One thing I had wanted to try in conjunction with reordering the arrangement of the arms, was to have each correspond anchor point offset, so that the anchor point’s relative height to its corresponding arm remained the same for all arms to anchor points.
So, this was a first go, this first picture shows the side arms from my frame, after removal from the frame, with the anchor point offsets attached.
After reinstalling the arms and remounting the M4. It was apparent that the vertical-ish orientation needs opposing corners to be as close as possible to the same height (see this comment Loose belts during calibration - #59 by md8n)
So I immediately swapped the top two mounts for this arrangement
Calibration for this came out fantastically well. However, on my 17 degree slope frame the sled noticably couldn’t sit against the frame with these offset anchors.
So, I’m making two changes
Going back to my original anchor mounts, except for leaving the slightly higher blue one in the bottom right. Which means I have to go back into the maslow.yaml file and change back the relative 0 offset for every anchor point to the usual ones (except for the blue mount/BR value). And also changing the arm order to (from closet to the sled to furtherest away) BL → TR → TL → BR (on the blue mount)
And altering my frame, so that it now sits at about 23 degrees.
I strongly suspect that offset anchor mounts would work really well for those who use their M4 ‘on-the-floor’. Maybe someone could give it a go?
Calibration for this came out fantastically well. However, on my 17 degree slope frame the sled noticably couldn’t sit against the frame with these offset anchors.
when you say the sled didn’t sit against the frame, which way was it tilting?
with the earlier maslow, we wanted the chains to be at the height of the center
of mass on the sled. I suspect something similar here.
If I’m right, if the top of the router tilts towards the top of the frame, you
probably want to have the arms going to the top lower, if the top of the router
tilts away from the frame, you probably want that pair of arms higher.
I didn’t quite follow this; does it mean that the calibration fitness value was exceptionally high (in spite of the sled not sitting flat on the frame)? Could you share the calibration grid size, area and fitness?
Sold. My M4 is on a concrete floor, with fitness of 0.54, so would love to take it to the gym.
Pls no judgement for my dumb question - can i confirm the aim is to get the belts as straight as possible relative to their exit from the encoder, as in, to account for spoilboard and typical material thickness and stack order and get each anchor’s height to match as closely as possible?
Because I’ll double check the 3D file heights for my particular heights
Pls no judgement for my dumb question - can i confirm the aim is to get the
belts as straight as possible relative to their exit from the encoder, as in,
to account for spoilboard and typical material thickness and stack order and
get each anchor’s height to match as closely as possible?
I have a floor mount and raised my anchors relative to the respective arm a little more than a month ago or so, and it improved the fitness score quite a bit. As a practical matter, from the cuts I’ve made since, I think the accuracy improved quite a bit as well. That being said, the taller the mounts are, the more flex can be introduced into them when the belts pull, so you must have some way to combat that flex. For those with 3d printers, you can print mounts with different heights that should help negate flex. For me, I used 1.75" long couplers (as well as some nuts and washers here and there for spacer) that I had used when making my mounts that I tightened down tightly and keeps the mounts really rigid (I use a 3/8" threaded rod going down to concrete drop-in anchors). Mine isn’t the prettiest, but so far it seems to get the job done.
That being said, I configured mine to line up when the machine is on the spoil, so when the machine is on a material there is some angle re-introduced. But I don’t want to try to re-adjust every anchor each time I put a material of a different thickness. I’ve considered boosting them all another .5"-.75" as an average material thickness to close that gap a little, but I also accept that it will never be perfect every time.
I have a floor mount and raised my anchors relative to the respective arm a
little more than a month ago or so, and it improved the fitness score quite a
bit. As a practical matter, from the cuts I’ve made since, I think the
accuracy improved quite a bit as well. That being said, the taller the mounts
are, the more flex can be introduced into them when the belts pull, so you
must have some way to combat that flex.
Note that the PR I just submitted will partially help this by fixing the belt
length calculation. The PR isn’t a complete fix because the belt pulling down at
a significant angle will be trying to flex the arms, causing additional wear and
friction if nothing else, but it should be a lot closer when you are near the
corners, so it should improve the calibration, which then improves everything
else.
Having offset anchor points is really trading one set of issues for another.
You’re getting the same maths and behaviour for each arm (i.e. getting rid of different maths and behaviour for each arm), which should improve calibration and performance.
You’re reducing the angle that the ‘highest’ belt goes through as it exits its arm, reducing wear and tear. In effect, the wear and tear for each arm and belt becomes the same, all other factors being equal.
However, the belts do need to go ‘down’ to their respective anchor points so that you’ve got some ‘pull’ down onto your workpiece, and you’re not just relying on gravity.
And also the potential for ‘flex’ in your now offset anchor points needs to be eliminated as much as possible.
I am trying it out with timber blocks to raise the 3 raised belts, then setting the Z value to the same value on each (75mm) which leaves a small down angle on the belts.
I did a spreadsheet to calculate the height of each block which included a 22mm offset between each block.
I didn’t have any specific down angle in mind, but I was aiming for a minimum, and I’m not sure that is the best idea.
What I will say about the down angle and offset is that you can calculate what the minimum could be.
Do the maths for the sled sitting at the corner opposite to the lowest block (i.e. arm closest to the sled in the stack has its belt fully extended) and allow for the belt to clear your waste board, your work piece and any spacer pieces around your work piece you may use