First, I finished the PCB for the STM32 board.
Then, for our project, we have chosen the general architecture: the board embedded on the propeller will contains a cortex A8 and a FPGA. Indeed, benchmarks I have found about ffmpeg, the library we will use to convert MPGE2 (or MPEG 4) received thanks to the wifi connection use about 80 % of a A8 cortex. The wifi will use about 5%. So, we can not use the A8 cortex to command the LED drivers.
The question is the following: we can chose working with a armadeus (http://www.armadeus.com/english/products-processor_boards-apf51.html) but we have to design a main board with wifi and, we will only begin to work when the main board will be ready or, we can use a gumstix (the same that we used for previous test) and we have to chose the FPGA and to design the main board with the FPGA but, we can start working on the gumstix now.
Finally, I think about the LED driver we have to use and, the LED board architecture. We decided to use 8 different bus to command the LED driver so, LED will be divided in 3 groups on each propeller. The frequency needed is about 10 MHz for each LED driver. The LED will be command with a PWM signal created by the LED driver.. A LED will have a frequency of about 15 kHz so a PWM frequence of 33 MHz will allow us to chose correctly the LED intensity.
Today we make an experiment with the Gumstix. Our problem was to know if in turning at 1000rpm in a ventilator the WiFi will work correctly.
The first was to get a wireless power source to the Gumstix.
We got a lot of problems with that. First of all, the power connector of our Tobi expansion board is damaged (thanks to Copterix ) which cause a lot of reboot at any movement… We solved this with a lot of gaffer.
Secondly our hacked connector (with sticky tape) for the battery was not very well either and caused some trouble. Thanks to the PACT project which order a lot more than necessary 9V battery connector we improved our power line.
And last problem for the power, our board was rebooting in loop before we had a chance to login in… Simple problem but hard to guess, our battery didn’t deliver enough current. It was easily solved with a second battery in parallel.
Now, add a lot of gaffer to secure into place the batteries, the Gumstix and the antenna and we’re ready.
Of course, Murphy’s law requires, at this exact time the Gumstix decide to no longer properly boot and the WiFi to stop working. But with a little of black magic that finally worked and we have done our tests.
Our ventilator was missing of power and can’t get to its maximum speed (~1100rpm) with our board mounted on it (we could do better with a better balancing of the weight). But at the speed that we reach we succeed to had no loss of bandwidth (to confirm with a faster ventilator) which is a good news.
The bad news is that even placed on a table at 20cm from the computer we can’t do better than 1.5MB/s of transfer rate in WiFi (and 3MB/s in ethernet…) which is some kind of disappointing.
Here are the slides for the initial projects presentation that took place on February 15:
Today we define our architecture. The main board with the micro-controller, we will use an embedded Linux to add a lot of possibilities (plug-in, video conversion, GUI, etc.) and will communicate with the FPGA embedded in the propeller. In the main board there will be only USB ports to communicate but we will try to design a VGA2USB plug-in (and maybe DVI or other if we have time). The FPGA transforms a raw image (VGA style) in a list of beams (of colors for the leds) and commands the leds.
Our propeller will be a 40cm diameter propeller (to define : how many led can we put on).
For the moment,
Sylvain will work on the “no wire” transmission of power and information from the main board to the propeller.
Jérémy will work on the main board.
And I will work on the FPGA/propeller board.
After our first presentation and our first exchange with our professors, we have new ideas for the general architecture of the ROSE-ACE.
We study a similar project : http://www.das-labor.org/wiki/Borg_Ventilator to have some ideas to solve some problems.
We have to choose an general architecture for next week.
FLX, Sylvain, Jeremy Sauget
That’s not very pleasant to discover that what we expected to do – a POV display that can be plugged on VGA just like a screen – has already been done.
Moreover in a bright way by this team.
Actually I right now feel like that.
Well these German people did pretty good job, let’s do better!
As a reminder, here are the project attributions.
Educative robot with plugins (Tutobot)
Model railroad (SaMoRa)
- Guillaume B.
Persistence of vision (Rose Ace)
Ball robot (Ball-E)
Wall-climbing hexapod robot (DHEXTROSE)
- Guillaume C.