What to research while building a BOOKSCANNER?

A project log for automatic BOOKSCANNER for Ethiopia

Design of a 500 Euro automatic BOOKSCANNER which can be build without expensive or complicated equipment or skills.

Karsten FuhstKarsten Fuhst 06/17/2016 at 21:270 Comments

While designing and optimizing a machine like a bookscanner we tried to identify possible spots where problems can occur.

Two major operations are:

The friction between the paper and the acrylic sheet results in required torque/ power of the motor which drives the book slide. We decided for acrylic sheet because of the

The material is available in different

The parameter to influence the friction is the surface texture.

We got samples of different textures and made experiments with different kinds of paper from tissue to whatever BUTTERBROTPAPIER is in english :)

In this video you can see the experiments and the results:

The result of the experiments was that a pearl structured surface had a significant lower friction with all tested types of paper. This acrylic sheet is now build into the bookscanner and the results from the experiments could be replicated in the real application as well.

For the page turning we made experiments with different shapes of so called coanda effectors. This unit blows a stream of air tangential onto a tubular. The resulting low pressure zone causes the lowest page to separate from the others. Next to the overall shape of the coanda effector the positioning and the speed of the air are important variables - Some of the work can be seen here:

The process of just separating a single page is pretty reliable. Just in case if less or more than one pages is separated this requires a reaction of the bookscanner. It should be tried to repeat that page separation till exact one page is flipped. To detect the number of pages we found out that an infrared LED and detector gives very good results. One needs to teach zero, one and two pages into the machine and possible failures during the book scanning process will be detected.

The following video shows how the page detection works: