The side panels are made from 18mm plywood (multiplex), with some special angles. The top slopes, on which the plexiglas platen rest have an internal angle of 110 degrees. The lower slopes have an angle of 35 degrees to vertical, making them perpendicular to the opposite top slope. This simplifies the alignment of the cameras, independent of camera-platen distance. The height of the side panels is approximately 226mm.
The two side beams are separated by rectangular 18mm multiplex plates. 100mm high and 335mm long and attached at the level of the lower slopes with sunken pozidrive screws.
The camera carrying beams are constructed from two 20x20 T-slot beams per side, 405 mm long and attached with sunken flat-headed M4 screws and slotted nuts to the rectangular plates. The lower ends are bridged with internal corners, more slotted nuts and a short 20x20 T-slot bridge, 123 mm long.
The camera platforms are raised with three M4 threads so the camera lens ends up in-line with the center of acrylic platen it faces. The threads are 145mm long. Because the beams are perpendicular to the opposing plate, no extra correction for the platen-camera distance is needed. The threads attach to the camera platform. Having three adjustment points makes alignment simple. To make adjustment even simpler, the camera platform nuts are spring-loaded.
The image quality is improved by reducing internal reflections, mainly by making most of the inside matte black. I use black paper, but there is special paint for this. But what certainly also helps is a light source illuminating only the platen and not much else.
The acrylic platen are supported and attached to the frame with 3D-printed rulers. Apart from the 55 degree angle where the two platen meet, they need no special treatment (a small 3D-printed tool is included in the zip to help getting the proper camfer). The platen dimensions are (335 + 2 * 18 = ) 371 x 254mm, and 4 mm thick. The 254mm dimension leaves some margin for the 35 degrees camfer.
The 3D-printed rulers come in three types:
on the inside, on top of the rectangular plates supporting the lower horizontal side of the platen,
on the outside, clamping the platen to the width of the frame (horizontal side),
on the outside, clamping the platen to the length of the frame (diagonal side).
The parts are designed in OpenSCAD. The design files are included in the zip. The parts are printed using PETG, as the standard PLA tends to deform when under pressure. The horizontal parts are split in two, as the whole part didn't fit on my 3D-printer.
All rulers are screwed to the multiplex plates with small pozidrive screws (less than 20mm).
The control computer is the standard R.Pi 3 with Pi-Scan, touchscreen version. The back cover is for the Pi 4, so extra holes had to be made. There is a provision for a foot switch, but currently I don’t use it. A 3D-printed tray for the Pi is attached to the carrying frame. At sometime I made an Arduino based light switch and capture delay, but currently don't use it.
The camera's are Canon A2500 with CHDK. They usually work Ok, but occasionally crash, requiring a power cycle for that camera. With two of these issues I could scan a 300+ page book in half an hour.
The frame is improvised from a small wooden table placed upside down, with plywood plates screwed to the legs to carry the bookscanner. The inside of the table ‘top’ surface houses the light box, camera mains adapters and an extension chord.
The light box contains a light source (mains powered LED floodlight, 10W, 700lm) in a box, with the top side carefully trimmed to illuminate only the platen, not the other parts of the construction.
The quality of the lighting is also improved witth more distance between light source and illuminated platen. But with the inverted setup, this is usually a compromise.
The light box is still under development and will be documented later (maybe).
fjkraan@electrickery.nl, 2024-03-26