DS90UB925Q-Q1: errors reported by back channel

Hello Jason,

What type of cable are you using between 925/926?

Thanks,

Casey 

It is a custom cable. ~6 feet long 26 awg standed twisted pair. It is one pair of a 8 pair cable.

More debugging shows that the crc error counter is increasing quickly in the 925 (register 0x0A and 0x0B). I see link appears to be solid so not sure what can cause this communication channel to have so many erros.

Hello Jason,

The forward and back channels for FPD-Link operate at significantly different frequency bands. You can learn more about it in this TIPL video: https://training.ti.com/ti-precision-labs-what-is-fpd-link?context=1139747-1138099-1139854-1139837  

CRC errors on the 925 indicates a problem with the back channel communications. One reason could be due to the custom cable insertion and return loss characteristics, particularly in the frequency range of the back channel communications. Usually we recommend high quality automotive STP cabling such as Dacar 535 to ensure the link can operate properly. I would suggest trying at least one other type of cable assembly to see if there is a difference. Also, there could be an influence from the connector/layout as well. 

Best Regards,
Casey 

I have a 925 eval board and connected my cable to this board. Testing showed no errors. This tells me the cable is good and the 926 is good. The issue seems to be on my board. Not sure what to check to see where the issue could be. I went through all pins and they are hooked up just like the eval board. 

Any suggestions on what to check to point me to the issue? I looked at some voltage levels and nothing stands out so not sure what to look at next? I have 40 boards and they all have the same issue. Could it be a cap value or voltage level that can cause this? How can I verify my video link is good and it is just my comm channel that has a issue?

Hello Jason,

Can you share the schematic for your board? We will review it and see if there are any standout issues. 

Best Regards,

Casey 

Here is the page with the 925. The 3.3v supply comes from a LDO. There is a common mode choke just like the ones on the ref design on a different page as well.

Please let me know what you find thanks for the help. I did disconnect almost everything else on the board. This page shows the only 2 parts that are still loaded and connected.

Please be sure not to share this as it is propitiatory. 

925_page.pdf

Hello Jason,

Can you please try changing R119 to 0 ohms instead of 0.1ohms?

Best Regards,

Casey 

What is the point of this resistor? It is on the eval board schematic?

Changing it made no difference.

Hello Jason,

CMF is a common mode filter for the high speed transceiver. I see that the EVM board schematic shows a 0.1ohm resistor but I would suggest going with the datasheet recommendation of only tying to GND with a 0.1uF cap. Other than that I don't see any significant issues with your schematic. One thing I would suggest is that if you are not using the DE signal, then tie it to either GND or VDD but do not leave it floating. 

One more question for you - what is the PCLK frequency being used here?

Best Regards,

Casey 

I just checked and the PCLK line is being feed with 4mhz. My lfmode is low. Looking at the data sheet I think this could be a isse right? This needs to be at least 15mhz right? Would this cause a issue to have it running at a lower speed? I do see if I put the camera driver into reset and the PCLk just floats high all the comm errors seem to go away?

Hello Jason,

Yes this could be the issue. 925 can actually support down to 5MHz PCLK but you must select LFMODE = H in the MODE_SEL strapping which is not being done currently in the schematic. However the device is not qualified for operation at 4MHz so I would not recommend this overall as we can not assure the reliability. 

Best Regards,

Casey