Only update of BurstSPI-Library
Dependents: WordClock TI_NEOPIXEL_SPI
Fork of PixelArray by
neopixel.cpp
- Committer:
- charly
- Date:
- 2017-11-02
- Revision:
- 6:b8f53cf2b54d
- Parent:
- 4:c3b314df3dfe
File content as of revision 6:b8f53cf2b54d:
#include <stdint.h>
#include "mbed.h"
#include "neopixel.h"
namespace neopixel
{
PixelArray::PixelArray(PinName out, ByteOrder byte_order, Protocol protocol)
: spi_(out, NC, NC), byte_order_(byte_order), protocol_(protocol)
{
if (protocol_ == PROTOCOL_800KHZ) {
// 800kHz bit encodings:
// '0': ----________
// '1': --------____
// The period is 1.25us, giving a basic frequency of 800kHz.
// Getting the mark-space ratio right is trickier, though. There are a number
// of different timings, and the correct (documented) values depend on the
// controller chip.
//
// The _real_ timing restrictions are much simpler though, and someone has
// published a lovely analysis here:
// http://cpldcpu.wordpress.com/2014/01/14/light_ws2812-library-v2-0-part-i-understanding-the-ws2812/
//
// In summary:
// - The period should be at least 1.25us.
// - The '0' high time can be anywhere from 0.0625us to 0.5us.
// - The '1' high time should be longer than 0.625us.
//
// These constraints are easy to meet by splitting each bit into three and packing them into SPI packets.
// '0': 100 mark: 0.42us, space: 0.83us
// '1': 110 mark: 0.83us, space: 0.42us
spi_.frequency(2400000); // 800kHz * 3
spi_.format(12); // Send four NeoPixel bits in each packet.
} else {
// 400kHz bit encodings:
// '0': --________
// '1': -----_____
//
// Timing requirements are derived from this document:
// http://www.adafruit.com/datasheets/WS2811.pdf
//
// The period is 2.5us, and we use a 10-bit packet for this encoding:
// '0': 1100000000 mark: 0.5us, space: 2us
// '1': 1111100000 mark: 1.25us, space: 1.25us
spi_.frequency(4000000); // 400kHz * 10
spi_.format(10); // Send one NeoPixel bit in each packet.
}
}
static void SendFourBits(BurstSPI& spi, uint32_t bits)
{
// Encode '0' bits as 100 and '1' bits as 110.
// We have this bit pattern: 00000000abcd
// We want this bit pattern: 1a01b01c01d0
uint32_t ac = (bits * 0x088) & // 0abcdabcd000
0x410; // 0a00000c0000
uint32_t bd = (bits * 0x022) & // 000abcdabcd0
0x082; // 0000b00000d0
static uint32_t const base = 04444; // 100100100100
spi.fastWrite(base | ac | bd); // 1a01b01c01d0
}
static void SendEightBits(BurstSPI& spi, uint8_t bits)
{
int zero = 0x300; // Encode zero as 0b1100000000
int one = 0x3e0; // Encode one as 0b1111100000
for (int i = 128; i >= 1; i >>= 1) {
spi.fastWrite((bits & i) ? one : zero);
}
}
void PixelArray::send_pixel(Pixel& pixel)
{
// Pixels are sent as follows:
// - The first transmitted pixel is the pixel closest to the transmitter.
// - The most significant bit is always sent first.
//
// g7,g6,g5,g4,g3,g2,g1,g0,r7,r6,r5,r4,r3,r2,r1,r0,b7,b6,b5,b4,b3,b2,b1,b0
// \_____________________________________________________________________/
// | _________________...
// | / __________________...
// | / / ___________________...
// | / / /
// GRB,GRB,GRB,GRB,...
//
// For BYTE_ORDER_RGB, the order of the first two bytes are reversed.
uint8_t byte0 = (byte_order_ == BYTE_ORDER_RGB) ? pixel.red : pixel.green;
uint8_t byte1 = (byte_order_ == BYTE_ORDER_RGB) ? pixel.green : pixel.red;
if (protocol_ == PROTOCOL_800KHZ) {
SendFourBits(spi_, (byte0 >> 4) & 0xf);
SendFourBits(spi_, (byte0 >> 0) & 0xf);
SendFourBits(spi_, (byte1 >> 4) & 0xf);
SendFourBits(spi_, (byte1 >> 0) & 0xf);
SendFourBits(spi_, (pixel.blue >> 4) & 0xf);
SendFourBits(spi_, (pixel.blue >> 0) & 0xf);
} else {
SendEightBits(spi_, byte0);
SendEightBits(spi_, byte1);
SendEightBits(spi_, pixel.blue);
}
}
void PixelArray::update(Pixel buffer[], uint32_t length)
{
for (size_t i = 0; i < length; i++) {
send_pixel(buffer[i]);
}
wait_us(latch_time_us_);
}
void PixelArray::update(PixelGenerator generator, uint32_t length, uintptr_t extra)
{
for (size_t i = 0; i < length; i++) {
Pixel out;
generator(&out, i, extra);
send_pixel(out);
}
wait_us(latch_time_us_);
}
}
