Musings on frame design

playing around with the simulator, there are a lot of different sources of errors, but the first one to try and address is the distance between the motors. An error of 2mm here can result in almost a 5mm error in the final cuts.

We measure this by using the chain, which is FAR more accurate than trying to use a tape measure.

However, this means that any flex in the frame that changes this distance i very significant. The stock frame depends on the plywood not warping, the arms not flexing, the screws holding the arms in place having no play, the screws holding the motor mount plywood into place having no play, the screws holding the motors in place having no play, and none of the wood pieces changing dimensions significantly due to humidity.

That seems to me to be far too many variables, with far too many different angles for something thatā€™s so sensitive.

So I am thinking that the first thing to do is to eliminate most of those variables by using a solid beam across the top to separate the motors.

Starting with a 10ā€™ long 2x10 beam (ideally LVL, but a good, straight board of other wood is probably ok, and still far better than the stock design), if you mount the motors directly to this beam, you have eliminated as many of the variables as you possibly can (itā€™s also worth considering using a hunk of metal for this, although Iā€™d want to look at the coefficient of expansion for the metal in questions)

From this starting point, the rest of the frame dimensions are not critical. You need to support the beam in place at an angle to match your workpiece, support the workpiece, and keep the chains from tangling

So I am thinking that the frame would have a 8-10 foot lower rail (2x2 or 2x4[1]), to support the workpiece and chain ends mounted to a couple of legs. The legs would be a bit taller than the stock ones (6.5 ft or so) so that they extend above the workpiece to support the upper beam.

I would put a 2x4 between the legs and bolt the beam to this 2x4 so that the beam could be moved in and out as needed to keep the chains parallel to the workpiece if you are dealing with especially thick or thin material

The legs can either be a-frames (like the stock ones, just a bit taller), or you can hinge the top of the frame to the wall/rafters

The stiffness of the main part of the frame should not be critical, so a simple 3/4 MDF/plywood board against a pair of legs is probably good enough, but adding some lightweight framing in a grid (say 1/2" thick) and attaching a 1/4" piece of plywood to the back would make a very stiff torsion box.

Thoughts?

[1] note that a 2x lower rail will work great if you are cutting 3/4 material, but if you are going to cut 1/2" or 1/4" material, the sled will catch on the lower rail, so you may want to make it thinner or have some provision to easily pull it off if you need to work with thin material)

3 Likes

I can confirm that there is flex in the frame between the motors. I just now measured the distance between my motors at rest and during the cal step with tension applied, and I found 12mm difference. Much of that seems to be twist created by the lever arm of the motor mount cantilever. FWIW, I measured 15kg of tension on the chain when the measurement was made.
Your upper beam idea sounds like a good one. In my neighborhood, though, an LVL beam tht size would cost as much as the Maslow :disappointed:. Iā€™ll have to cherry-pick at the lumber yard for an ā€˜nice straight oneā€™.

2 Likes

Iā€™m really liking this ideaā€¦ Iā€™m another one with a pretty fair collection of woodworking equipment and would prefer to just build a frame to begin with without the bootstrap step. Having a hefty beam up top to mount the motors to sounds like a plan.

I understand the goal here, but Iā€™m not understanding the description of how you are doing itā€¦ could you expand any?

Thanks,
Bruce

Hereā€™s an older photo of mine, similar to your description. I built mine on a steel frame, and need to improve the motor mounts, but overall itā€™s fairly rigid. I still would like to tie the top corners to the back wall with 2x4 stock to keep them from bowing inwards at high tension.

1 Like

I see a fair bit of torsion(twisting) in the frame. The cross beam across the top doesnā€™t fully constrain this motion. The ā€œAlternateā€ frame design looks like it might help more by having a frame around the entire 8x4 sheet. A second sheet on the back of the frame would add even more strength.

Iā€™m beginning to think the torsion box idea might be bestā€¦ of course to build a torsion box I fall back to curring parts with the Maslow, so maybe Iā€™ll be doing a bootstrap anyhow :slight_smile:

1 Like

what are the forces that are twisting the frame in your case?

Iā€™m thinking that with the beam on top, there is almost no force in the rest of
the frame.

All of the forces that affect accuracy would be removed from the frame and only
affect the top beam. (at least as I see it)

I know there have been some mentions of the ā€œAlternative Frame Designā€, but I hadnā€™t found the link until now, so Iā€™ll put it here:

Thereā€™s also a discussion in the old forums where itā€™s mentioned that moving the motors farther out horizontally would be desirable for lower corner accuracyā€¦ might play in the plans.

http://www.maslowcnc.com/oldforums#!/modifications:longer-arms-and-steel-mount

1 Like

I think that some of the chattering I see is caused or allowed by the twisting of the frame. If I hold a ruler up to the motor mount and against the wall behind the machine, I can see the arms moving in and out by as much as .5 inch in each direction.

here is a link to an onshape cad of what Iā€™m thinking of.

https://cad.onshape.com/documents/e635c24e358635f51da4b399/w/044390da4e634559b6692ad5/e/a41b321fe467ffd7af9ce0c4

The key is that the top beam is beefy and the motors have minimal overhang from the beam (I forgot to model the holes for the chain ends), and tension from the chains is pulling the motors towards each other against the strong axis of the beam.

2 Likes

that (the link to the picture on github) is what I was thinking of to start with, but I realized that the rectangular frame is putting a lot of material and strength in places itā€™s not really needed.

If you have a beefy top beam, the rest of it just needs to support the workpiece and keep the beam up in the air.

Just thinking out loud, but a couple ideas for the top beam instead of an expensive lvl (special order here).

Plywood stiffener, centered 2x4 on edge making a T. Might be enough left on the parts sheet?

Same with a heavier grade steel stud, steel tee (lots of predrilling), perforated angle, etc. SO except maybe the angle

Donā€™t need much of a lip on the bottom rail, plan to rip off a smidge and make it square, plus a backer board so 1/4 or 1/8 material will work.

Off to the lumber yard today. Maybe Iā€™ll drop by the welding shop and see whatā€™s in the scrap/odd stuff bin for a stiffener.

Just a quick mention before I go off to work. The deluxe frame design I posted on the old forums uses 10ā€™ unistrut rails for a planned panel saw option. I would think the ends of the rails would be nicely positioned for mounting the motors. I think my plans for our build (once our kit arrives) just changed againā€¦

You can find discussions concerning jwolterā€™s nice looking deluxe frame here:

http://www.maslowcnc.com/oldforums#!/?deluxe

and the file is:

https://github.com/MaslowCNC/CommunityGarden/blob/master/Deluxe%20Stand.skp

One question I have is how the lower unistrut is going to work with the panel sawā€¦ I love the panel saw idea, but you had to ā€œrecessā€ the lower unistrut a bit to clear the router and it seems like that would be an issue.

2 Likes

Just some musings about expansion between the motors.

Wood if fairly stable along it longitudinal direction, green lumber will shrink around 0.01% so over a 10 foot span the wood would shrink 0.012 of an inch in length.

BUT wood does have the ability to twist, bow, and cup. This changes the length between the motors far more than taking a board from green to dry. In addition common lumber will expand longitudinal .0000028 inches per degree of F.

A513 square tube will stay stable during humidity changes and has less chance to bow or twist. Steel will expand 0.00000645 inches for every degree F

So if a shop goes from 50F to 90F the 10 feet of steel will expand by 0.03 of an inch and 10 feet of wood would expand 0.01344 of an inch.

Moral of story, steel is better if you want predictable results or have changes in humidity. Wood is better if you are looking for a cheaper option and humidity is stable.

2 Likes

Wonder if 3/4" EMT (Electro Metallic Tubing, an Emergency Medical Technician wouldnā€™t stay still long enough) between the two arms would be enough to stiffen them? Think Iā€™ve got several lengths stashed in the barn rafters, and pretty sure Iā€™ve got a box of 3/4 straps somewhere.

No Unistrut at the lumber yard, didnā€™t drive the extra 20 miles to check the electrical contractor. $130 for plywood, 2x4s, and 1/4" OSB for a spoil board. Gonna make a big box store within 30 miles a requirement if I ever move again.

For a totally nutty idea, instead of EMT, a hefty piece of PVCā€¦ the thick wall stuff is pretty stiff. Lay it along a 2x4 with some straps and it would never compress.

Probably overthinking it, but just in case it gives somebody an idea :slight_smile:

Guys you can literally use the Maslow to cut tabbed pieces to form I beams or better wooden square tubing. They wonā€™t flex. If you engineer it correctly it can be wood and be stiffer than metal.

Hereā€™s just one of the designs Iā€™ve got.


another peice snaps in the top and bottom and a third shape the middle. They snap into each other to extend the length.
Itā€™s a small piece that I can throw in while cutting out other projects to not watse wood. That way Iā€™ll have customer technically paying for my shed construction. Adds more time to cutting, but so what. Itā€™s a robot. I can go assemble things while it cuts.

3 Likes

Make the tabs and gaps the same size and you can make a box with one type of piece.

2 Likes

@iRoc999 can you share the design files for that? it looks like the square openings are dogboned, is there something intended to connect to those?