a long week end

A few days have passed since my last post so let’s check what I have done the last couple of days.

We have huge reunions with the HARP group had here are our conclusions:

we have defined our PSSC
Can I transfer data from one computer to plate via Wifi?
Can I turn an LED on the stationary plate with the desired color?
Can I to know the position of my plate in real time?
Can I display a still image on the plate?
Can I throw a video on the stationary plate?
Can I display an image in a direction of the plate movement?
Will I get to see a video in a direction of the plate movement?
Can I  display a hologram?
Can I display a hologram animated with retinal persistence ?

With the bonuses :

Can I get to interact with the hologram from the PC?
Can I can use data from a 3D movie and display them?

Our structure would be
Number of LEDs: 16 * 16 * 2 = 512 leds
Number of frames per second: 16
Number of positions in our area: 512
Number of bits of color coding: 4

Flow Rate: number of LED * number of frames per second * number of positions * number of bits of colour coding
Temporary speed: 4MB / s

Switching speed: number of frames per second * number of positions per second = 8192 Hz = 12 micro seconds

LED switching speed of 0.12 ms = 7.5 micro-seconds between position effect we will successively display the 16 line of our matrix, we will return later.

“Screen” will have a resolution of 32 * 32, using the same trick that was used on roseace shifting our two matrices led led half a row-level and column-level in order to mix the two matrices .

Architecture 1.0:

the card receive the data  via WiFi.
The card decodes them through a FPGA, which will store them in RAM.
The same FPGA makes continuous readings of RAM reading successive addresses and will communicate the data to the drivers. To save space we will use a method inspired group of LED Cube: 16 lines 16 columns and with an enable  on the line.
therefore  the FPGA will monitor an enable on the lines through a multiplexer

The driver who initially controlled the matrix by column will now control it led by led. All calculations of the value of the LED has been done before the transfer to the card and storing them in the ram directly in the right order for the reading.

Advantages: no CPU,  only hardware=> very fast.
Disadvantages:  Very low level of code, we can not process the received data with the FPGA.


We have two drivers per side so each will control 8 colones of 16 leds.
The enable will allow us to pilot only one line at a time and playing on the persistence of vision we will  see all the pairs columns of our image and 512 positions later we will display the impairs columns with  our  second matrix LED, which is on the other side of the plate and is offset with the first of a half-led to the rows and columns. (As for the two RosesAce row LED was shifted)

his Datasheet:
Its speed: 1 Mhz so a priori it sticks for controlling the leds (

The Ram

Have two RAM read access and  one write access. because it requires that we can control the two side our surface and at the same time that we can update our train for the next form

Writing: what you get and is decoded by Wifi
Play: one access by face.
We still think about how we will store our data in order to optimize the playback speed.

Wifi protocol:

Send on each packet an image for a given position of the plate. which will be really light and fast

We have still lots of questions about the HARP project and try to find answer.
With the presentation of tomorrow we will have some new ideas to present so stay tune.

Pierre Paulin and Alex for the HARP project.

PS: For our project with xbee i have almost succeed to make my led twinkle green. but know it’s time to go back to the presentation of tomorrow.

2 comments to a long week end