advantage I see, you fix it on your kitchen table and start milling, no need for gravity anymore.
I think someone was suggesting a similar setup, but I’m reading that all on my tiny mobile, and eyesight is on the decline.

#30

New to this and am just setting up my machine now. I’ve got the frame built and have started calibrating so am no expert by any measure. I’m wondering if optical sensors could be used to determine a workspace and then to determine the sleds location relative to the determined workspace. Maybe have a sensor on the sled that looks for a certain colored sticker to determine an upper left and another to determine bottom right (throwing spaghetti at the wall).

#31

Another thing to consider rather than cable/rope is a custom printed ‘tape measure’… maybe out of fiberglass so you don’t have some sharp steel flaying around if something goes bad?

http://www.tapemeasuring.com/fiberglass-tape-measure/

Perhaps they can print a pattern on the tape measure that can be read by the optical encoders to determine position. Since they make tape measures, they might be pretty accurate at what they produce?

#32

Why not have 2 rods whose length is equal to the plywood sheet’s diagonal? (so, 9 feet)

Have impressions along the length of the rod that resemble the bike chain (tack-welding chain to a steel rod would be a decent prototype, or existing maslow kits could come up with some way of attaching the length of chain within a channeled dowel)

Put 2 motors with sprockets on the sled, integrated into a rod housing.

Have the sled “walk” up and down each rod independently to spatially locate the router bit on the surface plane.

The rods should be different distances from the surface plane, and the engagement of the sprocket teeth on the motors should be in-line with the center of the router bit shaft.

the only problem is that when the sled is on the farthest sides of the plywood, one of the 9-foot bars will be near-vertical. If the bar is mounted from the top, that means the bottom of the plywood needs 5 feet of clearance - so the bars likely need to be mounted on the bottom corners and then arc back and forth overhead.

…but then I’d put the mount points much lower than the bottom of the machining surface to keep triangulation, since 2 horizontal, bottom-mounted bars wouldn’t support the sled weight…which also means the bars need to be a bit longer

Future kits could likely drop all parts currently related to taking up chain slack, but would need a central structure for a handheld router and 2 “bar climber” assemblies.

Calibration could be as easy as ‘move the sled up and down each bar until the calibration marking on them both line up with the calibration marking on the housing’.

Installation could be very standardized to enable this: have an array hole guides built in to either side of one dowel/rod that precisely show where they need to be mounted (with the centers exactly 8’ 4.00" apart, with a marking on the rod dictating exactly 2’ lower than the machining surface, etc)

#33

Novel idea, but how many people can accommodate a 9-foot vertical height?

#34

In a garage? I’d think quite a few? Google says the average height is 8 feet for residential garages? the sled would still need gravity to stay on the cutting surface, if the bottom is more than a foot away from the back (measured on the floor) when it’s leaned back, they should fit

Now, all of this could be greatly downsized by reducing the width of the machining surface, of course. May be a case for a 4’ by 4’ “Maslow Mini”, but then the whole thing would be what, 5-7’ tall, before being leaned back?

But the mount points would basically have to be on the floor in either case, for sure.

#35

Can you sketch this having a hard time visualizing it

#36

No. I’m sorry if that was confusing. Those red lines are showing two different and distinct positions the sled can be in (actually they are showing three positions, but when the sled is at P1 or P3 the cables share the same line). When the sled is at P1 then the cable follows the A line (A1A2 -sled- A2A1). When the Sled is at P2 then the cable follows B (B1 -sled- B2).

The idea is that there is one cable that goes from ap1 to ap2 through the sled. The longest that cable would ever need to be is when the sled is located at P2. The shortest the cable would ever need to be is when the sled is located at any point on the line between ap1 and ap2. The take-up of that extra 26" of cable is the only thing that would ever need to be spooled at the sled.

The idea was to limit or reduce the amount of hardware and bulky spooling space needed at the sled.
If the cable passes through the sled we only need to manage 26" of extra cable on each of the two cables which equals 62" total.
If we have 4 cables that each terminate at their own spool on the sled then we need to manage 107" of extra cable on each of the four cables which equals 428" total.

#37

To put it another way, this is the sled at P1:

And this is the sled at P2:

The only “extra” cable is that small loop.

#38

Ah… got it. Neat idea… I guess my next question would be how do you hold onto the cable/wire/whatever? Pinched between rollers?

#39

Yeah, that’s the part where I think my idea suffers the most. I’m thinking pinch rollers or something like a capstan or a windlass might work (or some strange combination thereof). There’s no denying that not having access to the end of the cable causes difficulties.

Another benefit is that the spooling of the spare rope/cable does not need to be highly accurate as long as the measuring is happening at the motors or elsewhere. With this idea you could have a single spool (one on each cable) that could provide tension for a capstan type take-up motor to work while keeping the extra loop of cable out of the way.

Is not needing to spool ~420" of cable worth designing an accurate way to climb the rope?

#40

If anyone wants to get ahold of some Amsteel for testing/playing around with this idea, a have a 1200 foot roll of 7/64" I bought for making hammock stuff with. It’s much much stronger than steel and will easily hold me and my gear up in a hammock, so even the heaviest router and sled will be fine. I have plenty available to contribute to the cause here. Just pay postage.