Perfect built at last (after a lot of tuning) – oval circles and rectangulars

New Home Forum Troubleshooting Perfect built at last (after a lot of tuning) – oval circles and rectangulars

This topic contains 24 replies, has 3 voices, and was last updated by Profile photo of PureField PureField 2 months ago.

Viewing 25 posts - 1 through 25 (of 25 total)
  • Author
    Posts
  • #18964
    Profile photo of PureField
    PureField
    Participant

    I have been spending a significant amount of time trying to fix the initial issue where the Z axis assembly is not square.
    I finally found the right combination of screws to tighten and I am square in all directions. All distances and diagonals are perfect.
    Rails are parallel and all should be great.

    So I printed my test pattern but the Y axis seems to move too much making everything longer.
    Ovals vs circles, rectangles vs squares and so on. Plotted squares have 100% square corners but are not even length.
    I scanned the plot from the machine and overlaid it with the original in Inkscape. I attached the SVG.

    What can that be?
    Thinking about setting a variable for the multiplier to scale the Y axis down.

    Pictures from the Y-axis.

    #18969
    Profile photo of PureField
    PureField
    Participant

    X-Axis dimensions and center square.

    #18974
    Profile photo of Jasoroony
    Jasoroony
    Participant

    Painstakingly, I guess, but count off the tiny belt notches (mark them with a pen as you go, every 10 or 20) and measure the distance of the X’s and Y’s for the same count of notches from one end to the other. Your Y belts are longer, so there may be a bit more stretch in them.

    If you get different measurements on X’s vs Y’s, you could re-tension the belts or adjust the firmware… If the X’s are different to one another (and/or Y’s to one another) you will need to re-tension so they are at least “similar enough”. My machine isn’t rectangular, but square and the belts are about the same tension, so I’ve not noticed this, but now you bring it up I will be measuring and adjusting the firmware if needed, much the same as the extruder 100mm filament test that’s recommended.

    #18980
    Profile photo of vicious1
    vicious1
    Keymaster

    How far off is it?

    #19003
    Profile photo of PureField
    PureField
    Participant

    @vicious1 – the attache SVG shows the result of my test. I am attaching a screenshot of the SVG
    You can seee that the X direction is pretty close. The boxes on top are 90deg but not square. The circle in the middle is 50mm radius or 100mm diameter and is about 3mm off.

    Attachments:
    #19006
    Profile photo of PureField
    PureField
    Participant

    @vicious1 – the attache SVG shows the result of my test. I am attaching a screenshot of the SVG
    You can seee that the X direction is pretty close. The boxes on top are 90deg but not square. The circle in the middle is 50mm radius or 100mm diameter and is about 3mm off. I centered the plot on the top of the large circle and you can see the boxes are longer towards the top. The text at the bottom starts way lower too.

    Attachments:
    #19067
    Profile photo of PureField
    PureField
    Participant

    I have not yet been able to check the teeth on the belt. I had a few minutes to check the tension and it appears very similar and I was not able to stretch the belt in any direction to the extend necessary (3mm over 100mm) to cause this. Any other things I should check for next time I work on the machine?

    #19068
    Profile photo of Jasoroony
    Jasoroony
    Participant

    Everything appears to join up correctly in what you’ve drawn with the machine, so it’s not skipping steps. I’ll assume all the stepper motors are the same and working also, since if one of the 4 is off/different/bad it will not give you 90 degree corners in the drawing.

    That leaves the belts as the next thing to check, since they are what’s meant to give the machine precision in the X/Y drive mechanism.

    Sorry, I can’t help anymore than that. If they are all equal length for the teeth counted, there’s some bad magic going on in your machine!

    #19069
    Profile photo of PureField
    PureField
    Participant

    Thanks for the help. Makes sense, will take some time to check them tomorrow evening.

    #19078
    Profile photo of vicious1
    vicious1
    Keymaster

    The squares and circles seem to indicate a loose pen. What mount are you using and how much pressure is on the pen? When I use my pen mount its only 1-3mm past contacting the paper to apply minimal pressure, depending on the pen type

    The letter seem perfect but just started in the wrong place, were those all the same job, motors never turned off in between movements?

    The letters just seem to be disproportionately far from the intended location, compared to the others. The squares and circles are not all off in the same way, really just seems like the pen might be loose or too much pressure was used and you ware seeing a wobble, except the letters, they are way off.

    #19080
    Profile photo of PureField
    PureField
    Participant

    The pen is mounted in a small wood block screwed to the mount. it is pushed down with a spring to keep even pressure. it nas minimal play that accounts for wobble in the 1/2 mm range. The entire plot is about 4-5mm longer than it should be. It was printed in one continuous print.

    #19082
    Profile photo of vicious1
    vicious1
    Keymaster

    What were your settings for speeds and travel speeds? What software, are you sure you imported it correctly and didn’t scale it.

    Maybe try just a large square.

    #19097
    Profile photo of PureField
    PureField
    Participant

    I attached the sourcel

    #19126
    Profile photo of PureField
    PureField
    Participant

    Rebuilt the timing belts using steel wire instead of cable ties to remove the slack. Made them all the same length and measured the ends to be consistent.
    Test print shows the same problem however by amount but not percent. So this time the total discrepancy was 3mm over 400mm. The X axis is 100% correct.

    One thing that had changed since my initial work was the use of a new Arduino due to some end stop issues. Using the original Arduino shows only a discrepancy of less than 1mm. So there is a setting that is different between the software on the Arduino causing the problem. The old Arduino still has a problem with the end stops and I cannot seem to get it to print as it always starts the print by raising the print head. Likely due to a home gcode.

    How do I get the exact software for the Arduino that was on the one purchased?

    #19127
    Profile photo of vicious1
    vicious1
    Keymaster

    Marlin Firmware

    What were your settings for speeds and travel speeds? What software, are you sure you imported it correctly and didn’t scale it.

    Maybe try just a large square. Are you trying the same file over and over again? This could be a software CAM issue, or a speed issue, or a driver issue. The cable ties are rated to more than 40lbs, If you put anywhere near that on the belt you are going to stall the steppers.

    #19128
    Profile photo of PureField
    PureField
    Participant

    With the last test I just used the LCD to control the motors and move them directly. Used the indicated distance and measured the actual one to get the difference.

    I noticed the cable ties flex from circle to tensioned and hence cause a 1/2mm flex between relaxed and tight. The metal does not flex and has helped the consistency. The settings are the ones from the setup page you posted and entered into Repetier Host.

    The page has a few downloads. I used the first one on the list.

    #19130
    Profile photo of vicious1
    vicious1
    Keymaster

    At this point in time the lcd controls are not always accurate for movement, they don’t always obey firmware speed and accel limits, you should always test with gcode.

    Again, use very conservative speeds for travel, testing at max speed isn’t going to give you valuable information.

    You have an extremely long z axis, most of this could be coming from that.

    I feel at this point you are changing to many things at once and are going to frustrate yourself. Only 1 variable should be changed at a time.

    What speeds were you using for the initial test that were pretty close? What software did you use to generate your g code?

    #19131
    Profile photo of vicious1
    vicious1
    Keymaster

    It still looks to me like your first one was fine, that either your pen was loose, you say it wasn’t. So the next logical step was you where moving so fast it was swinging that long z axis, and so fast you skipped a step and caused the writing to be a few steps off.

    I suggest testing a pen at no more than 30mm/s most of us mill at 10-30mm/s, so testing at 15-20 is your best option.

    #19132
    Profile photo of PureField
    PureField
    Participant

    I am using Inkscape to create the pattern with the GCODE extension to export path to gcode. I then import the gcode into Repetier Host and convert it using the Save for SD Print option to put it on an SD-card to read on the LCD. How can I tell the speed?

    %
    (Header)
    (Generated by gcodetools from Inkscape.)
    (Using default header. To add your own header create file “header” in the output dir.)
    M3
    (Header end.)
    G21 (All units in mm)

    (Start cutting path id: path4404)
    (Change tool to Default tool)

    G00 Z5.000000
    G00 X190.616031 Y77.612219

    G01 Z-0.125000 F100.0(Penetrate)
    G03 X191.322088 Y78.182530 Z-0.125000 I-5.398729 J7.405952 F400.000000
    G03 X191.711573 Y78.590627 Z-0.125000 I-2.358425 J2.640772
    G03 X192.219280 Y79.246967 Z-0.125000 I-9.640186 J7.981640
    G03 X192.462603 Y79.631045 Z-0.125000 I-3.046680 J2.199259
    G02 X192.591853 Y79.776179 Z-0.125000 I0.472870 J-0.290998
    G02 X192.689981 Y79.810191 Z-0.125000 I0.098128 J-0.124547
    G02 X192.807944 Y79.762793 Z-0.125000 I0.000000 J-0.170491
    G02 X192.855343 Y79.651718 Z-0.125000 I-0.106447 J-0.111075
    G02 X192.819688 Y79.475603 Z-0.125000 I-0.452772 J0.000000

    #19133
    Profile photo of vicious1
    vicious1
    Keymaster

    F is speed

    You are using arcs, you should turn them off (I & J commands), and it is going to too many decimal places Don’t quote me but I think marlin only handles 2 decimal places maybe 4 I don’t remember its late here. That has your z axis spiraling down .125 each path, using a pen you should not be increasing depth. actually now that I look at it that gcode isn’t complete.

    Try estlcam, and the walk through. inkscape is vector based no need to convert anything just save it as a dxf or svg and estlcam will read it natively.

    #19135
    Profile photo of Jasoroony
    Jasoroony
    Participant

    100mm/s (F100.0) is being used for Z, that’ll be overridden to what’s max in the firmware.
    400mm/s (F400.000..) is being used for X/Y movement – also will be overridden to 195.

    For me, Marlin with the Graphic LCD has serious issues with the G02/G03 codes at anything above 30mm/s. I’ve got a non graphic 20×4 LCD that I tested also, and at full speed like you are running it, it had similar issues (the entire machine coming to a stop when the LCD draws) but I’ve not tested at which speed it starts to work properly. They have mentioned that with 1/16 stepping (instead of our 1/32 stepping) it will give back half the CPU used which will probably make it work much better. Bit of a nasty trade off though, and they are talking about implementing a built in slowdown thing now to stop the machine going bonkers.

    Change all F values you see to 30 (F30) and try it again.

    #19138
    Profile photo of vicious1
    vicious1
    Keymaster

    For some reason the firmware max is not always obeyed. sometimes it is obeyed by just skipping that movement, not just slowing it down. It is best to set actual values.

    #19139
    Profile photo of PureField
    PureField
    Participant

    Have not yet changed the speeds as I just saw your post. Swapped the Arduino back to the new one with the current firmware.
    Now with a full power cycle on the electronics and restart it prints what appears to be perfect.
    The hardware incl. timing belt is exact to 1/4mm (hardly measurable with a digital caliper).

    Testing more now. The machine does not vibrate hardly at all and runs very smooth. First test 3d print is nesting washers where the outer diameter of the inner one is the inner diameter of the next one. 4 of them stacked perfectly inside each other.

    We are making progress.

    For the record – all parts, service and support provided by Vicious1 have been nothing but great. Thank you.

    #19141
    Profile photo of Jasoroony
    Jasoroony
    Participant

    I mostly use GRBL for stuff like this now, so my Marlin experience isn’t really growing anymore, but if you can find some gcode where commands are being ignored due to the feed rate I’d be interested in seeing it so I can avoid doing it in the future. Everything else I’ve tried on both firmwares generated from InkScape seems to perform the same way once the speed is fixed to counteract the LCD CPU drain. I.E. A simple find/replace-all on the two F values.

    #19168
    Profile photo of PureField
    PureField
    Participant

    Thanks for your help so far. I am going to start using the MPCNC now and let you know if anything new comes up over time.

Viewing 25 posts - 1 through 25 (of 25 total)

You must be logged in to reply to this topic.