will try to back up in the firmware, good idea. I’m also curious to look through the code, honestly without a good understanding of what the machine is doing behind the scenes, its hard to tell whether or not I’m running into software or hardware issues
mfinn wrote:
I’m starting to think this is a new issue.
Original problem being too high friction between the arm and spool, which is now solved (I think)
New problem I think is the motors are just pulling way too tight
@dlang Can you walk me through the various knobs that could be affecting how tight the motors pull?
- Belt length for each arm is calculated using the height of the triangle
(combination of z-zero, spoilboard and work thicknesses?) and horizontal
distance to each anchor (from calibration)
yes, it’s
A = distance from Z zero + Z offset for each arm + workpiece thickness + spoilboard thickness
C = length of belt
B (calculatedd from A and C) horizontal distance covered by belt
total horizontal distance coveredd by belt = B + _beltEndExtension + arm length
(distance from center of bit to the endd of the arm)
- When the machine is “applying tension” is is retracting until it hits a
set tension value, or is it retracting until it hits this calculated belt
length?
it pulls until it hits the stop and ramps up to the retraction current limit
- Z-zero is used in order to calculate the above belt length, but the machine has no knowledge of the tool inserted into it
- it seems like z-zero should be 2 different values, but IMO there’s only the one in the software
- 1 value is used to calculate the height offset between each arm and its anchor
This is Z zero
- 1 value is used to set the tool zero so you cut at the right height
This is Z home
so you set the Z zero with the router all the way down, then you raise it, put
in the bit, and then move to where the bit just touches the surface (either
manually or with a touch-off sensor) and set Z home
- When does the machine calculate the belt lengths? I’d like to run some
tests to decrease the tension. If I update the z-zero, or spoilboard/work
thicknesses, will the machine update its correct belt lengths it should pull
on each arm? or do you need to restart the machine each time you update these
values
they should take effect immediately
I would relax tension, make changes, then apply tension
If the Z numbers are wrong enough during ‘find anchors’, you won’t have good
anchor locations. But I think it’s more likely that it’s a different issue.
David Lang
2 Likes
The easiest thing that I would start with is to run the anchor locating process again.
It will get a better fix on the anchor locations each time that you run it so it might just need another go to get better positions
1 Like
I’ve seen people in the past discuss retraction force and calibration force in the forums. Are these 2 different values that can be set? If so, where do you change the calibration force limit?
mfinn wrote:
I’ve seen people in the past discuss retraction force and calibration force in
the forums. Are these 2 different values that can be set? If so, where do you
change the calibration force limit?
They used to be two separate settings, now there is only one
David Lang
2 Likes
thank you for this explanation @dlang ! it was very helpful
I have never pressed the “set z-stop” button afaik, which I’m assuming is how you set z-zero. I’ve only been pressing z-home, so maybe this is my issue with the tension being too high. Will run a test later to check
When you have the Z down as far as it will go, Zm should be showing 0. The other Z is the value relative to Z-home which what you set at work level.
1 Like
mfinn wrote:
I have never pressed the “set z-stop” button afaik, which I’m assuming is how
you set z-zero. I’ve only been pressing z-home, so maybe this is my issue with
the tension being too high. Will run a test later to check
when you do calibration, it sets z-stop to whereever it is when you start
calibration.
any time you go through the retract-extend-apply tension dance, you also need to
lower the router all the way down and hit z-stop
David Lang
did set z-stop, reran calibration, seemed to work well. When calibrating the machine has no problem moving in straight lines.
Then tried jogging around, belts seemed a bit less tense, but still didnt move in perfect one direction movements (but maybe better). After every jog, it seems like the motors relax and it settles down a bit lower.
Tried to run a gCode (without cutting) and got maybe 2 minutes in before it got a bit jerky in its movements, the belts felt quite tight, and then it failed with the following error messages:
[MSG:WARN: Motor current on Bottom Left axis exceeded threshold of 4000]
[MSG:WARN: Motor current on Top Right axis exceeded threshold of 4000]
[MSG:WARN: Motor current on Top Right axis exceeded threshold of 4000]
[MSG:WARN: Motor current on Bottom Left axis exceeded threshold of 4000]
[MSG:WARN: Position error on Top Right axis exceeded 15mm while running. Error is 15.025mm Counter: 1]
[MSG:WARN: Previous error was 15.025mm]
[MSG:WARN: Position error on Top Right axis exceeded 15mm while running. Error is 15.050mm Counter: 2]
[MSG:WARN: Previous error was 15.050mm]
[MSG:WARN: Position error on Top Right axis exceeded 15mm while running. Error is 15.075mm Counter: 3]
[MSG:WARN: Previous error was 15.075mm]
[MSG:WARN: Position error on Top Right axis exceeded 15mm while running. Error is 15.100mm Counter: 4]
[MSG:WARN: Previous error was 15.100mm]
[MSG:WARN: Position error on Top Right axis exceeded 15mm while running. Error is 15.112mm Counter: 5]
[MSG:WARN: Previous error was 15.112mm]
[MSG:WARN: Position error on Top Right axis exceeded 15mm while running. Error is 15.137mm Counter: 6]
[MSG:WARN: Previous error was 15.137mm]
1 Like
after the machine paused after this error message the top right motor is hotter than the rest, and the bottom left belt is slack, but it wasnt when the machine was in motion
1 Like
This error tells us that the machine isn’t able to move the belts to the length that it needs to for some reason so it is shutting itself down to prevent something from breaking.
Had it moved to that part of the board before during those two minutes? Like did you get the error message the 1st time it went to that spot, or did it do a couple of passes through that spot before and then the error happened?
Where on the sheet is it happening?
1 Like
this was the second pass in that area @bar and its in the top right of the sheet
also, not sure if it helps troubleshoot, but every time i extend belts, the bottom left always extends more than the bottom right does, wondering if that might be an encoder quirk
1 Like
I think I know what is going on.
The encoder is reading the position of the magnet in the end of the roller, but if the magnet is slipping (not glued into the roller fully) then the position that the encoder is reading won’t be accurate.
The way to test this is to retract all the belts fully, then extend them all fully, and then retract them again. The machine will record some numbers in the log which will tell us if the magnet is slipping.
@bar checked the magnets, seemed to all be secure but I added more glue to be safe, but problem persisted
Ran a bunch more tests over the last few days along with the following changes:
- re-sanded everything for a second time, retraction consistently below 700
- replaced encoder on TL arm
- Replaced motor on TR arm
- Replaced my anchor mounts that were 3D printed with the default wooden ones
- Replaced the controller (but flashed existing YAML to use calibration)
- Recalibrated with new controller
- Reverted back to 1.17
- Have been running all my tests in vertical orientation, tilted it down to make it horizontal (didnt solve so put back vertical)
In all of the above trials, the same problem persisted when jogging or trying to run gCode (belts tights, jumpy movement not following commanded path exactly, TR, BL motor overcurrent, loss of accuracy on TR arm, TL arm clicking) see an example serial log below:
Maslow-serial(14).log (3.4 KB)
Today, I ran another test, where I swapped the entire TL arm with the BR arm. The TL I suspected might be a problem because it has been clicking, and might be the source of the jumpy behavior. Most of my cutting is in the right side of the work volume so I figured the BR location might be the least stressed arm, thus I swapped the seemingly bad TL to there.
Reran the test, and the following happened:
- Still exhibiting jumpy behavior
- clicking now happens on BR (so it followed the arm), and it seems to be jumpy
- the gCode was able to fully run (~5 minutes) whereas before it would only make it like 30s before failing due to loss of tracking on TR
- TR did not have any tracking issues
- Jumpiness and clicking got less bad as the sled moved away from the right side of the work volume
- TR and BL still hit current limits
see serial for the above test:
Maslow-serial(18).log (9.5 KB)
Swapping TL and BR definitely had an effect, but not entirely sure what to make of it yet, any thoughts? @bar @dlang @ian_ab
1 Like
see video that goes along with the most recent test^
should be moving in a straight line to the right
I’m traveling and will try to look into this in more detail when I get a chance.
David Lang
1 Like
is there any significance to this memory error messsage:
[MSG:WARN: Low memory: 13420 bytes]
I get it every so often, and wondering if the machine could be running out of memory and thus not able to hold it is commanded path well @bar
also- are these normal center point deviations, or is my BR messed up?
before cutting: [MSG:INFO: Center point deviation: TL: 0.987 TR: 0.701 BL: -0.457 BR: 5.742]
after gCode runs: [MSG:INFO: Center point deviation: TL: 0.742 TR: 0.839 BL: -6.387 BR: 6.467]
btw, it seems like there is a bug that prevents the config window from saving negative spoilboard thicknesses. I input it, but then after I click save and reopen the config window, it resets to 0. Not sure if this could be part of my problem preventing the machine from moving without jittering
Hmmm these seem pretty normal to me, nothing jumps out as a big surprise there.
It does seem like there is something going on with that arm.
The less clicking in the lower position makes some sense because there is less tension on that arm when it is on the bottom because the top arms have to fight gravity.
This also makes sense because there is less tension on the right side.
If you run the anchor locating process again does that have any impact?