DS90UB953-Q1: Cannot access serializer when BC frequency is 50 Mbps

Hi Alex,

Is this a repeatable issue? U1 never works with 50 Mbps and U2 always does?

Can you share the deserializer & serializer register dumps for these two situations?

I would expect both the 960's to work with the 953 with a 50 Mbps back channel.

Regards,

Zoe

Hi Zoe,

yes it's repeatable. It's a default issue on our Fusion Boards. You are correct.

Enclosed the register dumps:

Regdump.xlsx

Regards,

Alex

Hi Alex,

Thanks for sharing the additional information. The AEQ_STATUS (0xD3) is pretty different between the two 960s.

After initializing the serializer and deserializer, can you try setting the AEQ_RESTART bit (0xD2[3]) and then see what value AEQ_STATUS settles on for each of the deserializers? 

Regards,

Zoe

Also, can you confirm that in both cases the serializer was connected to Port 0 of the deserializer?

Hi Zoe,

we used Port 1 of the deserializer but there is no difference between the Ports.
After I restart the AEQ it doesn't change the AEQ status at 50 Mbps.
If I set the BC frequency to 25 Mbps. The AEQ status is stable with a value of 0x02 in adress 0xD3.

Regards,
Alex

Hi Alex,

The register dumps you provided had Port 0 selected for reads in register 0x4D. Some registers (including AEQ_STATUS) are port-specific. If you are using Port 1, it would be helpful to have the register dump after changing 0x4D[5:4] to 01 to select Port 1 for register reads.

Are you using or are you familiar with Analog LaunchPad (ALP)? If you are able to, please go to the Margin Analysis tab and configure as below, then hit the "Start Margin Analysis" and share the results with me for both the 960's using 50 Mbps back channel. Please check that you install both the main ALP program and the ALP-PROFILE-UPDATE.

Thanks,

Zoe

Hi Zoe,

now the Regdump with the selected Port 1:

Regdump_Port1.xlsx

The value in the 0xD3 register is stable in U2. In U1 the value is floatend.

And the Margin Analyse of U2:

How can I change the I²C - Adress in the ALP to select the U1?

Regards,

Alex

Hi Alex,

Thanks for the additional info. You can change the I2C address by going to the scripting tab and using the command board.devAddr = [device address]

I will look through the register dumps and let you know if I find any unexpected values. I am also reaching out to the Fusion board team to see if they have come across this before.

Could you also let me know what cable type & length you are using to connect the camera module to the Fusion board?

Regards,

Zoe

Hi Zoe,

I use a 5.50m coax cable.
If I use the command in the ALP the programm didn't change the device. In Information there is still 0x6C and the analysis also run for this device.

Regards,
Alex

Hi Alex,

After running the command, can you check the value of register 0x00 in the registers tab?

Another option is to directly run the margin analysis script:
1. Go to C:\Program Files (x86)\Texas Instruments\Analog LaunchPAD v1.57.0010\PreDefScripts\DS90UB960\ub960_margin_analysis_script
2. Copy this folder to your desktop or other location you can write to.
3. Open ub960_margin_analysis.py and change devAddr_DES = 0x7A to the desired address.
4. Go to the scripting tab, press the run button, navigate to ub960_margin_analysis.py and say OK. While the script is running, leave the scripting tab selected (don't go to the info tab, etc.)

Regards,
Zoe

Hi Alex,

Are you seeing this problem with multiple cables? Can you try swapping the cable and seeing if you get the same results?

Regards,
Zoe

Hi Zoe,

the issue was if I switch to the information tab the programm resets the I²C Adress to 0x6C.

Now I got the Analysis for U1:

I tried the same with an other sensor (IMX390). The cable of this sensor is arround 1m but it's the same graph like ahead.

Regards,

Alex

Hi all.

If you found solution, can you share it on the forum, I have almost same problem with UB960 and IMX390 sensor.

Best Regards,

Stefan.

Hi,

Any updates about this issue?

Regards.