Now that I have my M4 cruising around the frame, I’m noticing that depending on the position and height of the router, the belt entry angle into the arm is more or less aligned with the arm.
Of course, when the M4 is right in the middle of the frame and the Z is bottom out things look good and the belt angle coming into the arms is ok at around 2 to 4 degrees. (depending on which arm you look at)
But when we take the worst extreme case scenario which would be cruising towards the top left anchor at 20 inches from the anchor with 1.5" of material on the frame, the top left belt would be at about 23 degrees from horizontal.
#2 The closer we get to the corner, the bigger the angle and then the sled tends to tilt. (see clip)
.
Ok, enough with the observation, here are the questions.
What if I was to raise the anchor points to minimise that belt entry angle? Since the arms all have a 1" height difference,
I would:
add 4 inches to the top left anchor point
add 3 inches to the bottom right anchor point
add 2 inches to the top right anchor point
add 1 inches to the bottom left anchor point
This would normalize all the belt entry angles and the most extreme would be around12 degrees.
I think this would probably also help with the tilting. But how would this affect accuracy ? I understand that this would shorten the belts a little bit. Would this affect the calculation? Will calibration take care of the shorter belts? This is probably a @bar question.
I have not received my M4 yet, but I was thinking about this being an issue, as it is definitely an issue with my M2. Those same exact problems manifest themselves from old to new. Hopefully by the time I get my M4 this problem will be solved as I am planning on raising all my anchor points to meet the height they come out of the sled. Since my primary mission or purpose to own is working primarily with plexi, the depths have to be perfectly consistent all the way around due to the type of bit I have, its an issue that needs to be solved before I start using it for production work.
I have not received my M4 yet, but I was thinking about this being an issue,
as it is definitely an issue with my M2. Those same exact problems manifest
themselves from old to new. Hopefully by the time I get my M4 this problem
will be solved as I am planning on raising all my anchor points to meet the
height they come out of the sled. Since my primary mission or purpose to own
is working primarily with plexi, the depths have to be perfectly consistent
all the way around due to the type of bit I have, its an issue that needs to
be solved before I start using it for production work.
depths should not be a problem, it’s getting through calibration to get accurate
ancor point locations (and therefor accurate x/y cuts) that is the current
problem.
I understand the calibration issues and the priorities, I’m sure Bar and the crew will iron it out.
We can certainly bring this back into the conversation later on. I would not want to ease out the belt angle and compromise accuracy!
In the meantime, I did some scaled down cutting (under 24"x 24") over the week end on my modified M2 frame and M4 machine
Cutting rectangles, circles, 3D bowls at speeds between 60 to 80 inch per min.
Accuracy: It will come, but was not so bad, at ±1mm in most case (especially compare to the M2)
Duplication: is very good
Square: is spot on
Speed: is great
UI: minimalist but works (no worries there, I know things will tidy up, later)
Great to hear you’re making good progress!
Please share some photos/videos of your cuts when you have an opportunity, we’re super excited to see what people make with M4.
This would effect the calculation, but luckily it there is already a setting for it. In the settings you will see a that we have an X,Y, and Z coordinate for each anchor point. At the moment we’re pretty much ignoring the Z-component and using the defaults everywhere, but if you want to experiment and customize it that should be supported already
Seems like calibration is handling this and roughly adjusted it in the same increments and pattern as the motor heights. The anchor points height differences calculated at calibration are 20ish mm which is in line with the height difference of the arms. I will give a try to increase the anchor points when I get back next week!
One would assume the ideal anchor height be parallel with each arm (or slightly lower to help pull the sled into the work piece a little) when on your typical material and typical bit down in the cut position.
IF the software supports it. Which @bar says it does.
@dlang is experimenting with this and measured them to be 21.25mm apart. Though Bars initial Z heights are not 21/22mm. They are not even consistent according to above screenshot.
I have my M4 still sitting in the box. I am about to build the stiffest frame (vertical) I can think of as well as have the mounts printed or try the gate hinges with stoppers.
BUT I’m hoping there has been more progress on this.
At the very least how to determine the exact number to put in each Z.
How where the original numbers figured out?
Why are they not consistent spacing compared to the actual measured distances between each arm, or are they?
What would 0 Z represent for each arm/mount height? Is this perfectly level (with material and router cutting as explained above) and what we are striving for or?
Sorry so many ?. I have seen a few threads discuss this.
To me this seems to be a very important calculation for accuracy (as well as frame/anchor flex).
There hasn’t been enough testing to say. Lots of speculation, but only in the
thread ‘beyond the green’ do we have someone who is doing the actual testing.
It would be extremely helpful if you could test this.
I think that the default Z-values in the config (40, 60, 80, 100) are valid if you are using just a bolt to fasten the belt ends. Then there is no space between the spoil board and the belt ends. When using the 3d-printed anchors, there is some extra space between the spoil board and the belt ends. The belts are then a little (about 9 mm ?) closer to the arms so you could adjust the Z-values for this difference.
I have raised 3 of my anchors so they are now all at different heights:
For this setup, I used brZ = 34, tlZ = 38, trZ = 30, blZ = 26.
I found that the sled is now a bit more resistant to tipping over as the sled comes to the left side of the frame, but it did not make a big difference. I didn’t dare to make the anchors even higher, because the sled could then be pulled away from the frame.
@willemx This is exactly what I was thinking. Thank you!
2x4s (or even 2x6) on ends with plywood triangles (slightly larger from more support) on top and bottom.
And top stacked better align with arms for parallel cables, or slightly lower (all same offset for same cable angle) to possible help the sled “stick” better to workpiece with vertical frame.
A few questions if/when you have the time…
How did you come up with your Z figures you used?
How is you calibrations and frame flex?
Do you have more information on your frame build?
@dlang Exactly. Adds to my confusion, thank you for bringing this up!
Maybe we can get some programming insight and more details from @bar if/when he has a moment.
BTW, these were manually computed after some very precise manual measurements. The software calibration (after a dozen runs of full 9 x 9 points) would not produce acceptable results even though the fitness was fine (0.75).
Frame is nothing special, just 44 x 96 mm beams, 18 mm plywood reinforcements, 18 mm MDF spoil board 1220 x 2440 mm with extra flaps at the sides to cover the full width of the frame and two shelves (planks?) above and below the spoil board to support the sled when cutting near the top or bottom.
I just 3d printed some extra reinforced anchors with holes for 10 screws.
Good that you mention this, because this puzzled me.
Quoting the user manual about the Z-values:
“These define the vertical spacing between the anchor points and the position of the arms when the machine’s internal z-axis position is zero.”
I interpreted this as “the vertical distance between the center of the belt at the anchor and the center of the belt at the arm”.
In my frame, the thickness of reinforcement plywood at the corners is 18 mm. My waste board is also 18 mm.
Therefore, if I measure from the top of the reinforcement plate to the center of the belt at the anchor and from the top of the waste board to the center of the belt in the arm, I can subtract these values to get the corresponding Z-value.
I think that the thickness of the workpiece is irrelevant in calculating the Z-values, because this is accounted for when you home the Z-position before cutting.
Is this correct?
Quoting the user manual about the Z-values:
“These define the vertical spacing between the anchor points and the position of the arms when the machine¢s internal z-axis position is zero.”
I interpreted this as “the vertical distance between the center of the belt at the anchor and the center of the belt at the arm”.
That is correct
the math is belt^2 = x^2 + y^2 + z^2 where Z is the fixed offset plus the
movement from the router being all the way down.
In my frame, the thickness of reinforcement plywood at the corners is 18 mm. My waste board is also 18 mm.
Therefore, if I measure from the top of the reinforcement plate to the center of the belt at the anchor and from the top of the waste board to the center of the belt in the arm, I can subtract these values to get the corresponding Z-value.
I think that the thickness of the workpiece is irrelevant in calculating the Z-values, because this is accounted for when you home the Z-position before cutting.
Is this correct?
That is not correct, the maslow has no way of knowing how think the
workpiece/spoilboard are, Z=0 (when cutting) is the bottom of the sled, and it
has no way of knowing if you are on a 1/8" thick piece of material or a 4" thick
piece of material.
The maslow accounts for the Z travel from zero, but the workpiece/spoilboard
thicknesses need to be accounted for in the Z offset.
Now, when you have a large frame (suitable for a 4x8 sheet of plywood), the
difference in the belt length is very small, but it’s there.
