This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DS90UB949A-Q1EVM: Configuration for single coax

Part Number: DS90UB949A-Q1EVM
Other Parts Discussed in Thread: DS90UB948-Q1, DS90UB949A-Q1, ALP

I've got a DS90UB949A-Q1 EVM (serializer) connected to a DS90UB948-Q1 EVM (deserializer). I'm attempting to use a single coax cable as an interface between the two. I can successfully get the two boards to lock (Lock LED1 lights up on the deserializer board) if I have two coax cables connected, one to Dout0+/Rin0+ and one to Dout1+/Rin1+. However, if I remove one of these coax cables and just leave the one connected to Dout0+/Rin0+, the Lock LED1 will blink rapidly. I'm assuming the link is not valid at this point. Is there a register or configuration I need to set to get these chips to communicate over one coax instead of two?

The changes I've made to each board from factory defaults are:

1. On the serializer EVM:

          -Add 0 ohm jumpers between the FAKRA connector and the test points to connect the SMA connectors.

          -Set switch S6 to position 3 instead of position 1.

          -AC coupling cap on the active line (Dout0+) is 100nF. AC coupling caps on inactive lines (Dout0-, Dout1+, Dout1-) are 47nF. Inactive lines are also terminated with 50 ohms. 

2. On the deserializer EVM:

         -Add 0 ohm jumpers to R37, R38, R63, and R64 to connect the SMA connectors.

          -Set switch S2 to position 2 instead of position 1.

          -AC coupling cap on the active line (Rin0+) is 100nF. AC coupling caps on inactive lines (Rin0-, Rin1+, Rin1-) are 47nF. Inactive lines are also terminated with 50 ohms.

I also tried a configuration where the inactive positive line (Dout1+/Rin1+) was not terminated and used 100nF instead of 47nF (basically the same as the active line except there was no coax cable ) but got the same result of a rapidly blinking Lock LED1.

 

  • Hi Sarah, 

    Thanks for including all the details, the hardware changes listed should be fine. When you remove the Port 1 coax cable, are you doing a PDB reset on the boards after? The PDB button on the 949 EVM is on the left of S2 and on the 948 EVM, it's a toggle on S3.3 (red switch bank). 

    The default setting on the boards are to auto-detect the FPD-Link input mode but a reset is recommended if the cable is removed after power on. We also recommend to set register 0x5B[7] = 0 on the 949 regardless of the hardware configuration (single-ended or differential). 

    Best,

    Shu

  • Hi Shu, 

    When I remove the port 1 coax cable, I've been doing a clean power reset. So I power down the boards, remove the coax, and power the boards back up. Just to double check I removed the coax with the boards powered on and performed a PDB reset, but the behavior stayed the same. I also reset the 949 register to be twisted pair (0x5B[7] = 0), that didn't change the board behavior. 

    On that note, is there a way to set the 0x5B register through the Analog Launchpad tool? It's not listed in the Registers tab. But it looks like I could force Single Link Mode by setting bits 0 and 1.

    I'm starting to suspect the issue is that my HDMI output is too fast for a single link. When I unplug one coax, I can watch on the HDMI tab as it repeatedly flips back and forth between single and dual mode (see pics). It's showing HDMI input as 1280x720 and 73MHz, which I think is too fast for single link. Right now I can't adjust these settings because I don't have an output monitor on the other side (I'm still waiting for an adapter), but I'm hoping once I do I can adjust the resolution and/or refresh rate and slow down the link. I will keep you posted once I'm able to test this out. 

      

  • Hi Sarah, 

    The 949 - 948 pairing can support pixel clock up to 192MHz in dual lane (96MHz in single) so the 1280x720 resolution should be fine. This sounds like a signal integrity issue rather than SW configuration so I have some additional reworks to recommend that may improve the signal path with the single lane coax connection. 

    1. Remove the HSD connectors on both SER and DES

    2. Remove Common Mode Chokes on SER and short those paths with 0 ohm resistors

    Regarding the ALP question, it looks like the device profile is incorrect, I noticed in the provided screenshots that the device recognized/selected is a 925 serializer and not the 949. 0x5B should be shown in the 949 register page. Another way to set registers is through the scripting tab with the following commands. 

    board.WriteI2C(dev addr, reg addr, reg data)

    board.ReadI2C(dev addr, reg addr)

    board.WriteI2C(0x18, 0x5B, 0x00)

    Best,

    Shu

  • Thank you Shu! I removed both HSD connectors and the common mode chokes on the SER board, and the LockLED1 now lights up solid with only one coax. I appreciate your help with this!