From Lancaster And Morecambe Makers
Jump to: navigation, search

Mendel90 build page

This page is for anything related to the project to build a Mendel90 3D printer for the space. Gus and Andy are working on this at the moment - if anyone else wants to get involved, you're welcome. We are planning to build the dibond_E3D version of the printer - i.e. the frame is made of 3mm dibond sheets, which we have, and we will fit an E3D hotend.

Notes on the build manual

These are some relevant points that seemed worth remembering from the build manual.

  • The hardest part looks like cutting the dibond sheets. We are hoping to use the CNC machine when it's ready, but there are some things to do before it can be used.
  • When we do the build, check that we are both working from the same commit of the github repository.
  • The hex pillars that connect the heated bed to the Y carriage need to go into tapped holes, so we'll have to find a way to tap holes in dibond.
  • The extruder needs a 'hobbed' bolt, where a small portion of the bolt has been roughened so it can drive the filament. Luckily there is an identical one in the parts we've salvaged from the donated Mendel printer.
  • When we assemble the extruder, we may need to refer to the openscad scripts for the dibond_E3D machine, as well as the pdf build manual, which is intended for the plain dibond version.
  • We need to decide what electronics to use. The manual recommends Melzi electronics, but there are others.
  • We have some RepRap Gen7 electronics from the old mendel (white board). These should work but are quite out of date (latest revision of the page is 2013). . We may need to upgrade from v1.3 to v1.3.1, which just involves getting a new chip and burning some fuses differently.
  • We also have Sanguinololu electronics, version 1.3a (red board). These have blown capacitors but shouldn't be hard to fix. 1.3a is the latest version.
  • We want the 12V version of the E3D hotend.
  • Andy may have some suitable dummy load resistors for the power supply - will check.
  • The power supply we have is 400W, which is not as much as the 500W supply suggested in the build manual. We can always try the one we have though and see how it does.
  • Do we have a dial gauge for levelling the bed?
  • Also a large set square would be useful to make sure the axes are square.

Salvaged parts from the old mendel printer

Gus and Andy have been through all the metal parts from the donated Mendel printer, separated them from the plastic, and counted up how much we have of each part. This is a list of what we have. It looks like we will need to buy most of the parts for the M90, but I've put notes to say where there are things we can salvage.

Number Part Notes
20 10mm Hex M3 bolts Need 4 for M90
6 15mm Hex M3 bolts
3 30mm Hex half-threaded M3 bolts Need 5 for M90
10 25mm Hex half-threaded M3 bolts Need 2 for M90
3 M3 nub (grub) screws Need 6 for M90
1 60mm hobbed half-threaded M8 bolt Suitable for extruder on M90
79 15x10x2mm washers Dimensions are outer x inner x thickness
9 15x10x1mm washers ditto
1 19x8x2mm washer
22 7x2.5x0.5mm washers
6 30x8x2mm washers
87 M8 nuts Need 2 for M90
1 M8 nyloc nut Need 4 for M90
22 M3 nuts Need 7 for M90
3 Large diameter M3 nuts
3 M4 nuts
4 22x8x7mm round bearings (metal caps) 7 22x8x7mm bearings needed for M90. Caps don't matter
5 22x8x7mm round bearings (plastic caps)
2 13x4x5mm round bearings (plastic caps) 4 needed for M90
10 LM8UU Linear bearings 10 needed for M90
1 8x370mm smooth rod
1 8x380mm smooth rod
2 8x470mm smooth rod
2 8x442mm smooth rod
The smooth rods will do most but not all of what we need for the M90
2 8x500mm threaded rod
1 8x465mm threaded rod
6 8x390mmm threaded rod
4 8x350mm threaded rod
2 8x220mm threaded rod
1 2.8x303mm threaded rod
2 2.8x194mm threaded rod
1 2.8x185mm threaded rod
2m 1mmx5mm drive belt, 0.2mm tooth pitch
96cm 2mmx5mm drive belt, 0.5mm tooth pitch
90cm 2mmx5mm drive belt, 0.5mm tooth pitch

Workflow to create gcode for the CNC machine

I (Andy) have been playing with openSCAD and FreeCAD, looking for a way to create GCode for the CNC machine that will cut and drill the dibond sheets. So far, it's looking like the cutting will be possible, but I'm not sure yet about the drilling. I'm using FreeCAD 0.16, compiled against liboce-0.16

The workflow to get the sheet model into FreeCAD from openSCAD is as follows:

  1. Clone the git repository and generate the scripts for the dibond_E3D machine.
  2. Look in the dibond_E3D/sheets directory for the dxf file you want to copy. E.g. frame_gantry.dxf
  3. Make a file exporter.scad in the sheets directory, with these contents:
use <../../scad/main.scad>
linear_extrude(height=3) frame_gantry_dxf();

(substitute 'frame_gantry' for any other filename of a dxf file)

  1. render this and export to stl.
  2. import the stl into FreeCAD.
  3. Do Part -> Create shape from mesh
  4. Do Part -> Convert to solid
  5. Do Part -> Refine shape.

You should now have a native FreeCAD object which you can use as any other.

Then the next thing is to create a tool path inside FreeCAD:

  1. Select the Path workbench.
  2. Start a new Path project.
  3. Give the project a tools table. I made a table with a 2mm and a 5mm end mill cutter.
  4. Load the 5mm tool using Path -> Partial Commands -> Tool number to load.
  5. Select the top face of the sheet object, and create a Profile using the 5mm cutter. Path -> New Operations -> Profile.
  6. Edit the parameters of the profile. I used the defaults except for adding an extra offset for this (the roughing cut). ('Offset Extra').
  7. You should now have a tool path showing (thin green line around the sheet).
  8. Repeat with the 2mm tool and no extra offset.

If you rotate and zoom the model, you should now be able to see two thin green lines spiralling around the sheet. This is the path the CNC tool will follow. It should be possible to generate GCode for the router from this.

This is an image showing what it should look like.

I haven't yet managed to make a path for cutting the holes. The Pocket tool looks like it should do this, but it took too long to render on my (fairly high spec) machine. If it comes to it, we may be able to work around this by breaking the model into several segments and cutting the holes in each separately. Or there may be some other way.