Code is (c) Henner Zeller h.zeller@acm.org, license: GNU General Public License, Version 3.0
For details of the RGB Matrix library used and how to connect, see the github over at Raspberry Pi RGB Matrix.
Clone with --recursive to get the RGB matrix submodule when cloning
this repository:
git clone --recursive https://github.com/hzeller/rpi-matrix-pixelpusher.git
(If you did a git pull to get the latest state, also make sure to get the latest matrix code:
type git submodule update).
For wiring, please have a look at the library documentation.
You can choose the wiring you are using with the option --led-gpio-mapping
on the command line. E.g. if you are connected to an Adafruit HAT, then it is
--led-gpio-mapping=adafruit-hat or --led-gpio-mapping=adafruit-hat-pwm
depending on if you did the PWM modification (which you absolutely should to
reduce flicker.
Simply run the program as root (which is needed to access the GPIO pins). It will drop its privileges once it has set up the hardware.
$ make
$ sudo ./pixel-push
These are the available options
usage: ./pixel-push <options>
Options:
-l : Switch on logarithmic response (default: off)
-i <iface> : network interface, such as eth0, wlan0. Default eth0
-G <group> : PixelPusher group (default: 0)
-C <controller> : PixelPusher controller (default: 0)
-a <artnet-universe,artnet-channel>: if used with artnet bridge. Default 0,0
-u <udp-size> : Max UDP data/packet (default 1460)
Best use the maximum that works with your network (up to 65507).
-d : Same as --led-daemon. Use this when starting in init scripts.
--led-gpio-mapping=<name> : Name of GPIO mapping used. Default "regular"
--led-rows=<rows> : Panel rows. Typically 8, 16, 32 or 64. (Default: 32).
--led-cols=<cols> : Panel columns. Typically 32 or 64. (Default: 32).
--led-chain=<chained> : Number of daisy-chained panels. (Default: 1).
--led-parallel=<parallel> : Parallel chains. range=1..3 (Default: 1).
--led-multiplexing=<0..6> : Mux type: 0=direct; 1=Stripe; 2=Checkered; 3=Spiral; 4=ZStripe; 5=ZnMirrorZStripe; 6=coreman (Default: 0)
--led-pixel-mapper : Semicolon-separated list of pixel-mappers to arrange pixels.
Optional params after a colon e.g. "U-mapper;Rotate:90"
Available: "Rotate", "U-mapper". Default: ""
--led-pwm-bits=<1..11> : PWM bits (Default: 11).
--led-brightness=<percent>: Brightness in percent (Default: 100).
--led-scan-mode=<0..1> : 0 = progressive; 1 = interlaced (Default: 0).
--led-row-addr-type=<0..2>: 0 = default; 1 = AB-addressed panels; 2 = direct row select(Default: 0).
--led-show-refresh : Show refresh rate.
--led-inverse : Switch if your matrix has inverse colors on.
--led-rgb-sequence : Switch if your matrix has led colors swapped (Default: "RGB")
--led-pwm-lsb-nanoseconds : PWM Nanoseconds for LSB (Default: 130)
--led-no-hardware-pulse : Don't use hardware pin-pulse generation.
--led-slowdown-gpio=<0..2>: Slowdown GPIO. Needed for faster Pis/slower panels (Default: 1).
--led-daemon : Make the process run in the background as daemon.
--led-no-drop-privs : Don't drop privileges from 'root' after initializing the hardware.
This will advertise itself as a PixelPusher http://www.heroicrobotics.com/products/pixelpusher device on the network. Number of 'strips' will be number of rows, so 16 or 32 multiplied by the parallel panels (1 .. 3).
The -u parameter specifies the size of the allowed UDP packets. Some network
switches (and the original PixelPusher hardware) don't like large packets
so the default is a conservative 1460 here.
But since we have a lot of pixels, using the highest number possible is desirable so ideally we can transmit a full frame-buffer with one packet (use something like 65507 here):
sudo ./pixel-push -u 65507
Even if the network supports it, sometimes sending devices limit the packet size (e.g. iOS, 8192 seems to be the limit of packets to send; important if you use LED labs softare) so we have to change:
sudo ./pixel-push -u 8192
You can control these for instance with the Processing framework http://processing.org/. The processing framework already has a contrib library section that allows you to select PixelPusher supporting libs.
Another software supporting the PixelPusher support is L.E.D. Lab http://www.ledlabs.co/
If you use the artnet bridge, you can specify the artnet-universe and the
artnet-channel with the -a option:
sudo ./pixel-push -a1,1
Generally, if you want larger displays, it is suggested to first use the feature of connecting multiple parallel chains to one Raspberry Pi; the adapter in the underlying project provides three outputs.
If you have the Adafruit HAT, then you only can do one chain, but you can
arrange them in a sideways 'U' shape to get a more square display. This
then can be mapped correctly with --led-pixel-mapper="U-mapper".
Here are four panels arranged in a square on a single connector, typically something you might want do do if you want a 64x64 arrangement of four 32x32 displays on an Adafruit HAT (which only provides one chain):
[<][<] }--- Pi connector #1 (looking from the front)
[>][>]
(--led-pixel-mapper="U-mapper" --led-chain=4 --led-parallel=1).
This is how it looks wired up from the back:
How about 6 panels ?
[<][<][<] }--- Pi connector #1
[>][>][>]
(--led-pixel-mapper="U-mapper" --led-chain=6 --led-parallel=1).
This even works if you have multiple parallel chains. Here is an arrangement with two chains with 8 panels each:
[<][<][<][<] }--- Pi connector #1
[>][>][>][>]
[<][<][<][<] }--- Pi connector #2
[>][>][>][>]
(--led-pixel-mapper="U-mapper" --led-chain=8 --led-parallel=2).
The --led-pixel-mapper="U-mapper" option essentially gives you half the
width of a panel, but double the height.
If you have a Raspberry Pi 2 or later consider assembling a display using parallel chains, for instance using the adapter that is provided in the RGB matrix project