I just tried again to cut a test disk. I started the job towards the center of the board but on the left side. I had retracted the belts, extended them, applied tension, and it said it was ready to cut. When i clicked play, it disconnected. I was running the most recent firmware sent this morning.
I am first going to try again without a shop vac running nearby. (I read about a drill causing interference). Then if that fails I will be going back to the released version of 1.17 since I have had some success with that.
Thanks,
Shop vac had no effect. I started cutting the test cut with it off, and then turned it on and brought it close to the unit. No dice.
Bigger problem was that despite going through the gcode with no problems, it never cut the board. It danced around rocking on the upper limit of the sled.
Then it did something new: when I went to jog it 400 mm to the right so that I could see the test cut, it went in a straight line all the the to the right of the plywood. To stop if from falling off I unplugged the Maslow.
I am currently 3D printing a couple of arms to hold casters at the top of the board. If I’m correct, this will not allow the Maslow to tip and rock on this edge. If anyone has any suggestions on how else they have mitigated this rocking I would appreciate it.
More importantly if anyone can explain the runaway I would appreciate it. That kind of worries me.
Update: So now it is stuck in extended state. I have retracted, extended , and applied tension 3 or 4 times. I ended up having to increase the retraction force to 1000 because the lower right belt stopped retracting. Now it will not go to ready to cut.
Shouldn’t need to recalibrate, although it may improve things. I have had a similar problem when releasing tension in one of the corners, where I think 2 of the opposing belts pull the 3rd to far for Maslow to identify its position. Retract, Extend dance then.
Also, I think you may have to increase the support angle. People have had trouble with 25 degrees. I ended giving up and going to a table version with belts attached to the walls. I also extended the belts 1 metre each with aluminium square extrusion.
Currently, if you are cutting a file and hit stop, then hit play, the maslow will continue where you left off. This is desirable much of the time, but if you realize you started the wrong file, hit stop, load a new file, and then hit play, it continues to cut the old file. This is NOT desirable
This PR clears the internal queue when you load a new file. My home maslow is still down and there is a marathon happening in town this weekend so I don’t want to try going anywhere close to the makerspace
could someone please test this build.
Test process. On the stock 1.17, have a small file that you know what it will do. start the file, stop the file, start it again so you know what normally happens.
Then load this test firmware and start the file, stop the file, load a different file, and start it again. With the stock firmware, it would continue to cut the old file, with the new firmware it should start cutting the new file.
you don’t have to go more than a few moves into each file to see the behavior.
David with the existing V1.17 it behaves correctly, i.e. When you hit stop then play Maslow returns to the start of the print and then begins again. If you hit stop and load a new file when you hit go, it starts the new file.
Pause stops the print and then run resumes at the same place it was paused at. I think this was fixed a couple of changes ago.
I didn’t try your code changes as the problem you identified was not there.
Ian
First off, thank you for the explanation of why it could be coming so slack in the upper corners. I am hopeful that it will stop losing so much tension and going into extended mode as I get things dialed in. Would increasing the tension force improve this performance? With it built and attached to the wall it is pretty difficult and looks incredibly sketchy to put this at a flatter angle.
Will I need to recalibrate if I change the angle?
Second, does the Maslow work in the vertical orientation? If everyone is giving up on it and moving to a horizontal orientation, then that is a problem for me. I do not have the space to set this up horizontally. I know I have seen videos of it being set up at some vertical-ish angle, but do people keep them in this orientation?
I am currently printing two of these to hold casters above the sled to try and eliminate the dancing on the nose of the sled. The line up with the holes and ridges that already exist on the sled.
I plan to attach them before I recalibrate when I return to school.
David,
On Thursday the 12th I tried the released version of 1.17, as well as the 3 that were attached recently above. After testing and I was unable to get it to stay ready to cut in the place I needed to make a cut. The belts kept going slack and it would go to extended state. I believe the one you are referring to is listed as Serial Test 117 D-2 and the Serial Test 117 D2 Final.
I returned to the released 117 because that was the firmware I was using when I had a successful cut. I need to get this thing cutting for student projects, so I’m trying to get something to consistently work.
David with the existing V1.17 it behaves correctly, i.e. When you hit stop then play Maslow returns to the start of the print and then begins again. If you hit stop and load a new file when you hit go, it starts the new file.
Pause stops the print and then run resumes at the same place it was paused at. I think this was fixed a couple of changes ago.
I didn’t try your code changes as the problem you identified was not there.
Thanks, that’s interesting. I’ll have to update the machine at the makerspace
and try again.
First off, thank you for the explanation of why it could be coming so slack in
the upper corners. I am hopeful that it will stop losing so much tension and
going into extended mode as I get things dialed in. Would increasing the
tension force improve this performance?
no, what we are finding is that the lower the tension you can get away with
during calibration, the more accurate the calibration is. The tension setting is
not used when cutting.
Will I need to recalibrate if I change the angle?
you should not.
Second, does the Maslow work in the vertical orientation? If everyone is
giving up on it and moving to a horizontal orientation, then that is a problem
for me. I do not have the space to set this up horizontally. I know I have
seen videos of it being set up at some vertical-ish angle, but do people keep
them in this orientation?
yes, some people do. They have found that an angle of 20-25 degrees works better
than the 15 degrees Bar initially expected to work.
I placed a couple of buckets under the bottom of the frame to make it more horizontal. I believe it is closer to 30 degrees at this point. I don’t know if I will be able to maintain that because it looks pretty sketchy.
I updated the firmware back up to the 1.17 D (v1.17-122-gb02fde01-dirty) to try for my testing today. I could not get the Maslow out of extended state. I had to recalibrate.
It took 3 attempts to recalibrate as it kept failing. I noticed the belts kept getting hooked on the plywood supports on the frame. Once I shielded them with cardboard it worked.
I have a couple of sticking points when the calibration fails. First off, it seems to be stuck in this Find Anchors Computing mode. Then in the suggestions it says to check frame measurements. Where can I enter the frame dimensions?
As I was recalibrating again it dawned on me there has to be a better way. Having to release tension, support the sled, retract the belts, take it down from the frame, pull the bit, lower the Maslow all the way on the z axis, reset the z stop, re-support the Maslow sled on the frame, extend the belts, and then being able to click calibrate, there has to be a better way. This takes a lot of time, and I know personally that I don’t have much of this. Why do we have to remove the cutting tool to recalibrate? Would we be able to simulate the stability the Z being all the way down by placing a stop between the frame and the router so that it was at a calibrated distance from the board?
Thank you,
Update: 10 am. After taking an hour and a half to get a successful calibration. As soon as I got a bit put in and it hung back up on the frame it is now stuck in extended state again and will not cut. The Maslow is not ready to move. What would happen if I ran a calibration cycle with the Maslow elevated on the Z axis to the point where it is zeroed for the cutting tool?
Will you make a separate thread for the issue that you are seeing? It get’s hard to keep track of things and this thread is supposed to be a place to post different firmware releases for testing
Edit: Actually I think this is totally the right place for this comment because I think that a lot of what you are are bugs that were in the pre-release version which was last published here. Using the versions here can get you some new features early, but it can also lead to getting bugs.
Yes! I ran into that yesterday too and fixed it! It was super annoying.
Good catch, that is left over from an old system where we asked people to manually measure their frame which led to a lot of issues. I’ll take that suggestion out right now. Thanks for pointing that out!
You shouldn’t have to do the retracting and extending the belts unless things end up in an unknown state which should basically never happen. I think that is the real issue here.
This is pretty important. Having the z-axis all the way down gives us a chance to set the home position for the z-axis and makes the machine as stiff as possible. We could make it possible to locate the anchor points with the z-axis higher, but it would result in less precise measurements and the goal is that you really only have to locate the anchor points once and then never worry about it again. If you are finding that you have to do it often that is a sign that something else is going wrong. It should be a one time thing.
Here is an updated version which fixes the issues with getting stuck in the FluidNC “home” at the end of locating the anchor points which would prevent the machine from moving, and also making it so that a failed computation will exit gracefully to the “Extended Out” state instead of getting stuck in “Computing” forever.
There are also some great UI updates from @ian_ab in here.
I’d like to consider this a candidate for a full release so let’s check it extra carefully for bugs and then maybe publish it next week if everything looks OK?
Interestingly enough. I was able to successfully do a test disk as well as an entire project with the prior version. After I clicked send on the 10 am update, I accidentally hit extend from the position where it was hung up on the frame. This caused the belts to extend out, and then I retracted all, and extended, applied tension, and it was still in extended state. So I slackened the belts, then hit apply tension again, and it said it was ready to cut. So I cut 2 jobs from start to finish. After both jobs I attempted to jog the Maslow to the side so I could look at it, and it started to runaway. No hitting stop in any menu would stop it. So I ended up cutting the power.
After the second job, I updated the firmware and index. I like how the green button on the face guides you through to the next step. However My machine is currently stuck in extended mode again. I have tried loosening and then tightening, retracting & extending to apply tension, no luck to get back into ready to cut.
[MSG:DBG: Waypoint 0 coordinates: X=0.000 Y=ovf]
[MSG:INFO: Center point deviation: TL: 0.000 TR: 0.000 BL: 9.541 BR: 19.863]
[MSG:ERR: Center point deviation over 12.000mm, your coordinate system is not accurate, maybe try running calibration again?]
Basically the machine is checking that nothing has changed since it last calibrated and it’s finding that the bottom right side has changed too much.
The question is why is that measurement changing? Is there anything about the machine or frame that doesn’t feel solid?
The bottom right side keeps having retraction issues. I keep having to up the retraction for e to get that side to fully retract. It will start to retract, but then stop like it has hit the tension threshold of being all the way retracted. I can click retract all a second or third time but it will only pull in a few more inches until I up the threshold.
I assumed this was sub optional but manageable with the threshold setting. I thought this was the reason the retraction threshold was described as “as low as possible and still fully retract”.
So, what do I do to correct this?
Thank you,
The bottom right side keeps having retraction issues. I keep having to up the retraction for e to get that side to fully retract. It will start to retract, but then stop like it has hit the tension threshold of being all the way retracted. I can click retract all a second or third time but it will only pull in a few more inches until I up the threshold.
I assumed this was sub optional but manageable with the threshold setting. I thought this was the reason the retraction threshold was described as “as low as possible and still fully retract”.
So, what do I do to correct this?
what tension setting do you need to set it to for it to fully retract?
unfortunantly, the only way to fix an arm that’s not moving correctly is to
disassemble it.
common problems
the spool doesn’t move freely on the arm halves (sand the inside carefully,
with high grit sandpaper to give it some clearance) Make sure it’s lubricated
when you reassemble it
the motor position can move a little bit, if it’s too close to the idler
(especially on 4.0 arms) it can be very tight
the gear on the motor may not have the setscrew tight and can be grinding the
plastic, which gets into grease and becomes a sludge that gums everything up
Thank you for clarifying where I needed to clean the arm up. I had the retraction force set to 1100 so the lower right arm would retract. All the others would retract fully at 900. Also interesting to not, when extending, the lower right belt always moved a little slower than the upper right. Inside I saw no signs of griding. I spent the morning taking that arm apart, sanding it, reassembling, testing, sanding more, etc. When I initially took it apart it was difficult to move the spool, now it moves smoothly. I was able to retract it fully with only the 900 setting that I dialed the rest down to.
If they retract at 900 you should be OK, but they may get tighter over time. Most of us have had to do this to all arms. You need to run Find Anchors again, the tight arm will have affected the measurements.
Thank you for clarifying where I needed to clean the arm up. I had the retraction force set to 1100 so the lower right arm would retract. All the others would retract fully at 900. Also interesting to not, when extending, the lower right belt always moved a little slower than the upper right. Inside I saw no signs of griding. I spent the morning taking that arm apart, sanding it, reassembling, testing, sanding more, etc. When I initially took it apart it was difficult to move the spool, now it moves smoothly. I was able to retract it fully with only the 900 setting that I dialed the rest down to.
1100 should not be a problem, if you were talking 1800+ I would worry about it.
However I¢m stuck in extended state.
the 2nd log you posted has this:
[MSG:INFO: Requesting state change from Belts Extended to Taking Slack]
[MSG:INFO: Succeeded]
[MSG:INFO: Current state: 5]
[MSG:DBG: Belt positions marked as stale/invalid in NVS, state=5]
MINFO: { “homed”: true, "calibrationInPr09.79, “tr”: 1804.47, “br”: 1803.57,
“bl”: 1806.09, “etl”: -0}
[MSG:INFO: Measured waypoint 0]
[MSG:DBG: Waypoint 0 coordinates: X=0.000 Y=0.000]
[MSG:INFO: Center point deviation: TL: 3.873 TR: 2.941 BL: 3.060 BR: 12.416]
[MSG:ERR: Center point deviation over 12.000mm, your coordinate system is not
accurate, maybe try run]
that is saying that when it retracted the belts, the resulting belt lengths,
combined with the anchor positions, did not compute.
This means that:
the magnet is slipping so it’s not recording the belt movement properly
you do show a retract/extend/retract cycle that looks good, which reduces the
probability of this.
or
the frame is flexing so the anchor points are not the same place they were
when you did find anchors
can you post pictures of your frame (you may have and I’m just not remembering
them)
or
you powered the machine down between the find anchors and now and when you
powered it on, the belts were too far off
possible, but I would have expected it to be off more than it currently is.
I think I was pretty careful about putting the encoders back together. But with all the other things I have to do in a normal day it wouldn’t surprise me if I swapped it. I hope I do not have to disassemble it again.
The width of it is 10’, the height of the cutting plane is 8’3”. The top mount is a set of door hinges. That allowed me to change the angle that it is sitting at. Normally it is at 20 degrees. yesterday I was able to lift it and set buckets under it to raise the angle. That is when I was able to cut the 4.1 that is on the lower left of the sled and the JB coaster that is directly above the sled. The 4.1 that is above and to the left of the sled was cut with it in the 20 degree offset position. It does cut, it’s just few and far between cuts.
The cutting plywood is also shifted to the right to bring the area I am working on closer to the center of the cutting surface. As long as I put a piece of material that is the same thickness to the left of the board, the sled does not tip as it approaches the edge. The white cardboard on bottom is to keep the belts from catching on the various edges that are there during the extension, retraction, and calibration.
I do not believe the bottom is flexing. Currently there are 3 10’ 2x4’s holding it straight in place. The anchors do not seem to be wiggling.
I guess the next step is to follow Ian’s advice and to run Find Anchors again. I’m going to give that a shot and then come back to see what else I need to do.
Thank you all for your help,
