It was the second time in a row that I was present at this conference, and not different from last year’s it did not disappoint. Unfortunately, the ARM’s father, Prof. Steve Furber, was not there but the exhibition itself was great. All major companies in the embedded industry were: ARM, Intel, Reneseas, Fujitsu, ST, Green Hills, Infineon, ST and many others.

I did not have enough time to walk around asking questions about all the cool stuff being exhibited there, so I just used I random and easy criteria – if it shines, moves or even flies maybe I should get more information. And it worked pretty well.

The first cool thing was a proof of concept of the company Vector.

PWM over Flexray

It works more or less like this: When the light sensor is blocked, the analog signal is converted into PWM and send over a Flexray bus to a host PC, in which it is processed and reproduced on the screen. The goal of this is to proof that Flexray is deterministic, whereas other bus systems (e.g. CAN) are not. In other words, the when data packages are sent over the bus it is certain that they will reach their destination, making it the best choice for x-by-wire applications.

After that, I walked into Xilinx’s stand to check out this remote controlled car.

Xilinx's remote controlled car

This car has 4 fish-eye cameras: one in the front, one in the back and one on each side. Every camera captures its own distorted video and hands it over to a processing central powered by an Spartan 6 FPGA.  The videos are then processed in real time to fix the distortion caused by the fish-eye lens and combined to form the 2D representation of the scenario surrounding the car.

It is not required much imagination to come up with a few applications by extending it. I can think of a few: automatic parking, identification of holes or water on the highway, monitoring from inside of possible aggressors approaching the vehicle and so forth.

Processed video

If you pay attention to the picture you will notice that the area where the car is, which is a blind spot, is replaced by some other car’s image. Pretty cool, right?

Finally, I would like to finish this post by sharing with you a discussion in which I participated on whether or not Engineers are still important in the marked nowadays. It started as a simple talk on the increasing demand for engineers in the embedded industry, but throughout the discussion, it turned into a career orientation for upcoming engineers in their quest for the first job. It was like a cook book written by the human resources of all companies. It was no surprise to hear that what the German companies expect from an engineer, it is not much different from the companies in my country: not only technical but soft skills, team player, self-motivated, etc. I guess it is a world trend for engineers.

That’s it. Until the next post…

About the project

The project has the intent to provide a viable SoC (Sytem on Chip) solution to be used for different application domains. If compared to other SoC Kits with Processors of 8Bit that reach their limits very fast or of 32Bit that require very “big” FPGAs, Spartan MC is a very cost-effective choice. The choice of the 18 Bit architecture was to explore the maximum capacity of the FPGA’s Blockrams and multiplexers. There are also a large number of standard and special peripherals that can easily included as needed.

Features

• 18 Bit RISC Architecture.
• 3-Level Pipeline.
• No Cache required – since the fast BlockRAMs of the FPGA are used for the registers and main memory there is no need for caching
• Standard and Special Peripherals as (re)configurable modules available: UART, SPI, I2C, JTAG, USB, etc.
• Multiple instances of  peripherals is also possible.
• It can be applied to different FPGAs of Xilinx’s Spartan family (Spartan 3E FPGAs start at 2$)  – depending on the application requirements a chip with a higher or lower number of gates can be chosen.
• Easy to configure – a Java programme is provided to assist in the configuration so that it is not required for the programmer to now details about the hardware implementation.
• Simulator, also written in Java, allows the programmer to test and debug most functionalities on the host PC.
• Multicore systems are possible

Spartan MC Core and its peripherals

Soc Solutions vs. Traditional Micro-controllers

Disadvantages

• Small micro-controllers are still cheaper than $2
• Up to this moment, the power consumption of uCs are still lower than in FPGAs
• AD/DC-Converters in uCs are integrated on chip (although some new FPGAs provide these on chip as well)

Advantages

• Configurability – the configuration can be adapted according to application (or changed later if necessary)
• Different signal levels on the Interfaces are also configurable
• Scalability of Performance – if more processing capacity is required a more powerful FPGA can be chosen (of course there is a trade-off regarding the price)
• No need to integrate several uCs into your project, just include the  modules you need via configuration tool and you are good to go

This was just a brief overview on the Spartan MC. For more information click here to visit the official project’s web site. However, the english version is unfortunately not as up-to-date as the german one.

This is my first official post. I hope to be writing more stuff very soon.

Fabio Mayoral