Das krümelige Weblog - live aus der Parallelgesellschaft ..

Schade NVidia..

Wenn sich an der Haltung NVidias nichts ändert, wird die nächste Grafikkarte, die ich mir zulege eine ATI sein:

Während AMD mittlerweile die Spezifikationen für seine ATI-Grafikkarten veröffentlicht hat, bietet Nvidia weiter nur proprietäre Linux-Treiber an. Dabei wird es auch in Zukunft bleiben, heißt es in einer Stellungnahme, die gegenüber ZDNet abgegeben wurde. Der Treiber sei geschlossen, da nur so geistiges Eigentum Nvidias geschützt werden könne. Zudem erwarte man nicht, dass die Community Probleme mit dem Nvidia-Modul behebe, dafür habe man ein eigenes Kernel-Team.

Auf der anderen Seite hoffe ich auch einfach nur, dass ich mir demnächst endlich eine neue Grafikkarte zulegen kann..

Geometry acquisition using light

For the last month I’ve been working on my final year project (Diplomarbeit) which consists of a measurement integration for high precision forged crankshafts. The institute I’m working at has bought a shaft measurement system. The system from Hommelwerke (Jena) uses the principle of light projection and - using focused light-rays - can detect geometrical variations quite well. For this purpose, the part that’s being measured is shone above and it’s contour is detected by a camera at the other side.

The first image shows a test object. Looking at it using the silhouette method provided by Hommelwerke the system has the ability to see most geometry elements (second image). The object can be rotated around the z-axis (top-down) to provide more data for the geometry reconstruction.

The projection system has a flaw though which is, it can’t detect or measure concave elements. That’s why the existing system has to be extended by a geometry acquisition system using structured (mostly striped) light. Using different angles for light-projection and camera system, the relief of a given system can be detected and any variations in geometry can be measured.

Using the course of the projected stripes (which usually variate in thickness to avoid stripe-mixing) the system can calculate the resulting geometry. Again, to fully capture all geometry elements, the test object should be rotated around the z-axis. Additionally the projection system should be translated along the y-axis (perpendicular to the stripe direction) each time the test object is rotated.

For that purpose a four-way positioning system has to be constructed that will hold the additional sensor (light source and camera) and bring it up to the object (i.e. crank shaft).

..to be continued..

3d-printing and more open source hardware

This is something for the very near future. It’s also something quite interesting for everyone involved in modeling 3d-objects. Of course more than for theatrical designers 3d-printers are - used as prototyping machines - more useful for a better understanding of CAD-objects.

DailyTech reports of the company Desktop Factory which is planning to bring out a 3d printer to the market which will cost less than $ 5000.

The Desktop Factory 3D will build models layer by layer from bottom to top. The models are constructed using nylon which is mixed with aluminum and glass and then hardened by heat. The Desktop Factory 3D printer will measure 25″ x 20″ x 20″ and weighs less than 90 pounds. It can build 3D models up to 5″ x 5″ x 5″ constructed of layers 0.010″ thick.

The interesting part is that this is by far not the first such appliance. There are already open source projects which aim to bring prototype machines which cost much less AND are in their entirety open source and self-replicating. I know of two such projects: Fab@Home and RepRap.

RepRap will make plastic, ceramic, or metal parts, and is itself made from plastic parts, so it will be able to make copies of itself. It is a three-axis robot that moves several material extruders. These extruders produce fine filaments of their working material with a paste-like consistency.

RepRap, which is my favorite (mostly since it looks like something i could build) is planned to cost around $400 in material. Owners are encouraged to support new builders, which is actually a new dimension of the support-communities known in the software-scene..

Also, don’t forget to make two machines for friends or colleagues. It’s by no means required, but it is definitely in the spirit of the project. The project is free (as in speech), but unfortunately real stuff costs money, so the common etiquette is to pay for raw materials / shipping / construction time.

RepRap also has a bigger working area, which is around 300×300 mm² and is easily expandable.. Later on we’ll need a plugin for Blender to export objects created by Blender to the RepRap-Software.

Open Source Hardware - OScar

The idea of open source development is slowly getting hold of hardware development. There are still some hoops to take but in general the idea of opening the construction and electronic sources to the public and letting a wide and rather indefinite public decide which way they go has some charm.

Hoops

The biggest problem in an open source hardware project has to be the cost of ‘building’ (or copying) the product. While software-projects can basically be copied endlessly without further cost other than the initial time consumption for development, copying hardware implies buying the raw material and processing it with added time-investment for every new copy.

The manufacturing process from source to binary application revolves around some sort of ‘make’-batch plus some bug fixing depending on the original source quality and the platform for which the source is being built. Ideally, you won’t even need to build the project, since some other guy built the binaries for your platform and you can download them right away. In a hardware project you’ll need some expensive tools to manufacture the raw material into the finished product. The only cost-reduction in an open source hardware project would be the expertise and the development time.

Bringing 2.0 into the factory

Still, having a community think about needed features and how they can be achieved can be a upside to a product and give it a marketing advantage. On the whole opening up development cycles and sharing the knowledge of modern industries can bring that web2.0-effect to a business sector that has always been marked as a monopolistic market. What if a startup in the traditional industrial sector could achieve the same success story as did some startups in application and internet development sector?

Open Source Car

One such open source hardware project - and the most fabulous but yet rather immature one - is the Open Source Car.

The idea behind the OScar project is simple: A community of people plans and develops a new car in the web. The idea is about the goal to develop a simple and innovative car, but also about the way how this goal is achieved. We would like to convey the idea of Open Source to “hardware” and we want OScar to be the precursor for many different projects in this field. If you are interested in how the idea was formed, read the manifesto of 1999.

Unfortunately right now the OScar project is stuck in a position, where there is much discussion going on in the forums, but actual decisions don’t seem to be taken. For example, one major subject is the design of the car. While there are some suggestions mostly contributed in the form of 2d-sketches, nobody is taking up the task of actually modeling them into a 3d-design.

Open Source CAD/CAM

Another major problem is the lack of a good open source cad-software package. Since the project tries to use other open source projects as a foundation, so that everybody is legally able to contribute, they can’t really use PTC Pro/Engineer or Dassault CATIA. Surprisingly, there are some open source alternatives, though none of them is even half as sophisticated.

• BRL CAD - used by the US military.
• FreeCAD

I hope to get to know those applications in the near future and will hopefully continue the article about where open source hardware projects stand and what we can achieve with them. To be continued..