That is correct. I think I went for “middle of the belt” as my height mark, but I’m not 100% sure that is correct.
I’m disassembling now to chop the ears now…
Running calibrations and excited to see some frame flex measurements…
What is a "normal’ measurement? A “good” measurement?
From the past few days, I’ve collected the following numbers:
[MSG:INFO: Flex measurement: TLBR: 0.505 TRBL: 2.480]
[MSG:INFO: Flex measurement: TLBR: 0.048 TRBL: 1.466]
[MSG:INFO: Flex measurement: TLBR: 0.344 TRBL: 2.512]
[MSG:INFO: Flex measurement: TLBR: 0.298 TRBL: 1.487]
Are these a concern?
Those are about what I’m seeing too. Re: What is good? We’re still learning.
Also why is TLBR always lower than TRBL? I see the same pattern. Again, we’re learning
different angle to the frame causing more flex (either in the frame or in the
arms)??
David Lang
that’s lower than I’ve been seeing and I have a pretty burly setup. Looks good to me.
Will Puckett wrote:
[MSG:INFO: Flex measurement: TLBR: 0.298 TRBL: 1.487]
k
@bar, suggestion for the future, also list how much belt is out at this point.
This could be flex, or it could be belt stretch.
David Lang
@bar thanks for mentioning you’ve been seeing this pattern as well. I put one rachet strap on my frame when I rebuilt it over the holidays to square it up. It was on TLBR but then since the numbers were different I added one TRBL and it didn’t seem to change anything in the flex measurement. Was about to go to crazy land trying to get them the same.
I’m cutting now with the chopped ears. The second arm down wants to get stuck on the z axis lead screw while plunging when at the right edge, which I think is mainly an issue that applies only to me and my small frame…
Will Puckett wrote:
@bar thanks for mentioning you’ve been seeing this pattern as well. I put one
rachet strap on my frame when I rebuilt it over the holidays to square it up.
It was on TLBR but then since the numbers were different I added one TRBL and
it didn’t seem to change anything in the flex measurement. Was about to go to
crazy land trying to get them the same.
there are five seprate things that can cause the different belt length
measurements when more tension is applied:
- the frame flexes (frame bends)
- the belt stretches
- the arm flexes (plastic bends towards the frame)
- the arm tips (if the fit between the arm and the router is not tight, the arm
can rock at this joint) - the router (and possibly the entire maslow) tips due to the belt tension
being at differenet heights.
putting the anchors at the height of the arms eliminates #3 and #4, but could
make #1 worse if it gives the belts more leverage to bend the frame.
I’m cutting now with the chopped ears. The second arm down wants to get stuck on the z axis lead screw while plunging when at the right edge, which I think is mainly an issue that applies only to me and my small frame…
you could get some thin pipe and put over the lead screw an tack it down to the
bottom clamp to try and deal with this.
David Lang
@dlang that seems it would make sense in my case, with TLBR as my bottom two arms and TRBL as my top two arms, but it seems that people with (top down) TL → BR → TR → BL would have the opposite problem?
For others considering clipping the ears: I’m having some problems that have arisen for z movements since making the mod. I’m not sure if I’ve bent a z lead screw, or if the arm assembly is just not as rigid anymore, but only one of the two motors is moving on some plunges moves.
I’m going to try reinstalling the ears (I ordered a backup one) and see if that resolves the problem. I had a board that was showing intermittent function with one of the z motors, but I just upgraded to the new (black) board, so pretty sure that’s not it…
Revisiting this image, it seems your dimensions are generally larger in the y axis, and smaller on the x axis. I’ve been having the opposite problem, my x axis measurements have generally been larger, and my y axis have generally been smaller. I began to wonder if this was an effect of gravity applying more to the top two belts more than the bottom ones, allowing maslow to swing down and outward.
Last week, I cleared some other items out of my studio and put my frame on the floor, switched maslow to ‘horizontal’ mode, and got a couple of much less errorful cuts–two small squares that were NOT parallelograms (well, they were the special case of a parallellogram that was actually square) and were within 0.2mm of the programmed 100mm.
I then readded my ears at which time I also changed to trapezoidal arm order. I cut some over the weekend but it was a hurried project I haven’t had time to analyze yet. It kept erroring out so I disabled the 15mm position error. Closer to the end of the job, maslow ran off the workpiece. I’ll probably try recalibrating. I’m curious to see if different arm order generates different corner positions.
If you have room and energy to lay maslow flat it might be interesting to see if your dimensionality shifts as well…
What size box did you cut? Also, how far to the edge did you go? I’m mostly sseing the issues on the edges of the 4x8.
What were the dimensions of your horizontal setup? I think the vertical frame design isn’t wide enough to cut accurately at the L&R edges so I actually have been considering trying a horizontal setup.
I had a similar thought when I suggested on a different thread to measure pull forces on each belt individually. I was wondering if there is significant difference that may contribute to errors. But the idea hasn’t gathered much interest yet. Lol. I’m hoping to help when my 4.1 components arrive.
Dano
I cut 100mm squares in the center for the initial test. I had been doing poorly even with such small pieces near the center vertically. They came out much better horizontally.
I took my same 2260mm x 2760mm vertical frame and just laid it down on the ground. I would love to do something larger, but working with the same frame in different positions hopefully will help highlight external factors such as gravity if they are at play… And I just don’t have the space for anything larger at this point!
That’s an interesting idea. In the same vein of thought, I bet dynamically changing the pull force based on location would improve accuracy. For example, on the far right side of the board, TR and BR tension is primarily pulling in the Y axis and has very little strength to counter TL BL primarily pulling in the X axis. I believe this is why the machine comes up short near the edges.
@bar any idea how difficult it would be to change the tension of some belts depending on the location of the machine?
I did some preliminary FBDs last week, and especially as Maslow moves closer to the edge of the frame, the stretch becomes more concentrated in the shorter belt (at a factor of like 5x in my case). Even though there’s a shorter length to stretch (which is part of the Young’s modulus calculation), it’s enough more force to stretch it significantly further than the longer belt. Theoretically at least. It’s still difficult to determine what value to use for the modulus. It’s hard to meaningfully measure such a small number, and probably the only meaningful way to build it into Maslow would be to write the code, then iteratively measure the cut results, splitting the guess in half or doubling it as necessary, and repeating until the results converge to the tolerance of our measuring equipment…
This effect is minimized in wider frames, so I would guess that I’m experiencing it more than most with a 2760mm wide frame…
I think the PID loop may be doing this already… I’m trying to get the debugger connected so I can see the values working while running a program… I’ve gotten it connected, but I keep getting autobreak points in the web and I2C code that I need to navigate… It’s been a while since I’ve actually used the debugger for mcu code…
I’d be really curious to see the tension of each arm in a situation like this. The TR belt goes slack and the BR pulls pretty hard. Given the angle of TL and BL, it pulls the machine toward the center.
Kyle wrote:
@bar any idea how difficult it would be to change the tension of some belts depending on the location of the machine?
The machine doesn’t know the tension on the belts, only their length. It would
be a major change to figure out what the tension is (we can’t just measure the
motor because we don’t know how much force it takes to turn the spool, that can
vary a lot from one arm to another)
David Lang