As you can see from the images, the machine seems to be jumping around as it makes its way across the top of my cut (Basically cutting a square). This does not seem to happen on the left, bottom and right edges (those cuts were actually really nice!). The bit i’m using is a 1/4" spiraling mortising bit from Harbor Freight (See link)
One theory for why this might be happening is a combination of the bit i’m using, the direction the router is moving in this section, and the direction the router is spinning. The current thought is that the bit may be an upcut bit and could be “grabbing” the material with every rotation more than if it were a down cut bit. Furthermore, I’m wondering if the router is spinning opposite the direction of travel and this is also contributing to this issue. An additional piece of information that may be handy is that I didn’t realize i’d set my depth per pass to be 0.01" rather than the 0.1" I’d originally intended for the cut to be. Could this be contributing to my issue also? Can anyone comment on this and see if I’m in the ball park with my assumptions and what the potential fix is?
Another issue I have (I’m hoping @bar may be able to comment on this) is that after after stopping the cut and removing the router, I attempted to do a hard reset on the machine so that I could retract all the belts. All the belts retracted except for one and after multiple attempts of hard resetting the machine and plugging in/unpluggings the connections on that encoder, nothing seemed to want to respond (See picture below)
Finally my last question is how can I stop a cut midway and rehome all the axes to basically start over? I have found that I am unable to do this very effectively by hitting the stop button on the GUI, and I end up having to hard reset the machine, retract all the belts, reinstall the machine, and then rehome it. Can anyone provide any information on this? I feel like it shouldn’t be this difficult.
Do all four of the belts seem tight? Do they seem overly tight like they are fighting eachother?
I don’t think so, if the cut was too deep you might see something like that, but at 0.01 I wouldn’t expect the router bit to really be interacting with the wood much at all.
You can either increase the retraction force in settings or extend that belt again and retract it. Generally when that happens it just means the belt is wound a little funny on the spool and unrolling it and rolling it back up will fix it.
The pause button should let you resume a cut, but after pressing stop you do have to start over. It would be awesome if there was a way to start a file in the middle, but the tricky thing is that gcode relies on information in earlier lines so starting in the middle doesn’t always work right.
Thanks for your responses. Please see my responses below to your questions in the order they were asked:
All 4 belts seamed “tight”, to what extent they needed to be tight I’m not sure. Should it be tight like a guitar string or should there be some “play” in the belt? Do I need to increase the retraction force potentially? If so, how much? Not sure on how this value scales when changing it.
My reason for mentioning the shallow depth is perhaps having it cut deeper would keep the bit from bouncing around as much?
Thanks for letting me know this! I’ll try it.
Interestingly when I hit stop my machine doesn’t seem to respond. Pause and play are really the only commands that work consistently, not sure why Stop doesn’t actually seem to do anything. As for my question, my thought would be that if you hit stop, it would be nice if the machine would basically cancel the remaining G-Code file, rehome the Z-Axis, and then move the machine back to your original home position on your cutting surface to then be ready for the next file input. In essence, to your point, it does “Start Over” but not to the extent of a physical hard reset that would require diassembly/reassembly of the machine to the frame. Is this possible to incorporate?
They should be reasonably tight, but they don’t need to be guitar string tight. If you can get a video of it in action we can probably better deduce what is going on.
Maybe, but this seems like an issue with how the machine is moving to me.
Absolutely! It’s a little tricky because the firmware is built on FluidNC and FluidNC handles stops in a bit of a weird way so it’s tricky to get everything playing nicely. I kinda thought we actually already got that worked out, but I’ll have to double check what it’s doing.
I’ve definitely seen the same thing happen, and agree that “Stop” canceling and rehoming would be awesome. Had a cut last night where the Z didn’t set right (I was on my phone and for some reason it didn’t go through). The cut then dragged through the wood between cuts. Had to turn it all off, get the Z up above the material, then restart the whole retract and extend process.
@Bar I attempted to Extend all the belts and then retract them per your original instructions to me regarding the one belt not fully retracting back into the Maslow. Upon extending the belts fully and then retracting them back, this belt (pictured below) is now no longer even retracting at all and is completely stuck in an extended position. Weird that it extended the belt fine but will no longer retract. Thoughts? Bad Encoder? Board issue? Not sure. Thanks!
If it won’t move at all then there’s a chance the belt could have jammed up in the gears of the arm. I had that happen last week and had to take the affected arm completely apart to get it unstuck. It shredded some of the teeth on the belt, but still seems to be working well enough regardless.
The easiest thing to do is increase the retraction force in the settings which will probably make that arm retract, but if that doesn’t do it I agree with the folks who are saying that there is probably something jammed in that arm which is preventing it from spinning.
@Bar what is an adequate increment to step up the retraction force? Not sure
how much is too much, if there even is too much.
past advice has been to bump it up 200 or so at a time, up to 2500 or so
when you are cutting, it alerts if it hits 4000, so I think that’s a bit
conservative.
you want to minimize belt stretch and make sure that you aren’t flexing the
frame (and increase the calibration force as well when you are done)
in many cases, extending and retracting a few times wears whatever is causing
the retraction problems enough that you can lower the retraction forces again.
Hi Everyone! Been a little while since I’ve posted but I was able to get back to troubleshooting and although I have fixed a few issues, some new issues have come up.
For starters, I was able to get all the belts retracted fully by increasing the retraction force from 1300 to 2100. I decided to then up the retraction force again to 2500 and then performed a recalibration. After performing the recalibration, I ran the profile I had been running before and things were going fine until it started moving vertically upwards, then that’s where the issue came up.
The unit was unable to go up the full length before the bottom left belt completely became loose and the bottom right belt was also somewhat loose (see pictures below). The full picture of frame and unit shows where the machine failed and I also have a picture of what he cut looked like during it’s failure. Interestingly, I would like to note that during my calibration, this same belt (bottom left) was not winding up properly and I could see it not staying wound around the spool at all times.
For starters, I was able to get all the belts retracted fully by increasing the retraction force from 1300 to 2100. I decided to then up the retraction force again to 2500 and then performed a recalibration. After performing the recalibration, I ran the profile I had been running before and things were going fine until it started moving vertically upwards, then that’s where the issue came up.
The unit was unable to go up the full length before the bottom left belt completely became loose and the bottom right belt was also somewhat loose (see pictures below). The full picture of frame and unit shows where the machine failed and I also have a picture of what he cut looked like during it’s failure. Interestingly, I would like to note that during my calibration, this same belt (bottom left) was not winding up properly and I could see it not staying wound around the spool at all times.
This is becoming a FAQ
check that your frame is not flexing as the belts pull tight (especially during
calibration)
was the router all the way down against the sled and Z stop is set at that point
when you did the calibration
check that the Z offset values in your maslow.yaml file are correct
@dlang I am not seeing any noticeable flexure in my frame during calibration, and the router was at it’s lowest position during the calibration. What Z-Offset should I be looking for in the .yaml file?
@dlang I am not seeing any noticeable flexure in my frame during calibration,
and the router was at it’s lowest position during the calibration. What
Z-Offset should I be looking for in the .yaml file?
There are 4 Maslow_*Z variables, one for each arm. This is the distance from the
belt at the anchor to the belt at the arm in the Z direction.
The defaults are designed for the anchors being attached to the frame behind a
3/4" spoilboard with a 3/4" workpiece on top of that and the sled on top of the
workpiece.
As for the anchors, I designed them loosely off of the 3D printal options available on the Maslow site, the main difference being is that I added 3/4" from the base of that version to account for the 3/4" spoil board. There is a reason for this.
Currently (and something I need to look into changing in the future) I have a set of workpiece holders that attach to my frame to hold the workpiece in place during a cut (see the picture below).
At the present, these holders stick out too far to allow a complete Maslow calibration to be performed on the SPOIL board as the Maslow will have interference with the holders when it gets low enough. To account for this, my corner anchors were made 3/4" taller than the standard anchors found on the Maslow site, and when I do my calibrations, I perform it WITH the workpiece installed. With the workpiece installed and the calibration running, this allows enough clearance between the bottom of the sled and the bolts holding the workpiece holders in place to not interfere. Theoretically, this configuration should be no different than if I were to have the workpiece holders removed and the anchors were 3/4" shorter running the calibration on the spoilboard, although I do wonder now if the software is anticipating an additional thickness being added as the workpiece from the calibration file data to ensure adequate tension is being applied. Does this logic make sense? Here’s a side screenshot of my frame, anchors, and workpiece holders so you can see what i’m talking about better:
Another thing I noticed too while running my initial calibrations was that the older anchors were so short relative to the workpiece surface that the belts would contact/lay on the edge of the workpiece when the Maslow was at certain points on the frame during calibration. I didn’t think this was normal or correct, but could you shed some light on this and if you’ve ever experienced anything like this?
As for the anchors, I designed them loosely off of the 3D printal options available on the Maslow site, the main difference being is that I added 3/4" from the base of that version to account for the 3/4" spoil board. There is a reason for this.
Currently (and something I need to look into changing in the future) I have a set of workpiece holders that attach to my frame to hold the workpiece in place during a cut (see the picture below).
At the present, these holders stick out too far to allow a complete Maslow
calibration to be performed on the SPOIL board as the Maslow will have
interference with the holders when it gets low enough. To account for this, my
corner anchors were made 3/4" taller than the standard anchors found on the
Maslow site, and when I do my calibrations, I perform it WITH the workpiece
installed.
That is actually what you should be doing in any case, so the Z values above are
all too large by 3/4" (19mm), so try reducing them all and doing a calibration
again.
although I do wonder now if the software is anticipating an
additional thickness being added as the workpiece from the calibration file
data to ensure adequate tension is being applied. Does this logic make sense?
it does not, it assumes that you are calibrating on the top of the workpiece (or
that the difference is small enough to not matter, which I don’t believe is the
case)
I was able to get back to troubleshooting this unit and I went ahead and followed the direction provided by @dlang. I updated the YAML File to reduce the Z Values by 19 mm and reran the calibration. I went ahead and took video of the calibration as well as certain points and I want to point out the slack that forms in the bottoms belts, particular the bottom, left:
Is this behavior normal? I don’t believe it is…
After the calibration finished, I went ahead and homed the unit to the bottom left corner (41" left, 20" down from dead center of the board), loaded in my test g-code file, and then ran the file without the router running. No surprise, the unit got to the top and then had the same behavior it has been having when it hits that point. You can see the belts and the bottom become loose, the unit starts to shift left and right with a jerky motion, and then it quits out after a certain time.
I hope these videos help to explain what is going on and hopefully someone (@Bar) has a solution as i’m getting frustrated with this and would really love to start working on some projects. Thank you everyone for your help!
