• State TI Thinks Resolved
  • Locked Locked
  • Replies 9 replies
  • Answers 1 answer
  • Subscribers 34 subscribers
  • Views 482 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

DS90UB960-Q1: EMC- RE radiation 1.2 GHz - over-scale frequency point

Violet Lei
Intellectual 1856 points
Part Number: DS90UB960-Q1

Tool/software:

Dear team,

Here is my customer question, need your help, thanks a lot

customer's design:

935-960

question:

We found a sub-scale frequency point at 1.2 GHz of re radiation when doing EMC experiments, which was determined to be caused by the camera chip DS90UB960 output, please work with TI software to determine if any of the frequency points within the chip would affect it. Can software do frequency adjustments or other ways to handle this problem.

But the margin does not meet the 3 db requirement. 

  • 0
    TI__Mastermind 28320 points

    Hi Violet,

    Just to summarize the question, it seems that customers are failing their EMC requirements (3dB from the limit line) at 2.41795GHz, 2.4681GHz, 1.2066GHz, and 2.44765GHz in a system with the 960 device. Since this 960 device can only operate at up to 4.16Gbps (~2.08GHz), only the peak at 1.2066GHz is suspected to be related to the 960 device.

    However, has it been confirmed that the FPD-Link devices are causing this spike at 1.2GHz?

    What is the serializer being used in the system and what MODE are the FPD-Link devices operating at?

    If they pull the PDB pin to GND on the serializer, then does that spike disappear?

    Best,

    Justin Phan

  • 0 in reply to Justin Phan
    TI__Intellectual 1856 points

    Hi Justin

    Yes! Your Summary is great~

    1\has it been confirmed that the FPD-Link devices are causing this spike at 1.2GHz?

    Their test way:When they do EMC testing, a spike is generated when the camera is attached;
    Remove the camera, leaving only the coax and fpdlink and SoC. the spike is still there.

    2\What is the serializer being used in the system and what MODE are the FPD-Link devices operating at?

    935-960; you can get the MODE configuration in the .

    3\ If they pull the PDB pin to GND on the serializer, then does that spike disappear?

    They'll try this test,  I'll tell you when the results come out 

  • 0 in reply to Violet Lei
    TI__Mastermind 28320 points

    Hi Violet,

    Violet Lei said:

    1\has it been confirmed that the FPD-Link devices are causing this spike at 1.2GHz?

    Their test way:When they do EMC testing, a spike is generated when the camera is attached;
    Remove the camera, leaving only the coax and fpdlink and SoC. the spike is still there.

    If they remove the camera and the spike is still there, then that is actually proof that the EMC spike is not related to the FPD-Link devices. The serializer generates the Forward Channel signal that is up to 4Gbps (2GHz). The deserializer by itself only generates a Back Channel signal that can go up to 50Mbps (25MHz). If the serializer is not connected, then no FPD-Link device is generating any GHz signal. It must be something else in the system that is causing the EMC spike.

    Violet Lei said:

    2\What is the serializer being used in the system and what MODE are the FPD-Link devices operating at?

    935-960; you can get the MODE configuration in the .

    If they are connecting a 960-935, they are actually in the wrong MODE. You must use the CSI-2 Non-Synchronous Mode or CSI-2 Synchronous Mode when connecting a 960 deserializer to a 935 serializer. The RAW modes are for when you are connecting the 960 to a DVP serializer, such as the 913A/933.

    I would recommend submitting a schematic review on E2E, but this requires sharing the customer schematic. If you do not want the schematic posted on this public forum, then you can reach out to a local TI FAE and have them submit the request on your behalf through our internal E2E forum.

    Violet Lei said:

    3\ If they pull the PDB pin to GND on the serializer, then does that spike disappear?

    They'll try this test,  I'll tell you when the results come out 

    I am not sure of the specific methodolgy they are performing for EMC testing, since different companies have slightly different requirements, but when doing their EMC testing, we do recommend not disconnecting the camera module. Keep all of the metal enclosures around the PCB that you have setup. One way to test if the serializer is the issue is to pull the PDB pin to GND. This puts the serializer device in a low-power state, where all high-frequency operations are suspended. If the EMC spike still occurs, then it is not the serializer that is generating that spike.

    Best,

    Justin Phan

  • 0 in reply to Justin Phan
    TI__Intellectual 1856 points

    Thanks a lot ! It's a clear explanation！！(I am not sure if the customer has no problem, and I will turn this ticket off when I get an accurate response from them， thanks again）

  • 0 in reply to Justin Phan
    TI__Intellectual 1856 points

    Hi~Justin

    excume me~

    The customer still think the spike from  the 960, when they disconnect resistor, and the energy near the chip pin is higher ,you can see the video they sent.

     

  • 0 in reply to Violet Lei
    TI__Mastermind 24971 points

    Hi Violet,

    Justin is out today and will be back on Monday to continue this debug. Would you also be able to re-upload video or send over email? It seems that it cannot be seen through E2E.

    Best,

    Thomas

  • 0 in reply to Violet Lei
    TI__Mastermind 28320 points

    Hi Violet,

    Like Thomas mentioned, the video that shared is not visible through E2E.

    However, the key point that should be clarified with the customer is the fact that the only the Forward Channel FPD-Link signal generated by the serializer has a possibility of causing the EMC spike at ~1.2GHz. However, customer needs to confirm if the FPD-Link serializer is actually the cause. If the serializer is disconnected or PDB=LOW on the serializer, then there is nothing in the FPD-Link chipset that can is operating at that frequency. Even the max CSI rate of the DS90UB960 device is 1.6Gbps (800MHz).

    The customer should selectively disable different devices in their system and see which ones cause the EMC spikes to disappear. This will help identify which device is causing the EMC spike.

    Best,

    Justin Phan

  • 0 in reply to Justin Phan
    TI__Intellectual 1856 points

    Dear Justin,

    I sent the video by mail, perhaps because the file was too large, it was not sent successfully.
    Thank you for the response, the solution has been sent to the customer, but they have feedback that they cannot pull PDB low in the serializer.

    Any other actions to try?

  • 0 in reply to Violet Lei
    TI__Mastermind 28320 points

    Hello Violet,

    If the customer cannot pull PDB=LOW on the serializer, then here are some things they can try:

    1. Can they turn off the PoC power to the camera module?
    2. Can they try to change the FC rate?
      1. If they change the FC rate to 2Gbps (1GHz), then the FPD-Link serializer will no longer be generating a signal frequency of up to 2.08GHz. If the spike is still there, then they need to investigate something else in their system or setup that could be causing the spike at 1.2GHz.
      2. I am still unclear on the MODE they are using to connect to the camera module. RAW10 mode is only for connecting a 960 device to a 913/933 device. Do you know what MODE they are using to connect to the 935-960 devices?

    Best,

    Justin Phan