Laserbot - mLaser software issues

svogl · 2017-02-27 15:38:10 UTC

I received a Laserbot the last days and it assembled fine - great video instructions! The first test was the big M pattern with the scales - it came out fine and the scales were perfectly right - great!

The mLaser software looked nice (just wondered why I would need a password?), but here the troubles start: I intend to do some “serious” cutting (pcb stencils etc.), so I tried to import a Kicad print (via print to PDF/Inkscape detour), printed on paper an got

the original PCB is 99x99mm --> HOW DO I SCALE this to get correct results? Inkscape exports with full page as the viewport, computing a scale vector is plain impossible (tried different files/settings)?
I also tried the 40x40mm test square (I assumed you’d supply a file with a square at that size?..) and scaled it, and got:

hmmm – the software seems to drop waypoints – also, if you look at the first picture, yo can see random dropped segments (all rectangles are closed in the input file) – I reduced travel speed to 800mm/min to rule out mechanical issues.

It looks as if it is not a tension issue either, connected tracks print more or less ok:

(svg extracted from from https://en.wikipedia.org/wiki/Space-filling_curve#/media/File:Hilbert_curve.svg )

BUT if you compare you can see missing segments at the beginning and end of the figure, a dropped corner point (bottom left) and a series of dark dots - seemingly communication stalled here?

Does anybody have a serious driver software? Otherwise I’d script my own g-code generator
S

P.S. mLaser runs on a Windows 10 machine; I compiled the github repo on a Linux Mint machine, but it did not show any data in the preview window. Will retry with another laptop later.

tech_support · 2017-02-28 11:34:17 UTC

Hi svogl,

What are the original images for the first cut result and second cut result?
If you want to cut a image to a particular size, you can zoom in and zoom out the image which imported to the mLaser software.

When you test the Laserbot, do you use the latest version mLaser software V2.3?

svogl · 2017-02-28 13:27:07 UTC

Hi
sure, I saw that I can zoom around, the point is that I need mm-exact prints.

You can grab the files from here:
http://voxel.at/voxel/sites/default/files/Hilbert_5_10x10.svg
http://voxel.at/voxel/sites/default/files/paste_at42q1070.svg
http://voxel.at/voxel/sites/default/files/Square%2BTest.svg

I’m in office at the moment, but to my knowledge I used the latest software, but will check later.
Btw, I printed the Hilbert-curve unscaled at first, it came out at ±35x35 mm with precise cuts but the same errors (ends missing, the cut edge):

-> could be an svg parser issue as well…

svogl · 2017-02-28 13:27:44 UTC

Btw, @ belt tensioning: There are some nice hacks around, like this one:

svogl · 2017-02-28 15:33:43 UTC

Btw, I compiled MLaser on a Ubuntu 16.04.02 machine with QT5 (had to add a #include <qiodevice.h> to karchive), but when I load an SVG file, I don’t see any update in the UI, it simply stays blank.

Which QT version are you using?

furthermore, it would be great if you could check-in a default set of .ini files (material definitions, english locale

tech_support · 2017-03-01 08:32:30 UTC

Hi svogl

Thanks for your info.

For the mm-exact prints, here we tested several square image, the prints are correct.

As for the other issue, I have provide feedback to our engineers and will let you know when there is update.
Btw, is it possible for you to attach these svg files directly? It may be help for us to test. it looks like we can’t save the svg files from the links you provided. Thanks!

svogl · 2017-03-01 09:24:26 UTC

Would like to but your page says:

But you can pick them up at my page:
http://voxel.at/voxel/blog/makeblock-laserbot-laser-engravercutter-first-impressions

Hope this helps
Simon
p.s. I’d really like to get the application running under LInux, that’s our main working environment; if you have any clues I can give it a try.

charlieb · 2017-03-01 15:28:53 UTC

If you just want to go straight gcode and generate your tool paths with some other software (e.g. Inkscape) this thread and firmware might help:

svogl · 2017-03-01 19:00:55 UTC

thanks for the pointer – I saw the thread and this will be my next target, but I wasn’t sure if the XY plotter parameters are equivalent to the LaserBot & I thought it might be good to have a solution for all users, not only the hacking fraction
Best
Simon

tech_support · 2017-03-02 06:46:46 UTC

Hi Simon,

Thanks for the link.
We have recorded this issue, our engineers will work on finding a solution to this issue. Will inform you when there is progress.

For the mLaser software, there are only Windows and Mac version, no linux version.

svogl · 2017-03-02 19:16:57 UTC

@Linux - no problem, I’m into CS. The program compiles, but when executing, QT complains that several slots are unknown and imported graphics does not show up, nor do the zoom buttons work.
Which QT version have you been compiling this against? Will try to replicate.
/s

tech_support · 2017-03-03 07:02:46 UTC

Hi svogl,

It is version 5.3.1

svogl · 2017-03-04 09:39:18 UTC

thanks for the update; Ubuntu 16.04.2 uses 5.5.1;

On my Mac (Sierra 10.12.3) the application runs, but only if I start it via terminal. I copied it into the Applications folder, but when trying to open the following appears:

Started via terminal, I see the following first lines:

simac:src simon$   /Applications/mLaser.app/Contents/MacOS/mLaser
2017-03-04 10:32:04.444 mLaser[8010:251031] There was an error parsing the Info.plist for the bundle at URL <0x7f83177009d0>: NSCocoaErrorDomain - 3840
QMetaObject::connectSlotsByName: No matching signal for on_btnNext_clicked() "Bluetooth-Incoming-Port" "incoming port - Bluetooth-Incoming-Port" ""
QMetaObject::connectSlotsByName: No matching signal for on_toggleLaser_valueChanged(int)
QMetaObject::connectSlotsByName: No matching signal for on_btnOK1_clicked()
QMetaObject::connectSlotsByName: No matching signal for on_btnOK2_clicked()
QMetaObject::connectSlotsByName: No matching signal for on_comboBoxSerial_activated(QString)
QMetaObject::connectSlotsByName: No matching signal for on_btnItly_clicked()
...

trying to build on my Mac, I am stuck at:

simac:MLaser simon$  /opt/local/libexec/qt5/bin/qmake src/MLaser.pro 
Project WARNING: Fervor: falling back to application name 'mLaser'
WARNING: Could not resolve Info.plist: 'Info.plist'. Check if QMAKE_INFO_PLIST points to a valid file.

Could you please check this file into the repo?
thx!

svogl · 2017-03-05 10:35:49 UTC

Thanks for the pointer @charlieb,
I am currently looking at options and found jscut:
http://jscut.org/
Anybody tried that already?
S

charlieb · 2017-03-05 21:57:32 UTC

Just had a quick look but it doesn’t support polylines so you have to convert them to paths in inkscape anyway so you might as well just use the inkscape gcode plugin anyway.

Just my 2c

tech_support · 2017-03-06 07:24:55 UTC

Hi svogl,

You may check this link.

krouviere · 2017-03-24 16:59:48 UTC

I am experiencing a similar problem. Using mlaser for any work that needs accurate scaling seems nearly impossible without a bunch of work to try to compensate for this.

Here’s an image showing 3 items. You can see the 30mm scale bitmap test file loaded in mlaser 2.4, an svg file loaded in mlaser, and the resulting cuts. I loaded the 30mm test file, ran the laser, then loaded the svg file and ran the laser without moving the paper or making any other changes.

Note that both images of the mlaser UI show a mm scale at the top. I have placed arrows pointing out the 30mm spot on the mlaser UI in both instances. I cannot seem to load a dxf or svg file with correct scale. Putting that aside for the moment, this test case should be easy to reproduce using mlaser 2.4. You can clearly see that according to the scale marks in the UI, the 3 boxes in the svg file should be smaller than the scale test (look at where the arrow points to 30mm on the scale). The svg file should be smaller than 30mm but it is much larger. It does not make any sense.

Nevermind for now the fact that the files won’t load with the correct scale (each of those squares is supposed to be 15mm wide). The scale in the UI does not even make sense. One step at a time… please fix whatever is causing the laser to cut at a different scale than what is shown on the rulers in the UI. Then I’ll worry about getting the dxf and svg files to load correctly with proper scale.

I’m looking into other options to make this cutter usable, like perhaps grbl or benbox “if” I can get those to work with this machine, or just replacing the controller with something that works with MACH3. Otherwise all I can use this for is just engraving art or other things where scale is not important.

Can you please start by ensuring that the scale on the rulers in the user interface accurately reflects the size that will be cut?

Thanks

P.S. Looks like I can’t put 2 dropbox links into this message so I will follow with another message with a link to the svg test file.

krouviere · 2017-03-24 17:00:22 UTC

Here is the SVG test file.

tech_support · 2017-03-28 11:02:02 UTC

Hi krouviere,

We have provide feedback of this case to our engineers, will arrange the further test. Thanks for your detailed explanation and files.
Besides, there is only mLaser software can control the Laserbot, there is no other software support this robot kit.

svogl · 2017-03-28 19:52:42 UTC

hi @krouviere,
I tried your svg with my approach - it is to scale at least:

(I test on old thesis prints)
I loaded your svg into inkscape, used the gcodetools plugin with default settings to create a gcode file. This is the reason for the burn marks at the beginnings of paths - there is a penetration sequence first.

I also see the poor laser power control as in your shot - segments are partially missing
At this stage, consider it a toy, not a tool; have not given up yet though…
S