Hi team,
At present, customer use DS90UB954+953, and the CPU is Hisilicon's HI3516. In the process of using, it is found that some boards have the problem of low probability frame dropping. Read the HI3516 ISP error frame related register, and the frame dropping rate is about 1/10W.
Refer to the document " Power-over-Coax Design Guidelines for DS90UB953-Q1", try to adjust the POC filter circuit, adjust the model of magnetic beads and inductors, the results have changed, but the effect is not ideal, can not completely solve the frame dropping problem.
At present, the 954 end POC filter circuit is configured as NLCV32T-100K+MPZ1005S331+MPZ1005F470 (fully referring to the configuration of EVM board), and the 953 end is NLCV32T-100K+BLM18HE102+BLM18HE102+MPZ1005F470
Q1. At present, it is through reading the frame dropping information of HI3516 to judge whether the frame is dropped. The test time is long, and it is often an hour or even a day to find the frame dropping. I don't know if there is any way to quickly reproduce the problem.
Q2. Try the BIST function, the 954 register is configured as 0x4c: 0x01; 0xb3: 0x01, and then read 0x57 is 0, touch the signal line or even short the signal pin of FPD link to GND, and 0x57 is 0. I don't know whether the setting is correct.
Q3. Later, tried to modify the register(SFILTER_CFG) 0x41 of 954. This was originally the default value 0xA7, then it was changed to 0xA9, which was obviously improved, and then it was changed to 0XAA, which was improved a little bit. The description of the related content in spec is less, can you provide more detailed information? Is there any risk in changing to 0XAA? thank you in advance!
Q4. What suggestions do you have for the current frame dropping situation?
