4.1 Calibration Issues and Questions

[boslaw]

@bar

So, things are back to not working again.

I installed a bit in the router - I had to raise the z axis up all the way to do that.

I extended the belts and applied tension. One belt stays slack (not the same arm that previously had issues).

I get a center point error again - measurements off, try calibration.

I can’t calibrate without taking the bit out and lowering Z all the way.

I do this anyway (more wasted time).

I fail calibration.

Back to square one. So so frustrated. I have spent an entire 8 hour day fiddling with this machine disassembling reassembling, extracting and retracting belts, etc etc.

Reload good yaml. Add bit. Lower til bit is barely touching the wood. Retract All, Extend All, attach to Frame, Apply Tension - Center point deviation over 12.000mm - can’t do anything.

If it’s not one specific arm that is the problem (pretty sure it’s not at this point) then there is something else wrong with this thing.

To top it off, my power block seems to be randomly losing power now. Mostly only happens after I’ve tried to calibrate or apply tension. Things won’t work, I’ll go to remove my belts so I can retract them, and then the machine will lose power. Fiddle with the power block - back to normal again. This has NOT happened during calibration or any other time so that doesn’t appear to be the problem - it’s just another problem.

NOTE: my tape measure measurements according to @dlang website are:
tlX -27.6
tlY 2064.90
trX 2924.3
trY 2066.5
brX 2953.2

The Maslow auto calibrated measurements are:
tlX -5.0
tlY 2427.399902
trX 3653.699951
trY 2439.300049
brX 3651.800049

I don’t know why these numbers would be so very different.

[Deadset]

Please try increasing your calibration power, i was having similar issues with dodgy calibrations and centrepoint issues when applying power, raisinf calibration force worked for me

[boslaw]

@bar @dlang @Deadset

Here’s another failed attempt at calibration because I’m a glutton for punishment.

This one looked like it was going to succeed until the Maslow just turned itself off midway through the process.

maslow calibration attempt SD 480p

Sorry about the aspect ratio - I turned the phone to horizontal part way through so I had to cut the clip to rotate it.

Log is here:
Maslow-serial-14.log (22.5 KB)

At the beginning of the log, I was raising my Z to remove the bit I had inserted yesterday.

Once the bit was removed, I retracted all and extended all belts and attached them to the frame.

My config settings were:

I then pressed Calibrate.

The machine attempted to Calibrate, running through waypoints 1-5.
It tried Calibrating for a few seconds, then spit out some calibration values then it looks like it tried calibrating again on its own

It measured more waypoints (way more than the 3x3 would require) and started running calibration again.

That lasted for 10 minutes, then it spit out calibration values, said it was measuring yet another waypoint (15) and promptly turned itself off.

WTF?!

If the board or something isn’t bad then this just seems like it isn’t ever going to work for me. I appreciate everyone’s suggestions (try this, try that) but if there’s no objective reason why one suggestion should work over another, then I’m really just wasting time (and this thing takes a TON of time by the time you retract, extend, connect, disconnect, disassemble, calibrate, etc.

[caddad]

I received my new belt and tried to recalibrate, but got errors about the yaml. restarting didn’t help and when I tried to delete the yaml and index (with the intent to update to 1.04.1), it asked if I wanted to confirm deletion of a text.nc file. I cant seem to delete the yaml or index to start a fresh calibration.

[caddad]

Here;s the serial file
Maslow-serial(1).log (3.1 KB)

[boslaw]

I get the same thing on my 4.0 when I delete the yaml - test.nc is deleted, no mention of maslow.yaml

[Justin_Walker]

Same here!

[Otto_P]

Same here!

[fdoepke]

Before reading this thread I started trying calibrate my Maslow 4.1 upgraded to 1.03. And all I get is that the machine is making an absolute mess of the belts. I am in a horizontal sheet of 8x4 plywood and it starts tightening one belt then another. Then it seems release some belts and they all just tangle and I have to take the whole machine apart. I am getting very frustrated with this as I don’t know where to start fixing it. Sounds like it only works in vertical? Or is my frame too small?

[caddad]

@fdoepke I lost a belt to that. Apparently you need to assist the belts during horizontal calibration and pull them out as they loosen. After calibration, this shouldn’t be an issue. This isn’t in the instructions and I haven’t had a chance to confirm due to the yaml issue I now have.

[boslaw]

@caddad

I cant seem to delete the yaml or index to start a fresh calibration.

Download a new yaml from the v1.0.4 files. Upload to your Maslow. It will overwrite the old one. You can confirm by uploading and then checking the config in FluidNC. The numbers should all be standard instead of your calibrated numbers.

[caddad]

@boslaw Can you provide the link to the file?

[dlang]

the machine first pulls all the belts tight.

Then it tries to move to the next position. Since it doesn’t know where it is,
it feeds out some belts (more than it needs to, to be sure it feeds out enough)
and then pulls the belts tight in it’s new location.

the very oldest firmware did this in a way that fed out a LOT of belt because it
was trying to make big movements. Long ago this shifted to move only from one
point to the next closest point, so not a lot of belt needs to get fed out

after it does the first pass of doing this (I think the 3x3 grid), it then has a
better idea where the anchors are, and so in the next pass it feeds out far less
belt.

what firmware version are you running?

David Lang

[caddad]

I updated the yaml and index, now i get these errors. I turned off the power, unplugged and replugged all 8 encoder cable connections to no avail. First it was the BR, then after reconnecting the cables its the TL that is not reading.

[dlang]

that’s an error I’m not familiar with, we need @bar to comment.

David Lang

[boslaw]

@dlang @bar

Any ideas on my last post or do you think my Maslow will just never work?

[dlang]

many people have it working, so we just need to figure out what you are doing
differently.

I am in the middle of moving (have to be out in a couple of days) so my
responses are going to be less common in the next week or so.

Bar is going to do a step-by-step video, when that comes out, try to follow it
and report what is happening different than what the video shows.

David Lang

[boslaw]

@dlang

Thanks David. I did post a video showing my latest attempt, along with the log file. That should be enough to determine what (if anything) I’m doing differently than everyone else. I’m not convinced I’m doing anything different based on all of the advice and instructions but I guess it’s always possible.

[danFromJersey]

@dlang
I feel like I am struggling in very similar ways to @boslaw. I have a horizontal frame with 3/8 drop-in anchors on a 12’x8’ grid. The following is the procedure that I follow when attempting to calibrate.

1. Turn on Maslow
2. Press RETRACT ALL
3. Press EXTEND ALL (and extend the 4 belts to 2150mm)
4. Press RETRACT ALL (again, so that I can make sure all 4 of my offsets are small)
5. Press EXTEND ALL (and extend again)
6. Connect the 4 belt ends to the 3/8 bolts at the corners
7. Press CALIBRATE
8. Refresh the browser page and press CALIBRATE (again, because the 1st time doesn’t seem to send the command)
9. Watch as the fitness calculation keeps trying and trying but failing to get a good frame size

I have measured the bolts several times. Here is what I have,
bl->br = 143.125" = 3635.38mm
tl->tr = 143.875" = 3654.43mm
bl->tl = 96.375" = 2447.93mm
br → tr = 95.875" = 2435.23mm
bl->tr = 172.375" = 4378.33mm
br->tl = 173.0625" = 4395.79mm

Using Anthropic’s Claude 3.7, I got the following coordinates for the corners
bl = (0, 0)
br = (3635.38, 0)
tl = (20.79, 2447.25)
tr = (3658.93, 2433.91)

I have the following questions about calibration.

1. If I enter these numbers into the maslow.yaml file, does the calibration use these measurements as the starting point?
2. Could someone give me a pseudo-code version of calibration?
3. I set the calibration grid to 500x500 with a grid size of 5x5. Shouldn’t the calibration measure 25 points? It is only measuring 5 points?

I have attached the serial printout from the calibration. Hoping this helps someone (and in turn helps me!). Because I tend to have to refresh the browser to get the calibration to start, you lose the output from step 1-7 above.

Thanks!!
Maslow-serial.log.txt (34.6 KB)

[dlang]

Dan Strassberg wrote:

7. Press CALIBRATE
8. Refresh the browser page and press CALIBRATE (again, because the 1st time doesn’t seem to send the command)

for those who are having problems with their browser:

when you connect to the maslow SSID, you connect and your browser detects that
it doesn’t have Internet access and asks you to ‘login to the network’

the browser that it opens at this point is NOT able to do everything, it’s
deliberatly crippled by the OS/browser ‘for your protection’ (in case the
hotel/etc you are connected to is a bad guy)

you need to open a new browser window (after telling your OS/browser “yes, this
doesn’t have Internet access, I want to use it anyway”)

9. Watch as the fitness calculation keeps trying and trying but failing to get a good frame size

I have measured the bolts several times. Here is what I have,
bl->br = 143.125" = 3635.38mm
tl->tr = 143.875" = 3654.43mm
bl->tl = 96.375" = 2447.93mm
br → tr = 95.875" = 2435.23mm
bl->tr = 172.375" = 4378.33mm
br->tl = 173.0625" = 4395.79mm

Using Anthropic’s Claude 3.7, I got the following coordinates for the corners
bl = (0, 0)
br = (3635.38, 0)
tl = (20.79, 2447.25)
tr = (3658.93, 2433.91)

I have the following questions about calibration.

1. If I enter these numbers into the maslow.yaml file, does the calibration use these measurements as the starting point?

yes it does, but since the purpose of calibration is to find these numbers, you
should be able to just start operating.

2. Could someone give me a pseudo-code version of calibration?

measure the belts at a bunch of points. The more points and the wider the area
they cover, the better (until you hit an area where the belts aren’t straight
with the arms)

make a guess at the frame size (if your yaml files are the default)
start matching the measured arm lengths against the anchor positions, nudge them
towards a better fit.

repeat

3. I set the calibration grid to 500x500 with a grid size of 5x5. Shouldn’t the calibration measure 25 points? It is only measuring 5 points?

In that case, you aren’t doing the full calibration. The fact that you are
having to reload the browser and hit calibrate again is telling me something is
wrong. There is state maintained in both the browser and the firmware during the
calibration, and trying to restart it doesn’t work well.

I have attached the serial printout from the calibration. Hoping this helps
someone (and in turn helps me!). Because I tend to have to refresh the
browser to get the calibration to start, you lose the output from step 1-7
above.

someone other than me will need to look through that.

David Lang