@bar I was wondering what the difference is between the triangle we are calculating (as if the belt were really a straight line from the anchor to the Z axis of the bit) vs reality (where the belt is a straight line from the end of the anchor plastic to the end of the arm) so I did a quick onshape doc to play around with this and it looks to me like this can be enough of an error to be significant when near the anchors
if the arm length is 150mm, the Z offset is 100mm, and the true x/y distance is 600mm, then the difference between the two calculations is 2.77mm with 2.7mm more belt needed
(with the X/Y distance 1200mm, the error becomes 0.59mm)
This is a pretty simple thing, I set the X/Y and Z dimensions, find the length of the theoretical belt, then I set a horizontal ‘arm’, subtract the arm length from the theoretical belt, and plot where that would hit, reporting the difference between the two actual x/y lengths
This should be setup so that you can view it from the top, enter numbers for the x/y, Z and arm length values and read the error
it should be a fairly simple change to make to the calculation, backing out the
extra length added to the belt before calculating the belt length rather than
just adding it the belt length.
float Maslow_::computeBL(float x, float y, float z) {
//Move from lower left corner coordinates to centered coordinates
x = x + centerX;
y = y + centerY;
float a = blX - x;
float b = blY - y;
float c = 0.0 - (z + blZ);
float length = sqrt(a * a + b * b + c * c) - (beltEndExtension + armLength);
return length; //+ lowerBeltsExtra;
}
to:
float Maslow_::computeBL(float x, float y, float z) {
//Move from lower left corner coordinates to centered coordinates
x = x + centerX;
y = y + centerY;
float a = blX - x;
float b = blY - y;
float c = 0.0 - (z + blZ);
float beltExt = (beltEndExtension + armLength);
float xy = sqrt(a * a + b * b) - beltExt;
float length = sqrt(xy * xy + c * c)
return length; //+ lowerBeltsExtra;
Hi David, rookie question here. Does the code already account for Z height change when actual bit is installed and at some cutting depth?
sort of, it is in the calculation, but the point of this thread is that that
calculation is flawed, so it introduces some error.
if you are cutting with an arm 150mm up from the workpiece, 600mm from the
anchor, the current math calculates that you need 618.46mm of length, it then
subtracts the arm length from that (say it’s 137mm, which I think is close from
memory) leaving a calculated belt length of 481.46mm to extend.
according to my calculations, this results in a position error of 5.5mm of error
and the belt needs to be extended an extra 5.23mm to 486.69mm
if there wasn’t any compensation, the result would be off by 25mm (600-137 for
the belt length)
so it’s close, close enough that it seems to work, but I think the subtle error
is throwing off the calibration, and therefor throwing off everything else as
well. Not badly enough to cause problems for people doing artistic stuff (say
set design, holiday yard decorations) but badly enough to cause problems for
people who need more accuracy
Cool thanks for the explanation. I understand the theory. I wasn’t sure if the higher z height would add to the error.
I was thinking in my head of mounting the anchor points with threaded rod and a couple nuts to raise or lower the anchor to be parallel to sled or standard height for testing the error. But honestly I haven’t even created a frame yet. Just following the forum boards.
Great timing, I’m going to make some large test cuts on scrap wood tomorrow for some parts where I’m looking for dimensional accuracy.
@bar don’t want to make you work on the weekend, but if you end up releasing a test firmware release tonight, I’d be happy to be guinea pig and report back!
I was thinking in my head of mounting the anchor points with threaded rod and
a couple nuts to raise or lower the anchor to be parallel to sled or standard
height for testing the error.
If you do only that, the threaded rod will flex and you will have more errors,
add a stack of plywood disks that get clamped hard under the lower bolt (and
have them glued together so they act as one unit and don’t slide relative to
each other)
the threaded rod will flex and you will have more errors
I am using a frame that consists of 4 cement anchors with threaded rods inserted in each. I have trimmed the bolts to be short, about 1-2" above the ground. Reading this thread, I am wondering if you believe that replacing them with slightly taller rods, with nuts and washers that would fix the plastic at the same height as the Maslow arms would help with this error.
As to the rod flexing, I doubt that 3/8" thick threaded rods that are securely anchored in cement would flex from a 4-5" height
And if so, should I then set all 4 z-offsets to 0?
the threaded rod will flex and you will have more errors
I am using a frame that consists of 4 cement anchors with threaded rods inserted in each. I have trimmed the bolts to be short, about 1-2" above the ground. Reading this thread, I am wondering if you believe that replacing them with slightly taller rods, with nuts and washers that would fix the plastic at the same height as the Maslow arms would help with this error.
As to the rod flexing, I doubt that 3/8" thick threaded rods that are securely anchored in cement would flex from a 4-5" height
And if so, should I then set all 4 z-offsets to 0?
There is more force on these than you think, so check them.
Until the math gets fixed, making the Z offsets all 0 will help. After the math
is fixed, it may or may not help (is it better to just pull sideways on the
sled, or down on the sled, testing needed. Personally I think it will be better
to just pull sideways)
I used 3/8" threaded rods down to cement anchors for my mounts. I actually made the exact switch you describe last weekend putting in longer pieces in the top left and bottom right (i.e., the tallest arms). The rods do actually want to move around. That being said, tightening everything down really tight eliminated any movement for me. When I set up my mounts, I got a pack of 3/8" couplers off of Amazon (link below). I had some extra lying around, so I used those as the spacers. The coupler took up much more of the space (~1.75" long) and, when tighened down, made everything very solid. In the top left (i.e., the tallest arm), I used a coupler and some nuts as spacers (mostly because I was testing if it would work and not necessarily as a permanent solution), and the nuts are definitely the weakest link in the chain. I might get some more couplers that are a bit shorter and replace the nut-spacers with another coupler (or maybe look for a really long coupler that does it all in one piece).
EDIT: I tried to put the Amazon link, and it kept putting up a different item. Regardless, you should be able to find something comparable easily.
I think most of the time there will still be some upwards angle going to the arms. For example, I have my mounts even with the arms when the machine is on only the spoil, but, after adding a cutting material, I’ll have to adjust the z values a smidge to compensate for the extra height of the material. Unless someone is only ever cutting one thickness of material and decides to fix their anchors at that exact height, I think that is bound to always be the case.
