Chain sag inaccuracy

Technical Details
82

…but Maslow can cut light gauge aluminum and other soft metals, where 0.1mm accuracy is not unusual.

I agree with you on the deflection and swell/warp of wood, which is why I plan to use Unistrut or steel channel for the frame. And your metal linkages should help also.

There are higher accuracy #25 chains with tight tolerances. And if #25 isn’t stiff wnough, we can go to bigger gauge chains.

Maybe we add one more LVDT-based encoder at the top between the motors and link it to the sled via thin wire for additional position feedback to help with the router hitting knots in the wood like you said.

Just some ideas. I’m sure the people on this forum have more.

Even with all the above, the price of a Maslow would still be significantly less than any comparable CNC router I’ve seen (for example see: https://www.kickstarter.com/projects/2130625347/goliath-cnc-an-autonomous-robotic-machine-tool-for/description )

I think +/-0.1mm is achievable (ok, let me be more accurate: I FEEL that +/-0.1mm is achievable). The encoder resolution is there. And there are great minds on this forum ironing out the inaccuracies one by one.

Even though this started as a hobby CNC, I think by the time we’re done it’s going to be “professional grade” :slight_smile:

We’re going to get there…

83

Yes, I understand both play a part, but I was thinking that line stretch may be easier to compensate for with software than catenary deflection and the harmonics it creates. Those chains are like big heavy guitar strings out there. Every movement creates a wave that resonates along its length. Reduce the mass of the lines and you reduce the resounding waves.

84

I think if we compensate just for static sag we will be way ahead of where we are now in terms of accuracy.

That said, there is a balance:
Heavier chain is stiffer, so less stretch, but more sag.
Lighter chain means less sag (and a higher resonance frequency like you allude to), but more stretch.

Let’s do the math just for static sag for starters, and then decide.

85

I don’t have the power to move comments from one thread into a new one :slightly_frowning_face:…I would recommend creating a new topic and quoting the original posts

2 Likes
86

FYI - My practice is to suggest it. If that isn’t effective, I create the thread and point to it. " here - I made this please move here for X and keep Y here "

Thank you

87

note that the cable will be at different heights from the workpiece depending on
how far it’s been fed out. If you add a guide to limit this, you are adding a
pivot point to the line.

88

You could make the spool go back and forth like a fishing line.
And is a pivot point a problem?

89

the ‘spool and cable’ concept is simply a non starter. Chain and sprockets is the best method. lets stay focused on increasing the accuracy of the primary design.

Its clear that a horizontal top bar, a linkage kit (multiple methods) and some ‘tightening up’ of the math is getting us closer and closer to the mm line.

lets keep the momentum going in those directions at least until we seem to run out of known sources of inaccuracies…

I expect that at that point we’ll be just around a single mm, and that perhaps thats good enough for plywood, at which point we work on increasing machine speed while maintaining best accuracy.

4 Likes
90

You could make the spool go back and forth like a fishing line.

that sounds like a lot more parts to get right.

And is a pivot point a problem?

It can be, it complicates the math and means that the line needs to make a sharp
bend at that pivot point (or if it can’t make a sharp bent, you loose accuracy)

1 Like
91

Hey, i’m not saying i like the idea, just my thoughts on solving parts of it. But i don’t really like the idea as a whole, throwing actors and camera’s across a set is a whole different thing as moving a router with half a milimeter precision. Also we’re way further in development of the chains, then starting some completely new idea

1 Like
92

In theory as long as there is the option of relative calculation it should not matter what you use. Chain, rope, cable. My focus is on what I call a Standard build. Meaning the most common, easily available and practical parts. The project should remain open to alternate setups. People working in alternate designs also understand they may have to work on some of their own development. They may need to adapt. As in some of the different size builds people have mentioned.

My 2 Cents

Thank you

2 Likes
93

Has anyone measured to see how much force it takes to pull a router+sled across a sheet of plywood with and without a bit engaged in the plywood?

My calculations show that the friction between the sled and workpiece would result in a few pounds of resistance. This would be significantly lower with a waxed sled bottom.

The major variable that has yet to be defined is the force due to cutting the work piece. Obviously, this will vary due to the feed rate, the depth of cut, spindle speed, bit type, how sharp the bit is and what kind of material we’re cutting.

All that being said, we have numbers for the static tension in the chains. Some ballpark numbers for the cutting force would help us put everything into context.

Ultimately, we may learn that the cutting force plays a bigger role in the inaccuracies than just the weight of the chain causing sag. The solutions to those two problems are very different.

94

we may find that the cutting force and sled friction are significatn factors,
but before we worry about that, let’s work to correct for the factors we do know
about and then measure the accuracy of the resulting machine. It may be that we
then nee dto dig further, or it could be that we find that the resulting errors
are not measurable.
k

95

My understanding is that the Maslow is less accurate with larger depths of cut and/or higher feed rates. Those two variables are chain sag agnostic. That’s an indication that cutting force matters. Especially if the end goal is to increase the speed of the machine.

96

if you go too fast it can be a problem, however it’s not clear that the problem
is cutting force as opposed to the chain angles getting steep enough that
gravity doesn’t apply enough force to move the sled at the speed that we are
looking for.

We also don’t have any acceleration planning in the maslow firmware, it assumes
that the sled goes from stopped to full speed instantly (and also full speed to
stopped instantly)

the fact that the sled doesn’t actually do this leads to inaccuracies, and they
get drastically worse as the speeds go up.

trying to deal with friction before we deal with acceleration planning is
focusing on the wrong thing

1 Like
97

Point taken on the acceleration planning. That’s definitely a problem that will affect accuracy everywhere on the work piece.

However, since this thread is about inaccuracies due to chain sag: if gravity (*mass of the sled) is not supplying enough force for the sled to move it at the speed we want, that’s precisely because of the resistance due to cutting forces and friction.

Chain sag is a function of chain tension and length. Any chain sag calculation will need the chain tension. We know the static tension. Cutting forces and sled friction make up the dynamic portion of it.

Less tension and longer chain leads to more sag. If the sled is in the bottom left, the chain on the right side already has low tension. If the sled is moving to the left, the resistance due to cutting forces and friction will lower that tension even further leading to more sag.

Sled friction is easy to calculate to within a pound. If we have a ballpark number for cutting force (at a given depth of cut, feed rate, etc.), then we’d have enough variables defined to create a rough dynamic calculation. If all we want is a static calculation to start with, we’ve already got all the math done earlier in this thread.

1 Like
98

However, since this thread is about inaccuracies due to chain sag: if gravity
(*mass of the sled) is not supplying enough force for the sled to move it at
the speed we want, that’s precisely because of the resistance due to cutting
forces and friction.

not entirely, you could have zero friction and still have the situation where
gravity is not supplying enough force to accelerate the sled fast enough

Chain sag is a function of chain tension and length. Any chain sag calculation
will need the chain tension. We know the static tension. Cutting forces and
sled friction make up the dynamic portion of it.

A big question is how accurately we need to know this tension. I was hung up on
this problem for a while, but it was pointed out that even if we are off by a
fairly high percentage, we would not be that far off in terms of the resulting
effect.

it takes quite a bit of sag over a pretty long distance to make a noticable
effect, so even if the tension is off by 50% we are still making a significant
fix over the stock approach.

Less tension and longer chain leads to more sag. If the sled is in the bottom
left, the chain on the right side already has low tension. If the sled is
moving to the left, the resistance due to cutting forces and friction will
lower that tension even further leading to more sag.

true, and moving the motors out wider will reduce the angle to the near motor,
resulting is significantly more gravity force to move the sled.

Sled friction is easy to calculate to within a pound.

That I don’t believe. You don’t know the suface area that’s in contact, because
you don’t know what’s been done to the surface you are riding on. You don’t know
how smooth the bottom of the sled is. You don’t know how smooth the surface you
are riding on is.

That’s far too many variables (most of which will change over a given run) to be
nearly that accurate.

you can’t calculate it from the motors because you don’t know how much power a
given voltage results in, you only know the resulting movement.

If we have a ballpark number for cutting force (at a given depth of cut, feed
rate, etc.), then we’d have enough variables defined to create a rough dynamic
calculation.

don’t forget to factor in the type of wood, how much glue, how sharp the bit is,
etc.

again, far too many unknowns.

2 Likes
99

If there is zero friction and zero cutting force, it would be equivalent to having the sled suspended in midair supported only by the two chains. That situation is the absolute limit for how fast the current incarnation of the Maslow can move. If you’re saying that gravity is still not going to accelerate the sled fast enough in that situation, then the sled either needs to be heavier or have something actively pulling it down to the bottom corners.

As for the frictional force between the sled and the work piece: contact surface area size does not affect frictional force. Empirical testing shows wood on wood static coefficient of friction is between 0.25-0.50 (source linked in my earlier post). Kinetic friction would be lower. Waxing the sled bottom would significantly lower that, sawdust would also lower that.

My calculations showed that a 20 lbs sled on a frame angled at 15 degrees with a coefficient of friction of 0.25 resulted in a frictional force of 1.3 lbs.

Tl;dr friction between the sled and work piece is less than 2 lbs. Possibly less than 1 lb if you wax the bottom of your sled.

1 Like
100

If there is zero friction and zero cutting force, it would be equivalent to
having the sled suspended in midair supported only by the two chains. That
situation is the absolute limit for how fast the current incarnation of the
Maslow can move. If you’re saying that gravity is still not going to
accelerate the sled fast enough in that situation, then the sled either needs
to be heavier or have something actively pulling it down to the bottom
corners.

or che chain angles need to be changed.

remember, it the chain is only 10 degrees from vertical, only 17% of the force
of gravity is working to swing the sled closer to vertical, the rest of gravity
is just pulling down on the chain.

As for the frictional force between the sled and the work piece: contact
surface area size does not affect frictional force.

in theory yes, in practice it’s not that simple (the theory assumes that the
surfaces do not deform on contact)

and many examples how that spreading the weight of something across a larger
area will make it easier to slide

Empirical testing shows
wood on wood static coefficient of friction is between 0.25-0.50 (source
linked in my earlier post). Kinetic friction would be lower. Waxing the sled
bottom would significantly lower that, sawdust would also lower that.

and chips or tear-out sticking up would increase that)

My calculations showed that a 20 lbs sled on a frame angled at 15 degrees with
a coefficient of friction of 0.25 resulted in a frictional force of 1.3 lbs.

Tl;dr friction between the sled and work piece is less than 2 lbs. Possibly
less than 1 lb if you wax the bottom of your sled.

If that really is the case, it’s unlikely to be a noticable factor in chain
tension.

1 Like
101

That’s not how gravity works. Gravity pulls on a body in proportion to its mass…so in the end everthing from a ball bearing to a boulder will end up accelerating at the same rate regardless of mass.