Mendel Replicator - one step closer

In the time just after christmas, we had some free time to improve our 3D scanner system. We added some gears to get a smoother and slower scan with the linelaser. Also we made a small turntable to let us turn the object between the scans so we would be able to make a fully automated scan of the whole object.

We only had one Motor available, so we used the turntable like a spool, and connected the thread on this spool to the gears of the linelaser. So when we now let the linelaser turn more than normally needed for scanning, it pulls on this thread and turns the turntable.

Then we created a small automation script to control the motors and click on the right buttons in the DAVID-Laserscanner software so it makes scans from all sides of the object, which can be easily fused into one 3D model.

When this was ready, we started a test to see how much work and how fast it would be to get from one real object to a printed copy.

We started scanning a little dragon model (~6 x 4 x 4 cm³) and after around 2 hours of fully automated scanning, we had 8 scans from all sides on our computer. The scans where so good, no cleaning was needed. We then spent around 1hour combining those scans and preparing the resulting model for printing. Although on a faster computer this would probably go much faster. You can see the result here:

When the model was ready, we sent it through skeinforge and started to print it on our Mendel. Skeinforge needed 15mins and the print time was also around 2 hours. You can see a few pictures of that here:

The resulting object looked really good, and pretty similar to the original object, as you can see here:

So all in all we needed 2 hours for automated scanning, 1 hour manual work for cleaning and preparing the print, and 2 hours printing time. This means we have a time from a real object to printed copy of around 5 hours, depending on the size of the object. Sounds great :-)

You can find the scanned object at thinigverse thing: 5374


laserpointer carriage for steppermotor

Last friday a friend of mine brought me some nice electronics.

It is the first version of a steppermotor controller with one control unit.
With one steppermotor controller it is possible to ease the 3D scanning with a laserpointer.
So I quickly designed and printed a carriage for the steppermotor and the laserpointer and mounted it to a m8 bar.

The bar was taped to my desk right beside the camera and the background of my 3D laserscanner.

The motor has a step angle of 0.9 °. The controller has a substepping of 16, so one complete rotation is divided into 6400 steps.
This is just enough to move the laser in a distance of about 40 cm with ~0.5 mm steps.

The firmware was ready to compute easy commands. After testing the angle and the speed needed for scanning,
we spend some time to modify the firmware to compute a chain of command.

Now it works just fine. Press the "on" button and the scanning sequence starts. After finishing the sequence the electronics shuts down automatically.
The scanning sequence can be easy modified in the firmware.

The scanning software we use is the original DAVID Laserscanner software which was included in the complete laserscanner kit I bought at their homepage. The laserpointer is a red linepointer with an integrated focuswheel. It was also included.

We tested the accuracy of the steppermotor with a scanning resolution of 1280x1024 on an object with a size of about 11 cm.
After 2 times up and down (4 scans in about 5 minutes) it looks like this (interference was removed):

Last time I scanned with this resolution one scan was about 45 h and I moved the laser as slow and accurate as possible - manually of course. The result was horrible and far away from this scan!

The stl files for the laserpointer/steppermotor-carriage can be downloaded at thingiverse here.

Next visit we want to test a complete version with 6 control units - but not before christmas. I hope to find an easy way to scan with 3 motors: The first one for the movement of the laserpointer as now tested and proofed, the second for the rotation of the object and the third one for z-movement. The rotation of the object should be possible with a simple plate. There is no need for high accuracy of the z-axis, so the construction shouldn't be very complicated.

greetings from the lake of constance!




More Articles...

  1. almost a mendel replicator