Zusammenfassung
There is a lot of activity going on concerning the use of Linux in consumer audio products. In some cases Linux already entered the living room through a SetTopBox. However, the vast majority of current consumer audio equipment is still based around microcontroller based MMU-less solutions where programming is done without an Operating System. Also one often still uses assembly code and proprietary tools of the microcontroller suppliers. The big disadvantage of this approach is that it takes many man-hours to get a product finished, to solve bugs and let alone to improve the functionality during the test period.
This talk describes in detail the development process of the embedded application software for a consumer product called a surround sound decoder. Incoming audio and video signals are being routed and processed such that a video signal from a DVD player is displayed on the proper TV or monitor and the sound can be auditioned using stereo, Dolby Digital EX, DTS, THX, Dolby Prologic 2, Neo:6 and other audio signal processing algorithms. The front of the product has a vacuum fluorescent display (VFD) and buttons. Information can also be displayed using the TV or monitor connected to its video output, the TV/Monitor then is used as on screen display (OSD). The product can be operated with the buttons on the front or using a standard IR HiFi handset.
In this particular case ArdisTech used an MMU based Power PC processor, the Linux kernel which was stripped to fit together with embedded/Qt and the application software in 16 Mbyte flash eprom and to run in 64 Mbyte DRam. For both the VFD and the monitor/TV information embedded/Qt was used. We managed to bring down development time from 2 years to less then 6 months. Because of the use of Linux/eQt in the earliest possible stage of the product development we could offer the client an emulation environment on a standard PC computer which was: what you see is what you get. The big advantage was to have a first final version of the software available while the hardware was still in the designing stage. The second noticable advantage was the reliability of the software versions despite the many changes which during that times occurred both in the hardware and software. The third advantage was that functionality changes could quickly be build into the software during functional test time. The turn around time was often minutes/hours instead of days/weeks. Since the Power PC processor chosen supported 10 Mb ethernet all prototype machines could run as clients of an NFS server which allowed simultaneous updates of the software on all machines anytime in a very easy way without having to go through the elaborate process of flashing eproms per machine.
Über den Autor
Jan DeWit is a biophysist with a Ph. D in Nuclear Magnetic Resonance of Tobacco Mosaic Virus. He has been involved in R&D of Multimedia type products ever since. Together with his collegue Bart Thissen he owns ArdisTech. ArdisTech consists of software and electronics engineers who specializes in doing technical-ICT projects for third parties either in an assignment or in cooperation mostly using Linux one way or another.
