Maslow Home Maslow Community Garden

Tensioning system to improve performance in bottom corners

chain
tension

#92

@Robert: Thats good info… you should make a copy of your posts in a dedicated thread (so my bookmarks make sense when I go looking for it later)

Personally, I’d always prefer to have extra driven axii… (IMHO you can never have enough e.g. Youtube: Cable driven parallel robot) but barring that, I’m going to give a tensioner a try once I get everything calibrated fully.

My choice for maslow is primarily space savings. I already have an older VMC and a small CNC router, but I needed something that does full 4x8 boards and I really don’t have the space for a full router table… Plus 3d printers have taught me that patience is a virtue so speed isn’t a major issue for me…

My original thoughts were to use a servo with a bungee, and basically make a sloppy reverse maslow on the bottom, but Jacob’s block and tackle tensioner off the idler seems like an interesting solution. The only question is how its going to work with the droop calculations in maslow’s firmware as the tension looks non-linear.


#94

The math is actually even easier than you suggest, but the part that kept this from being a mechanical “home-run” is the unpredictability of the pulley friction, and the cord-to-pulley friction. The advancements are choppy.

The forces generally work out (I bought a fish scale and put it in line to measure forces on the sled) to be even across the range of motion left to right, only changing with height as is the design.

To summarize, you can’t count on forces to be predictable while using an elastic cord of this length with this level of mechanical engagement. There are other designs that should provide a better chance of improvement (eg, the single-pull-down with a ring and weight) but the discussion covering the beam length increase holds the best chance of improving the design for most people, assuming the space isn’t an issue.


#96
  1. Limit the accuracy to about 1/16"

the aim is 1/64

  1. Be simple to build
  2. Be easy to operate.

Again the design is aimed at simplicity and even though the unistruct is used for the top bar, significant deflection will occur on the ends of the bar where the motors are mounted because the vertical supports are too far from the unistrut ends, limiting its accuracy. If the vertical supports were moved toward the beam ends and 1 or 2 more vertical supports added to support the unistrut and the cutting area, accuracy would be improved. But this would add cost and complexity, which might be in conflict with our list items 1 & 3.

That is not the stock design, the stock design has the legs further apart (IIRC
closer to 80 in), which gives the top beam more support. If you implement the
full build of the design you show (I think it’s @bee’s 80 design), he had two
2x4’s at angles to each other to make the top beam more rigid

with a unistrut top beam you could screw a 2x4 below it to add strength.

Lastly, understanding the machine cutting forces is extremely important to
getting consistently good parts. The gravity feed cut is a bit troublesome in
these forums but it doesn’t need to be. Just slow it down on the gravity feed
side of the cut by adding “F10” (feedrate of 10inches per minute or less) or
the equivalent to the g-code. And speed it back up when its done with the
gravity feed cut by adding “F15” or “F20” . Going slow with the router doesn’t
cause fires. Industrial routers go very slow all the the time for certain
types of cuts at speeds that are well below chip load recommendations even
down to 5 inches per minute.

This is a CAM issue, but right now the motors aren’t that much faster than
gravity, so it’s not usually a problem


#97

see the spreadsheets for the forces involved with different dimensions


#98

Would it be possible to have GC attempt to implement these ideas, or could they be added another way, say via post processing script Fusion360?

they would need to be done in the CAM software, you don’t want the machine to
re-arrange your cuts.

  1. “Feed Rate Zones”
    whereby on the outer edges of the workspace GC enforces rates of travel known
    to be more accurate (obv more testing needed to determine where these borders
    lay, and what speeds produce the best results)

the limit on speed depends on a lot of things, including how well the sled
slides on your workpiece (fastly easier on a sheed of oak plywood than a sheed
of OSB for example), so I don’t think this is practical

  1. “Preferred Direction of Travel.”

To be clear @Robert , I don’t think that any of the proposed tensioners have
proven themselves in the real world yet. (readers, please correct me, if i’m
wrong) It seems like this would be the most difficult of the 2 proposals to
implement… but along the lines of the Feed Rate Zone, this would have GC
enforce a ‘clockwise travel zone’ on the lower left side and a
‘counterclockwise zone’ on the lower right.

actually, this is easy (but should be in the CAM software, not in GC

the maslow can cut faster and more reliably when the motors are pulling on the
chain. When the motors are releasing chain you are limited to the the force that
gravity can provide, when you are pulling you are only limited by the force the
motors can provide.

As people build beefier motor controllers and motors, this will be even more
significant.

David Lang


#99

I’ve run the feeds and speeds #'s (through various online calculators) and
we’re still pretty much pegged on the high end (ie: not moving through
material fast enough) in every scenario, inc with a single flute, @ 1/4" bit
diameters.

meaning, that everyone should be at the minimum RPM in nearly every situation.

correct

Now, what @Robert has said about full depth cutting (ignoring the friction
concerns for now) would probably put us into a use case where the full power
of the (default) router will be in play.

the default router has an electronic speed control, as the bit bogs down, the
router will have mroe power applied to keep it at that speed, so you really
should have it at the minimum speed pretty much all the time.


#100

the key things to look at here are the min and max chain tension that are highlighted in blue (which is related to the angles)

there are two configs on the left side, and then the differences in a grid to the right.

in general, things you do that increase the forces in the bottom corner (which helps) will also increase the force in the top center (which hurts), but different things affect these two areas in different ways, so you can tinker with this to find the combination of changes you are happy with.


#101

Because the Maslow is intended for entry level hobby CNC, to be used with entry level software like Makercam and Easel; I agree firmware should have ability to automatically slow down speed when chain lengths are at far lengths, indicating a bottom corner cut.

That being said, today i tried a very slow bottom corner cut at only 9 IPM (instead of usual 35) and unfortunately had same failure as two other pieces in red danger zone. Half tempted to try again at 4 IPM… any thoughts other than adding medium duty retractable lines to bottom ?

Sidebar: Dog Leash for Chain Tension


#102

The problem you’re seeing sounds like it would be addressed by the earlier comments from @Robert about cut direction. It’s odd, but form the mobile platform I’m on right now, I can’t find those comments, only references to them.
If I remember, avoiding downward cuts which rely on gravity for force and don’t have much tension from one of the motors to overcome pull from the bit would apply.


#103

Thanks I remember reading similar but can’t find, as if he was deleted. “avoiding downward cuts which rely on gravity for force and don’t have much tension from one of the motors to overcome pull from the bit would apply.” Is only an explanation why every downward cut near bottom corners is a problem with current setup… we aren’t taking that as a nail in the coffin right? :wink:

Any thoughts on what else we can try besides 12 foot header (which requires new chains and loses ability to use standard software? Otherwise we have dog leash for bottom tensioning to try


#104

you need to reduce friction.

trying to change the cut direction would give you grief as both the cut down and the cut to the left are problems.

I would guess than in both cases (left and right) the object was being cut in a counter-clockwise pattern.

on the left you see the standard symptom of the sled sticking in a down cut (the cut wanders), on the right you see the standard symptom of the sled sticking on a cut towards the outside (the cut stops, and then ‘cuts the corner’ as the chain starts pulling the sled up)

you can sand the bottom of the sled, wax it, put a different surface on it, add weight to the sled, tilt the frame towards vertical, all of which will help in these cases


#105

the 12’ header does not require different software, just different numbers
during the calibration step

But a longer top beam will not help the downward cut problem, only the cut
towards the outside problem.

The downward cut already has full gravity force being applied, the sled is
sticking and the chains both go slack, until the router vibration (and/or the
sled tilting away from the workpiece) jars it loose to slide down a little
futher (but because the chains are slack, you have no control over where it
goes, thus the wandering)

David Lang


#106

Is this a tension issue or a friction issue (I guess actually both). I am still yet to build my Maslow but have been looking at the options prior to build such as the 12’ beam. I have been looking at the sled as it is I admit the one item in the Maslowverse that has me wondering. I have looked at PTFE pads and your post just had me think of roller castors to reduce friction down. Kinda like the image below but have several with a reasonable size ball so that it will drive over lips and uneven joints. Has there been any attempt on something like this in the past? I am thinking an inverter roller transfer table.(just added an image)

imagehttps://www.google.co.uk/url?sa=i&rct=j&q=&esrc=s&source=images&cd=&cad=rja&uact=8&ved=2ahUKEwj0wtjf6qfcAhVLblAKHf_RAN8QjRx6BAgBEAU&url=https%3A%2F%2Fwww.ebay.com%2Fitm%2FLEGO-Technic-STEEL-BALL-CASTER-Steering-joint-socket-robot-wheel-Mindstorms-NXT-%2F332320754604&psig=AOvVaw20B0f4ckEcJ-mRfcboQvSV&ust=1531975052221280

imagehttps://www.google.co.uk/url?sa=i&rct=j&q=&esrc=s&source=images&cd=&cad=rja&uact=8&ved=2ahUKEwjQ5cHs7KfcAhXRa1AKHVVfDGMQjRx6BAgBEAU&url=http%3A%2F%2Fwww.ball-units.com%2Fproducts%2Fbtb-range%2F&psig=AOvVaw2zl930FCCpKdeUg5K3GhIO&ust=1531975797951776


#107

Is this a tension issue or a friction issue (I guess actually both).

The left side (problems cutting down) is a friction issue, both chains are
completely slack.

The right side is both, the sled is sticking, so decreasing friction would help,
but you also have very little tension of the long chain, so going to a wider top
beam would help (but not enough on a sled like this that is sticking when going
down)

your post just had me think of roller castors to reduce friction down.

you are cutting away the material the castors would be trying to roll on, it
would be very easy to end up with even several of them dropping into a pocket
you have cut.

Several people have suggested this in the past, nobody has shown it working

(you also have the problem that they are much thicker than the sled, so you now
have to use a much longer router bit, or get much more creative in your
mounting)

David Lang


#108

Can/has someone gone and cut vertical lines, starting in the center and working out at ~6" to ~1’ intervals, to observe where the machine begins deviating? Think about cutting several parallel, vertical lines. At some point each vertical line, the sled would wander. With this, we could calculate the angle of the close-side chain at which the sled deviates. This would provide quite a bit of information about the machine’s capability.


#109

I have been thinginking about running paracord to pull ya on the bottom corners and then eventually to the other end of the opposite chain. Has anyone tried this? I would use a hanging weight to maintain tension on both the sled and the chain.


#110

Hmm, I’m curious about this. I’m sure someone has had to have tried this? But I can’t think of what the disadvantage or problem would be with something like this. Having two weights connected to the bottom with pulleys in the bottom corners and then going up to near the motor and back down to suspend the weight. Obviously would have to have enough length between the top and the weight to be able to extend to the far opposite upper corner, but would this not help maintain downward and sideways tension?

So here is a quick and dirty picture, although while drawing it out I thought of a couple of things that would have to be different. I didn’t draw in the current chains, everyone knows where they are I think :stuck_out_tongue:

So, one thing that would need to be different is the pulleys at the top would have to be higher than the motors most likely in order for there to be enough length of cord to make it to the far corner. My rough estimation is it would need at least 10 or maybe a little more feet ~9 just to go from one corner of a 4 x 8 to the other, plus a little extra because you’d want to space such a thing off to the side a bit to make sure it adds some side tension as well. That is problem 1 because most machines aren’t that tall and like in my case I only have 8’ 6" to my ceiling anyway. Someone better with simple machine design than me might be able to come up with a solution to that?

Second thing is, I think it would need a second ring system on the bottom side of the sled for this as well. Fixing it on one point would be counteractive to the top rings design I believe. If it’s doing what I think it’s doing (this is just guessing since I haven’t actually built my Maslow yet).

However, it seems like if tension could be added at these points in some way, it would solve a lot of the lower corner issues as I understand them (but probably create new issues that I’m not seeing immediately).

EDIT:
So I just noticed that these forums apparently don’t load older posts by default unless you scroll up to get them to load. Looks like @yaddatrance Had a similar design above, but with using elasticord for tension I believe? Seems like the problem with that though would be the same as has been said for the elasticord as it is used now, not enough tension where it’s needed, and a lot where it’s not?


#111

I could be wrong, but my initial thought is that much of the the tension from the right would mostly be cancelled out by the tension from the left. And in this scenario (assuming equal weights), there would be an overall, slight pull to the downward left (vector sum of the two tensioning strings) rather than to the right, which I think is what’s desired.


#112

Hmm, yeah I’m not going to claim to have an understanding of the math involved without doing some research on it. But I can see what you are saying. The weight on the right would being pulling down and right more in the down direction, where as the weight on the left would be pulling left and down, more in the left direction. They wouldn’t be equal opposite forces to balance each other as desired. To achieve something close to that, the pulleys at the bottom would have to be spaced far enough out to make the pull direction differences negligible, which then runs into the space issues again.

Oh well, it was a nice thought while it lasted :slight_smile:

The bottom corners are my biggest concern with this design. To me, they are unusable as milling regions with default configuration results that I’ve seen like the images from ChuckC above. Making the default Maslow more like a 4’ x 4’ CNC with some extra regions in the upper corners, than a full 4’ x 8’. Honestly, if I had seen ChuckC’s results before I ordered the Maslow, I don’t think I would have ordered it lol. I was under the impression that it had a sixteenth to a few sixteenth of an inch inaccuracies. It appears, it’s far worse in the lower corners than I believed.

As I said though, I haven’t built my Maslow yet, but I might not in the immediate future now. I’d at the very least have to order more chain to make a 12 foot top beam before I built it because I know myself, I’d get pretty frustrated with it if I get results that bad in the lower corners hehe.


#113

is the Sledge being supported? , as its cutting close to the edge
side skirts might help