Snapped belt, bent anchor point

Troubleshooting
1

Hi,
this is my first post here, as I’m just getting started with my Maslow 4.1. Yesterday I wanted to do my first cut, but it didn’t go well. Here’s what’s happend:

Setup: I have drilled holes into my concrete courtyard (roughly 3000x4000 mm) and installed 8mm bolts, because that’s what I had at hand. To prevent the belts from rubbing on the ground and collecting too much dirt, I added spacers to keep the ends of the belts about 20mm above the ground.

Calibration: I did that a few days ago and I think this went fine. Fitness 0.59, tlX=44, tlY=3229, trX=3850, trY=3179, blX=0, blY=0, brX=3830, brY=0. Unfortunately I haven’t saved all the serial output, so I can’t say anything about the stiffness measurement or other information printed during calibration. I guess that’s part of my learning curve …

Yesterday’s crash: Temperature was about freezing. I wrote a simple gcode file to move to the four corners of my part so that I could align my wood. Extend the belts, take up slack, Maslow says it is at -89, -5 (out of my memory, maybe there is a sign or coordinate flip, but that’s the numbers). Maslow jogs fine, I start my gcode and as Maslow approaches the first corner, I push “pause”. Maslow stops and loses network connectivity. I can’t reconnect, so finally I decide to power cycle Maslow. I’ve seen WiFi problems in cold conditions on ESP8266 and (less) on ESP32 before, and there is another thread here mentioning it: Communications and belt retraction issues on a cold day.

Anyways, after the power cycle Maslow reconnects to WiFi, so I do “release tension, take the belts off the anchors, retract belts, extend belts, put them back on the anchors, take up slack” (is this actually required after every power cycle?). Maslow reports “0,0,0” as position, which I find surprising, so I try to jog in y-direction. First time nothing happens, so I try jogging again. Now one belt snaps just a few mm from its anchor point.

After the belt had snapped, I realized that the bolt I used as anchor is now bent by ~30°. One other anchor bolt was also bent, but much less. I believe that this bending contributed to the snapping belt by causing much more tension on one side of the belt than on the other side, thus tearing it across its width rather than just pulling along the main direction of the belt.

Now I wonder:

  • Did I underestimate Maslow’s pulling force and my 8mm bolts are too weak (especially as they are not supported at the top)? Do I need stronger anchors?
  • Or does this sound like an issue with the belt tension measurement?
  • Did the cold weather contribute?
  • After the crash I also realized that Maslow complained about a “center point deviation (BL -13.2, BR 6.7)” when setting it up again after the power cycle. Was that the reason it didn’t want to jog? Were the bolt already bent after taking up the slack (they were still straight when I put the belts on)?

Thanks a lot,

Hendrik

2

hendrik wrote:

Hi,
this is my first post here, as I’m just getting started with my Maslow 4.1. Yesterday I wanted to do my first cut, but it didn’t go well. Here’s what’s happend:

welcome

Setup: I have drilled holes into my concrete courtyard (roughly 3000x4000 mm) and installed 8mm bolts, because that’s what I had at hand. To prevent the belts from rubbing on the ground and collecting too much dirt, I added spacers to keep the ends of the belts about 20mm above the ground.

Calibration: I did that a few days ago and I think this went fine. Fitness 0.59, tlX=44, tlY=3229, trX=3850, trY=3179, blX=0, blY=0, brX=3830, brY=0. Unfortunately I haven’t saved all the serial output, so I can’t say anything about the stiffness measurement or other information printed during calibration. I guess that’s part of my learning curve …

the calibration should probably be done over, when you added the spacers to keep
the anchors out of the dirt, you probably ended up with a different height from
the anchors to the sled, so the Z offset values in the settings need to be
adjusted

Yesterday’s crash: Temperature was about freezing. I wrote a simple gcode file
to move to the four corners of my part so that I could align my wood. Extend
the belts, take up slack, Maslow says it is at -89, -5 (out of my memory,
maybe there is a sign or coordinate flip, but that’s the numbers). Maslow jogs
fine, I start my gcode and as Maslow approaches the first corner, I push
“pause”. Maslow stops and loses network connectivity. I can’t reconnect, so
finally I decide to power cycle Maslow. I’ve seen WiFi problems in cold
conditions on ESP8266 and (less) on ESP32 before, and there is another thread
here mentioning it:
Communications and belt retraction issues on a cold day.

Anyways, after the power cycle Maslow reconnects to WiFi, so I do “release
tension, take the belts off the anchors, retract belts, extend belts, put them
back on the anchors, take up slack” (is this actually required after every
power cycle?). Maslow reports “0,0,0” as position, which I find surprising, so
I try to jog in y-direction. First time nothing happens, so I try jogging
again. Now one belt snaps just a few mm from its anchor point.

you are the second person we’ve had report that the belts snapped in cold
weather. One can be coincident, two is looking like a problem @bar

After the belt had snapped, I realized that the bolt I used as anchor is now
bent by ~30°. One other anchor bolt was also bent, but much less. I believe
that this bending contributed to the snapping belt by causing much more
tension on one side of the belt than on the other side, thus tearing it across
its width rather than just pulling along the main direction of the belt.

wow, pulling hard enough to bend an 8mm bolt, it may be that you are just plane
pulling too hard.

After you setup again and are ready to do another calibration, touch base with
us to help make sure your Z offset values are plausable and please post the
resulting maslow.yaml file after you finish calibration

Now I wonder:

  • Did I underestimate Maslow’s pulling force and my 8mm bolts are too weak (especially as they are not supported at the top)? Do I need stronger anchors?
  • Or does this sound like an issue with the belt tension measurement?
  • Did the cold weather contribute?
  • After the crash I also realized that Maslow complained about a “center point deviation (BL -13.2, BR 6.7)” when setting it up again after the power cycle. Was that the reason it didn’t want to jog? Were the bolt already bent after taking up the slack (they were still straight when I put the belts on)?

I think I answered these questions above, if not, ask for clarification.

David Lang

3

Hi Dave,

thanks for your quick response!

The spacers have been there during calibration, too, and I don’t think it’s a z-offset issue, because these angles are small and if that mattered we’d have to recalibrate for every material thickness we want to cut.

Anyway, I will of course calibrate again after fixing the bent anchor points (tomorrow, weather permitting).

I tried attaching my maslow.yml from the first calibration, but as a new user I don’t yet have permissions for uploading files to the forum. I’ll try again after the repairs. Here’s the relevant section (I hope …):

Maslow_vertical: false
Maslow_calibration_grid_width_mm_X: 1500.000000
Maslow_calibration_grid_height_mm_Y: 800.000000
Maslow_calibration_grid_size: 7
Maslow_tlX: 13.100000
Maslow_tlY: 3014.000000
Maslow_trX: 3980.800049
Maslow_trY: 3013.100098
Maslow_blX: 0.000000
Maslow_blY: 0.000000
Maslow_brX: 3970.000000
Maslow_brY: 0.000000
Maslow_tlZ: 100.000000
Maslow_trZ: 56.000000
Maslow_blZ: 34.000000
Maslow_brZ: 78.000000
Maslow_Retract_Current_Threshold: 1300
Maslow_Calibration_Current_Threshold: 1200
Maslow_Acceptable_Calibration_Threshold: 0.450000
Maslow_Extend_Dist: 2400.000000
Maslow_beltEndExtension: 30.000000
Maslow_armLength: 123.400002
4

hendrik wrote:

The spacers have been there during calibration, too, and I don’t think it’s a z-offset issue, because these angles are small and if that mattered we’d have to recalibrate for every material thickness we want to cut.

the problem is that the z offsetsare assumed, not measured, and since the
calibration is trying to map the measured belt lengths with the assumed Z offset
to figure out the X,Y coordinates of the anchor, errors in the assumptions at
that point cascade to bigger errors than just having it wrong at cut time.

But I admit I’m reaching a bit here, we just don’t have experience with belts
breaking

Anyway, I will of course calibrate again after fixing the bent anchor points (tomorrow, weather permitting).

I tried attaching my maslow.yml from the first calibration, but as a new user I don’t yet have permissions for uploading files to the forum. I’ll try again after the repairs. Here’s the relevant section (I hope …):

That is the right section. the fact that tlX is 13mm says that your frame is
quite a bit out of square (it should be able to handle it, but…)

David Lang

5

I think that this is could be a factor. I’ve actually removed the Pause button from the next version because it’s causing strange behavior

At this point the machine shouldn’t allow you to move because it knows that something is off. I’m not sure what exactly could have happened to make it move.

I’m working on a full re-write of the state machine part of the code right now which should help to eliminate weird behaviors like this.

The machine can pull quite hard, but if everything is working normally it shouldn’t pull that hard. The type of concrete and the type of anchor are going to have a big impact on how hard is too hard

6

That’s good to know, thanks!

I’m actually not sure if it tried to move. Maybe the bolt slowly bent more and more, and at some point the belt snapped, unrelated to my (unsuccessful) attempts to jog.

By now I’ve come to believe that the bolts were simply too weak, especially with the spacers I had. With the spacers the belts didn’t pull right where the bolt comes out of the concrete, so there was some leverage.

Today I’ve fixed the Maslow, made the anchors much beefier, and ran a new calibration. I can already say that it looks better than before, but there is one more thing I’d like to check tomorrow before writing a full report. Today the sun went down and I ran out of time.

1 Like
7

hendrik wrote:

By now I’ve come to believe that the bolts were simply too weak, especially
with the spacers I had. With the spacers the belts didn’t pull right where the
bolt comes out of the concrete, so there was some leverage.

you need to be very sure the bolts don’t move, that is going to require that
they be tightened down against something (either the bottom of the hole, or a
nut tightened against the floor, and if you have a stack of washers to raise the
location of the anchor, put the nut above the washers to stiffen things)

you also need to make sure the Z offsets in the maslow.yaml file are correct for
your anchors (before calibration)

David Lang

8

Here comes the promised update. I decided to open a new thread about the new calibration and dimensional accuracy (see Feedback on calibration and resulting dimensional accuracy). The conclusion for this topic is that my first attempt to use single 8mm bolts as anchor points was too weak. My new anchors are wooden blocks (40mm high) bolted to the concrete, with a 10mm hole where I can drop a M10 bolt in to attach the belt. These anchors seem to work very well:

anchorpoint
anchorpoint1000×750 209 KB

2 Likes
9

the fact that you show different flex in different directions seems significant
to me.

Also, I would suggest that you consider putting different height blocks on each
corner to make the anchors level with each of the arms when doing the
calibration.

I suspect that that would significantly improve the calibration accuracy

David Lang

10

That’s on my list for the future. The current installation is temporary and I didn’t want to spend too much effort on it. I plan to build a proper workshop in fall. Then the Maslow will have a new home with level anchors and coordinate system marks on the floor.