Hello Leila,
1. The INIT sequence duration depends on the number of LEDs (max. 4072) and the time propagation delay (max. number will be defined by Inova but should be aprox. 8μs). The INIT command travels downstream and takes 4072*8μs and is being processed in every LED (~ns, can be ignored). After this, the 4072 response frames (length of 50 bits + 16 bit latency) travel upstream and take a tpd of 4072*8μs again. Due to fabrication tolerances, the max. bitwidth is +30% of the nominal value (500ns). The total result, for 4072 LEDs is approximately 245ms.
2. The "Set" commands all last the same amount of time to be transmitted. The transmission mode, however, can impact CPU load during the transmission as follows:
- Interrupts mode, CRC enabled:
The total time is the interval between the A1 and A2 markers - 68 us. This is the time between calling the digLED_Set_RGB() function, and digLED_Callback is called with the "TRANSMISSION_COMPLETE" event. The second channel represents the CPU usage during the transmission (Frame formatting, CRC calculation, and the transmit interrupts - in total 18.5 us)
- DMA mode single frame, CRC enabled:
Total time is 77 us, with the total CPU load 18.7 us. DMA mode is not more efficient than interrupts mode for single commands.
- Block mode DMA, CRC enabled:
This capture represents the Block transmission of 80 SET_RGB or SET_DIM commands using the single "digLED_Send_Cmd_Block" function. This is the most efficient way to address every LED in the strip. In this case, all the processing is done before the transmission begins (512 us) after which DMA sends the 80 frames (4.14 ms) - so about 6.4 us CPU processing time for each frame. The screenshots above were taken using S32K144 running at 80 Mhz, 40 Mhz FlexIO clock, cache enabled.
The times you are citing from the release notes refer to the CRC calculation time for one frame. It was measured on S32K116 as it is the least powerful. Those times are included in the CPU usage shown in the graphs.
Hope this helps.
Best regards,
Dragos
