This is basically a faster / more sensible version of the driver code for the 3D RGB Xmas Tree from The Pi Hut [1]. The Pi Hut provided their own on Github [2], but it is super slow - it only allows you to change one LED at a time. In The Pi Hut's code, changing a single LED requires the whole tree to be refreshed, which requires the generation of SPI command byte string. Refreshing the whole Christmas tree seem to be quite slow. Furthermore, every time the SPI command byte string gets generated, 5 list comprehensions are performed*. This could be slowing the program down.
* I could be wrong on this one - I am not very good with Python. This is based on my understand of line 88-90 of tree.py
Put tree.py in the same folder as your code. To instantiate the driver, use the following:
from tree import FastRGBChristmasTree
tree = FastRGBChristmasTree()To set a LED, you can do the following:
tree[22] = [255, 0, 0]Of course, the above is the equivalent as
tree[2,0] = [255, 0, 0]You can also set individual LED's brightness (which you cannot do using The Pi Hut's code).
tree[22] = [10, 255, 0, 0]You can also set multiple LEDs at once, e.g.
tree[:,0] = [[255, 0, 0], [255, 255, 0], [0, 255, 0]]The LEDs are set using a list in the format of [R, G, B] or [Brightness, R, G, B], where Brightness is an integer between 0-30 inclusive, and R, G, B is an integer between 0-255 inclusive.
The statements above only configures the command buffer, which still needs to be sent down to the SPI bus. To commit your configuration, do:
tree.commit()There are also a lot source code examples in the repository that you can follow.
The indexing schemed is explained in the section below.
I decided to include a numpy array (__led_config) to help with indexing the LEDs [3]. Vertically, the LEDs are separated into layers, based on the height from the base of the tree. The bottom layer is layer 0. The layer below the star is layer 2. If you look at the Christmas tree from top down, you can see that the tree has 8 "vanes". The numpy array stores the index for each LED in the format of [layer][vane]. If you orient the tree in such a way that the Raspberry Pi is towards you, you can index the LEDs using the table below:
| Layer | 0 | 1 | 2 | 3 | 3 | 2 | 1 | 0 | ||
| Vane | 0 | 1 | ||||||||
| 0 | 24 | 19 | ||||||||
| 1 | 23 | 20 | ||||||||
| 2 | 22 | 21 | ||||||||
| 3 | 7 | 12 | 11 | 10 | 3 | 3 | 9 | 8 | 7 | 2 |
| 3 | 6 | 6 | 5 | 4 | 3 | 3 | 2 | 1 | 0 | 3 |
| 2 | 13 | 18 | ||||||||
| 1 | 14 | 17 | ||||||||
| 0 | 15 | 16 | ||||||||
| 5 | 4 | |||||||||
| R | A | S | P | B | E | R | R | Y | ||
| P | I |
The existing example code from The Pi Hut is awful. There is no datasheet. I would have expected better documentation and drivers for the £18 I paid.
If you want to go even faster, you might want to mod your Christmas tree. Please have a look at issue #2.