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.

DS90UB954-Q1: Vpoc noise below 10mVpp

Part Number: DS90UB954-Q1

Dear TI team, 

Acc. to the datasheet of DS90UB954 (SNLS570B):

"The VPOC noise must be kept to 10 mVp-p or lower on the source / deserializer side of the system. The VPOC fluctuations on the serializer side, caused by the transient current draw of the sensor and the DC resistance of cables and PoC components,must be kept at minimum as well. Increasing the VPOC voltage and adding extra decoupling capacitance (> 10 µF) help reduce the amplitude and slew rate of the VPOC fluctuations."

Currently we are using an active differential probe and limiting the BW of the oscilloscope to 20MHz in order to focus in the noise produced due to load changes (in our case it is due to increase of consumption of the cameras when sending the frames). 

The measuring point used is right before the inductor of the Vpoc network. 

Currently we have a floor noise of ~15mVpp with some spikes produced by the loads (cameras). Noise Vpp is around 50mVpp (measured in the described test point, with BW limited to 20MHz and using active differential probe). 

I'd like to ask the following:

  • What is your recommended measuring method for this voltage noise in the VPOC?
  • How strict is this requirement to have a proper link in our system? Is it possible to "relax this requirement" and still be in a good position to ensure a proper FPD-link functionality? We have also measured the eye diagram of back-channel and forward-channel and they are both inside specs. 

BR,

Vicent Climent

  • Hi Vincent,

    The proper way to measure 10mVpp PoC Rin+ noise is by following the steps below:

    -Connect the deserializer and serializer using a cable, turn on the power supply
    -Turn off the deserializer and serializer (by pulling the PDB low)
    -Measure the noise at the Rin+ pin of the deserializer
    -Set the scope to 0 to 50MHz bandwidth
    -Use a short probing tip for ground
    -Account for noise floor in the probe (Actual noise from PoC = measured noise – noise floor)
    Best,
    Jiashow
  • Hello Jiashow,

    Thank you very much for your quick reply. 

    I have measured the Vpoc noise as you suggested. Unfortunately we are still outside your specification. For me it looks like our PS for the Vpoc is not able to follow up with the transcient caused by the sudden change in the load (camera sending frames). So I'd like to address you my last question from the previous message:

    • How strict is this requirement to have a proper link in our system? Is it possible to "relax this requirement" and still be in a good position to ensure a proper FPD-link functionality? We have also measured the eye diagram of back-channel and forward-channel and they are both inside specs.

    I have still one further question, in your answer you are recommending to measure this noise in Rin+. However in the datasheet Vpoc signal is described before the coil of the Vpoc network (see pic below). 

    Shall I understand that the DS it is talking about Vpoc noise coupled to RIN+ pin of the deserializer?

    Thx!

  • Hi Vincent,

    Yes, PoC noise going into the Rin pin can cause signal degradation which may reduce the maximum cable length the device can support, but it's difficult to quantify that. Have you tried adding more decoupling caps to bring the noise down? Have you used the margin analysis program to look at the eye margin?

    Best,

    Jiashow

  • Hi Jiashow,

    Yes, I have tried to to increase more decoupling caps but it  doesn't look to help much. The only thing that was giving us some minimum improvement was improving the transient response of our power supply. However as said the improvement was not much, from 50mVpp to ~40mVpp, and this solution could bring us problems at "extreme" voltage conditions. So we would like to avoid it as much as possible.

    Regarding the MA, yes, we have performed it as well as a measurement of the eye diagram in the CML output. Both are OK acc to your specs. As reference, we have used the following guide to validate the margin analysis results.

    http://www.ti.com/lit/an/snla301/snla301.pdf

    That is why I was wondering if it was possible to ensure the proper link functionality with margin analisys and eye diagram measurement.

    BR,

    Vicent

  • Hi Vincent,

    If the MAP tool shows good results, then your specific system should work fine.

    The PoC noise requirement is based on the worst condition (ie. longest cable) where it signal can potentially degrade when noise is too high in link.

    Best,

    Jiashow