-
Attempting to calibrate the Maslow on a vertical table.
-
Connected to Maslow AP
-
Retract all (success)
-
Extend all (1800mm) success
-
Mounted Maslow on a vertical table (4 anchors)
-
Run find anchors - get an error, see serial
Maslow-serial.log (4.0 KB)
-
somewhat lost as to what to try next.
extract:[MSG:INFO: Requesting state change from Retracting Belts to Belts Retracted]
[MSG:INFO: Succeeded]
Find Anchors
[MSG:INFO: Requesting state change from Belts Retracted to Finding Anchor]
[MSG:INFO: Cannot start calibration until the belts have been extended]
Find Anchors
[MSG:INFO: Requesting state change from Belts Retracted to Finding Anchor]
[MSG:INFO: Cannot start calibration until the belts have been extended]
Extend All
[MSG:INFO: Extending all belts]
[MSG:INFO: Requesting state change from Belts Retracted to Extending Belts]
[MSG:INFO: Succeeded]
Stop
I think that this is the issue here:
[MSG:INFO: Requesting state change from Retracting Belts to Belts Retracted]
[MSG:INFO: Succeeded]
Find Anchors
[MSG:INFO: Requesting state change from Belts Retracted to Finding Anchor]
[MSG:INFO: Cannot start calibration until the belts have been extended]
Find Anchors
What is happening here is that when you press âRetract Allâ the machine is trying to pull the belts all the way in to set their home position.
You only want to use retract all when the belts arenât connected to the frame.
A fair bit has changed since this video, but this is a good starting point for understanding how things should work:
Updated firmware, index files. ( am I required to update the Yaml file as well?)
Retried, retract, extend, mounted on board, then find anchors.
Seemed to go further and began to calibrate.
I did notice a few skips where the sled was lifting at one side at different times???
Is this a vertical = false/ true issue setting?
There is no need to edit maslow.yaml normally. A successful Find Anchors will do that for you.
Just to be clear on the exact sequence initially (until you get a successful Find Anchors)
- Click on Setup
With Maslow not connected to Anchors
Retract - Extend and connect to anchors
- Click On Find Anchor Locations
Maslow should drop a little then pull all belts tight (here it is using gravity to determine if Maslow is vertical or horizontal).
It should then do a slack belt movement to pull around a rectangle. (here it is making a rough estimate of the approximate location of your Anchor Point). It sends the data back to your computer for processing.
Next it should go around your work surface 3 times, pausing after each circuit to send data back to your PC for processing. Each circuit is larger than the previous. Belts should stay reasonably tight during this process. Maslow tipping may affect the accuracy. - After 3 iterations it will make 2 more stops and end up back in the center.
- If successful it will then go to Ready To Cut.
Before trying to cut any thing I suggest you raise the Z axis to a reasonable value (40mm) and set Z Home. Load a GCode file (without a bit in) and run it, just to get a feel for how the Maslow operates
When you are comfortable load something to cut
Adjust the Z until bit is just touching the material Set Z Home AND
Have fun.
this is the serial output if this helps - I do see. âCalculated Fitness Too Lowâ, my frame is based on this design but in 15mm plyboard (2420 height by 3000 wide)
Maslow-serial.log (52.9 KB)
Maslow+Frame+Metric frame.pdf (142.4 KB)
Please just stick with one topic, until we get you up and working. Otherwise, it makes it difficult to remember if we have asked for particular information before. This one will do. Can you put a copy of your maslow.yaml file here. I think you have posted it before, but I canât find it.
What I want to try is seeding a standard frame into the maslow.yaml with the following values:
tlX: 0
tlY: 2400
trX: 3000
trY: 2400
blX: 0.000000
blY: 0.000000
brX: 3000
brY: 0.000000
and see what it does when you run Find Anchors.
Can you also try (on the bench for this) retract, extend and pull out the belts, repeated 3 times, then post the resulting serial file here. What we are looking for is deviations less than 1 for the 2nd two cycles. This will give us an indication of whether the arms are OK or not.
Also, what retraction force have you set?
Retraction force is used during Find Anchors and the lower it is the better, consistent with all belts retracting completly.
I will send the yaml file tomorrow - the unit is at my workshop
most settings are factory - force is 1300 from memory
when - the pieces are attached to the belts there os a slight overlap - mine are quite small- could this impact this ?
Not sure what you mean here
What you can try here is set retract force to 700, extend and retract. If all belts retract, excellent! Otherwise increase it by 100 and try again. What we are aiming for is the lowest possible value where Maslow retracts all belts consistently.
What do you mean here. Are you talking about where the belts attach to Anchors? Maybe a photo would help.
yes - belt anchors- i did trim (the are all under 5mm) so I think it maybe a non-issue
I changed tension to 700- re ran the process - retract- extend- find anchors, appeared to work this time- i did not a slight lift on the sled once during calibration, however nothing like the previous one. (log attached)
Maslow-serial (2).log (32.5 KB)
Not a great result, but a pass. Re-running Find Anchors may improve it a bit. Belts need to rotate around Anchor points, too tight will cause problems.
The anchors are a little tight for rotation, may have to attend to this.
Can I simply re run locate Anchors again.( no need to retract extend) release tension first?
What do you mean by not great?
They really need to be free to rotate. The maths depends on the belts going straight from the Anchor Point to the centre of the router. Any angle deviation messes up the accuracy.
Jog to the middle (or set home to middle (0,0) then Click Home. Release tension then you can re-run Find Anchors. It will go through the same routine again, but use the existing values for the initial seed and hopefully get a bit better.
Originally the limit for fitness was 0.5. This was lowered to 4.5 a long time ago. Most people are getting fitness values above 0.8. The higher the better. You can have a look at this simulator
Maslow CNC Calibration Simulation
Paste the CLBM line from your serial log file into simulator.
Data points in the simulation should look pretty rectangular after a few seconds. Yours look distorted.
What it is attempting to do is find the optimum intersection point for all of belts in each position, much the same way as GPS works.
Understand - i run these and see where it all lands
Checked Anchors, some were less âfreeâ the man others. These are the 3D printed anchors from the not shop. Additionally I purchased the quick release pins from Amazon ( link in not shop also). It appears to be sticking on quick-release pins more than the 3D print. To confirm, is it expected the pins rotate in the 3D printed anchor points? ( and the belt anchors rotate with the pins?)
Also I noted the Maslow sled, regardless of location apart form centred home, is slightly tilted- normal? Fix?
Maslow-serial (1).log (21.4 KB)
Ian Tolond wrote:
Checked Anchors, some were less ¥free¢ the man others. These are the 3D
printed anchors from the not shop. Additionally I purchased the quick release
pins from Amazon ( link in not shop also). It appears to be sticking on
quick-release pins more than the 3D print. To confirm, is it expected the pins
rotate in the 3D printed anchor points? ( and the belt anchors rotate with the
pins?)
it doesnât matter which part of this is pivoting, but the belt end should rotate
so that as the belt is pulled, the pivot point of the belt end is at the pin.
the pin, the belt, the arm, and the center of the router where the bit is are
all in a direct line
Also I noted the Maslow sled, regardless of location apart form centred home, is slightly tilted- normal? Fix?
the sled can rotate wihtout the bit moving, so itâs rotation doesnât matter.
Now, as you move around, one or more arms will hit the frame, and what should
happen is that when one arm hits the frame, the frame conforms to the movement
of the arm. If it canât (either because there is another arm hitting the frame,
and they canât move, or because cable/hose/etc stops the rotation), then the
fact that the arm and belt are no longer straight does introduce some error.
Itâs usually not enough to matter, but enough of it can matter. the tool at
http://lang.hm/maslow/maslow4_frame.html lets you put in dimensions and see the
places where adjacent arms will hit along sides and opposite arms will hit in
the corners, the bands you see indicate the amount of angle between the arm and
the belt. above the graph is a chart that gives rough indication of how much
error in effective belt length is introduced at each band. As you see, going a
few bands in is unlikely to be noticable, but going a lot of bands in is more
likely to cause grief.
David Lang