ESP32-HUB75-MatrixPanel-DMA/FM6126A.md

The mystery of control registers for FM6126A chips

Datasheet for this chis chip is in chineese and does not shed a light on what those two control regs are.

An excellent insight could be found here

https://github.com/hzeller/rpi-rgb-led-matrix/issues/746#issuecomment-453860510

So there are two regs - REG1 and REG1,

one could be written with 12 clock pusles (and usually called reg12, dunno why :))

the other one could be written with 13 clock pusles (and usually called reg13, dunno why :))

So I've done some measurmens on power consumption while toggling bits of REG1 and it looks that it could provide a fine grained brighness control over matrix.

There are 6 bits (6 to 11) giving an increased brighness (compared to all-zeroes) and 4 bits (2-5) giving decreased brighness!!!

So it seems that the most bright (and hungry for power) value is

bool REG1[16] = {0,0,0,0,0, 1,1,1,1,1,1, 0,0,0,0,0}; and not {0,1,1,1,1, 1,1,1,1,1,1, 1,1,1,1,1} as it is usually used.

I'm not sure about bit 1 - it is either not used or I was unable to measure it's influence to brightness/power.

Giving at least 10 bits of hardware brightness control opens pretty nice options for offloading. Should dig into this more deeper.

Here are some of the measurments I've took for 2 64x64 panels filled with white color - reg value and corresponding current drain in amps.

REG1	bit value	Current, amps
REG1	0111111 00000	>5 amps
REG1	0100010 00000	3.890 amp
REG1	0100000 00000	3.885 amp
REG1	0011110 00000	3.640 amp
REG1	0011100 00000	3.620 amp
REG1	0011000 00000	3.240 amp
REG1	0010010 00000	2.520 amp
REG1	0010001 00000	2.518 amp
REG1	0010001 10000	2.493 amp
REG1	0010000 00000	2.490 amp
REG1	0010000 11110	2.214 amp
REG1	0001100 00000	2.120 amp
REG1	0001000 00000	1.750 amp
REG1	0000100 00000	1.375 amp
REG1	0000010 00000	1.000 amp
REG1	0000000 00000	0.995 amp
REG1	0000001 11111	0.700 amp
REG1	0000000 01111	0.690 amp
REG1	0000000 10000	0.690 amp
REG1	0000000 11110	0.686 amp

/Vortigont/