Test tubes won't stand up on their own. A tool to hold them is needed. It can be easily laser-cut from a sheet of acrylic.
It was decided to use a sheet of 1.5mm acrylic as a source material. The thickness was chosen based on availability and cost.
OpenSCAD was chosen as the design tool. The parts then can be programmed, assembled from individual cube/sphere/cylinder elements. The scripting approach allows parametric generation, specifying dimensions and element counts as variables. The program can output DXF files, for a laser cutter.
Multiple parts can be programmed in a single draft, then shown either individually or in an assembly.
OpenSCAD can be also executed from a commandline. This allows script-based generation of multiple DXF files from a single drawing.
A suite of DXF utilities was written for manipulation of the DXF files. The utilities can convert consequent individual lines to polylines (the cutting software tends to cut them out of order instead, taking time). Conversion to bitmaps for review and publishing, and assembling of multiple source files into a larger one for single-operation cutting, is also possible.
The software that controls the laser cutter. It is fairly crappy but does the job if it is not asked to do too much.
The parts are drawn in OpenSCAD. Full assembly is rendered as a model, to check if everything fits together, if the holes are matching.
A shell script was used to command OpenSCAD to produce DXF files for the individual parts; dxfline2poly was used to convert the lines-only file to polylines.
Another shell script was used to assemble the standalone DXF files to a large single-panel one.
(The single-panel cut was later not used; parts were instead cut one by one, with manual positioning of the acrylic sheet. The reason was the availability of an oddly shaped off-cut from another project.)
The linked images are at 300 DPI, printing-ready quality; print from something that preserves the actual dimensions, e.g. Irfan View.
Tube plate | Base plate | Back plate | Side leg |
Side plate | All merged in one panel |
testtuberack1.scad, download here
The design utilizes snap locks for assembly. Pairs of shaped pads are used, with cutouts along their sides to make them long enough to allow flexing without breaking. Round edges have to be used for strain relief, to avoid breaking off of the pads.
The concrete dimensions for 1.5mm acrylic were found by trial and error.
At single-pass cutting (8 mm/sec, 11 mA tube current) the thin acrylic tends to melt on the edges. The edges are overheated by the laser beam. The accumulated heat then spreads to the surrounding areas, melting them for a while.
The little snap-lock pads tend to deform; the single-pass cut heats their both sides too soon after each other, and the material softens through the thickness. The pad then can bend or twist a little.
The molten edges protrude a little out of the plane, making it slightly thicker. The matching holes are fairly accurately sized, so the pads then don't fit. Few strokes of a small file remove the protrusions easily.
Leftover acrylic, detail of deformed material | Leftover acrylic, detail of deformed material | Cut parts, detail of deformed material | Cut parts, detail of deformed material |
The design can be easily changed to other kind of locks, including mere protrusions into matching holes, when the assembly is to be glued instead of snap-locked. The lock shapes are defined in three separate functions; one for the protrusions that do the snap locks, one for the cutouts for allowing bending of the pads, and one for the cutouts in the matching parts.
The design holds well. The side legs aren't even strictly necessary; the assembly was fairly stable even without them.
The snap locks perform surprisingly well. One leg of one lock broke off; it was deformed thermally during cutting, and insufficiently filed to flatness. Excessive force during the assembly then caused the damage.
The holes for the large test tubes were unnecessarily large. Two millimeters less would be better. The SCAD file for download was amended to reflect it and the DXF and PNG files were regenerated.