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.

TIDA-00421: Video gets freeze when IR is turned on.

Part Number: TIDA-00421

Hi,

I've designed a board using TIDA-00421 as reference and added 4 IR LEDs (2 series) to this board. Whenever the IR is turned on the video gets freeze. The rated current is 100mA. I've set the current to 90mA(Coaxial voltage 5.8V). Current was reduced to 75mA but the video get freeze in 5min(Coaxial voltage 8V). Then the current was reduced to 40mA the video gets freeze after several hours. Could please let me know what may be the reason for this.

Thanks

  • Hi Rohith,

    I am interested in several things. Just to confirm, the video freeze happens later and later after operation (IR LED on) with increasing Coaxial Voltage, after 5 minutes of operation at 8V, after several hours of operation at ~15 V? And the video freeze happens immediately or soon after operation at 5.8 V?

    Could you please monitor and provide screenshots for:

    -3.3-V Rail voltage during this pink frame

    -1.5-V Rail voltage during pink frame

    -LOCK signal on deserializer before, during and after pink frame

    Finally, could you please share schematics and PCB files?

    -Luis

  • Hi Luis,

    Sorry for the confusion. There is a correction it is not 8V its 5.8V.

    And the video freeze happens immediately or soon after operation at 5.8 V?

    When the IR current is 90mA it freezes immediately when IR is turned on.

    -LOCK signal on deserializer before, during and after pink frame

    DS90UB964 is the serializer used.

    I will provide the screenshots.

    The schematics and PCB files are confidential.

    Thanks
  • Hi Rohith,

    Thanks for clarifying. So we have established that the video freeze happens immediately with a lower PoC voltage, and after 5 minutes - several hours with higher PoC voltage.

    I will await the screenshots.

    Thanks,

  • Hi Rohith,

    I now see the 964 does not have a LOCK pin.

    Do you have a way to report the LOCK status, either to a GPIO or via SW? Knowing the LOCK status will help us eliminate a failing communication link as root cause.
  • Hi Luis,

    The video freeze happens immediately with a higher IR current , and after 5 minutes lowering the IR current - several hours with Lower IR current.
  • Hi Luis,

    I tried changing the camera (Swapping and changing) and the Pink frame is visible for this particular camera.
  • Hi Luis,

    The voltage waveform does not have any change during the pink frame. Even in trigger mode waveform change was not seen.

    Thanks
  • Hi Rohith,

    So a video freeze occurs with higher IR current, and after a longer period of time with lower current. Few more questions and confirmations:

    Which voltage was measured, and was it AC or DC?

    There is only one particular camera that is experiencing the Pink Frame, but all cameras are getting video freeze?

    What is the LED control topology (linear, boost, buck, etc.), and is the current coming directly off of the PoC filter, or off a separate system voltage?

    Would you be willing to share schematics with me through email?

  • Hi Luis,

    Which voltage was measured, and was it AC or DC?:-
    DC
    There is only one particular camera that is experiencing the Pink Frame, but all cameras are getting video freeze?
    Yes. Pink frame and Freeze are different issues. Pink Frame occurs even IR Leds are OFF.
    What is the LED control topology (linear, boost, buck, etc.), and is the current coming directly off of the PoC filter, or off a separate system voltage?
    LED is turned on and off using a load switch. Yes, the current is directly coming off of the PoC filter.

    Note: I tried to load this POC voltage (Instead of IR led current) then also the video gets freezed.

    Would you be willing to share schematics with me through email?
    Yes, I can share the Image file of the schematics.
  • Hi Rohith,
    Thanks for the information and schematic shots. I need a few more things from you before we can determine what is causing the video freeze. Knowing the LOCK status before and during the video freeze (transition into video freeze) will allow us to determine whether or not the link is dropping.
    Can you read the LOCK status from the '964 via SW or i2c sniffer before and during a video freeze? If the LOCK signal is low, this would indicate the link has dropped.
  • Hi Luis,

    Sorry I missed a point to mention.

    When IR current is low, the video freezes after several minutes. During this time the video starts to slow down. The video gets lagging and then freezes.

    Thanks
    Rohith
  • Hi Luis,

    Register 0x4D is checked during the IR ON-OFF activity. The value read is 0XC3 when IR is off. Value 0XD3 is read rarely(a flicker is seen in one or 2 of the cameras) when IR is ON. During freezing it can be either 0XC3 or 0XD3.

    Thanks
    Rohith

  • Hi Rohith,

    Excellent, thank you for this critical info. Bit 4 in register 0x4D is set whenever a LOCK status change has been detected in the receiver (964) since the last read of the register.

    So the value 0xC3 read from register 0x4D means the link is LOCKED and its lock status has not changed since it was last checked. 0xD3 means the link is currently LOCKED, but the link DID recently change status from LOCK to not locked.

    We can now confidently say that we are losing channel LOCK with LED ON-OFF activity.

    After reviewing the schematic and knowing that we are losing LOCK, I want to focus on FB5. It is recommended to use a ferrite bead with at least 1-k ohms at 100 MHz in 'front' of the PoC filter, closest to the connector (where FB5 is).

    Could you try replacing FB5 with the recommended 1-k-ohm at 100 MHz ferrite bead (looks FB6 would work, footprint allowing) and seeing what the change in operation is? If not, use something at least greater than 120-ohms.

  • Hi Luis,

    Replaced the FB5 with 1K@100MHz, still the issue is not resolved. Even tried with exact same filter as reference but still not resolved.

    During the pink frame the 0X4D register is C3.

  • Hi Rohith,

    Was there any difference in operation with the recommended filter? One more point about the filter schematic, it is also recommended to use a 0.1 uF decoupling cap after the 100uH inductor.

    The next thing I would like to see is the AC voltage ripple on VPOC, and across and C55, along with the DC input voltage to the 3.3-V supply during LED on condition.

    Also, is the IR LED current a constant DC load, or turned on/off quickly (load steps)?

    Regarding the pink frame, it looks like LOCK is not lost, so the cause is likely upstream from the PoC filter, perhaps one of the supply rails to the imager.

    When exactly does the pink frame occur? I know it is only one camera, but is this random, or during LED on/off cycles?
  • Hi Luis,

    Also, is the IR LED current a constant DC load, or turned on/off quickly (load steps)?
    Turned on and off quickly using the load switch. I tried with putting load resistor in VPOC. Video still gets froze.

    When exactly does the pink frame occur? I know it is only one camera, but is this random, or during LED on/off cycles?
    Random. Even when IR is OFF.

    Rohith

  • Hi Rohith,

    Is this load resistor switched on and off like the IR LEDs, or is it a DC load?

    I will await more voltage measurements.

    Could you also please share your PCB layout files with me via email?

  • Hi Luis,

    Please find the waveforms.

    Is this load resistor switched on and off like the IR LEDs, or is it a DC load?

    I tried both as DC load and by plugging the resistors. Both had freezing issue.

    Was there any difference in operation with the recommended filter? One more point about the filter schematic, it is also recommended to use a 0.1 uF decoupling cap after the 100uH inductor.

    I tried using exact same filter as in TIDA-00421. Even tried with 48MHz oscillator as in TIDA-00421.

    Thanks

    Rohith

  • Hi Rohith,

    The voltage ripple across C55 is far too large for the link to operate properly. Based on visual approximation, it appears your voltage ripple correlated with the imager frame rate (30fps) is roughly +/-200 mV, or 400 mVpp, and even larger during LED on. This is what is causing the disruption in video. The target ripple on the link (across C55) should be roughly 50 mV, 100 mVpp.

    To achieve this, there needs to be considerably more bulk decoupling capacitance by the IR LED load, and more inductance in between the PoC network and IR LEDs. Quick test could be swapping the 4.7uH inductors for 100 uH inductors. Be careful the DC drop is not too large for the 3.3-V supply. Increase C69 considerably as well, if possible up to 220uF. I would also recommend going to the highest possible PoC voltage you can support as well.

    Let me know what the results are after implementing these changes.
  • Hi Luis,


     Quick test could be swapping the 4.7uH inductors for 100 uH inductors.

       This didn't solve the issue

    Increase C69 considerably as well, if possible up to 220uF.

      This didn't solve the voltage dip.

    I would also recommend going to the highest possible PoC voltage you can support as well.

    Increased voltage to 12V. Still freeze issue is there.

    Tried to increase the Vpoc cap to 470V. Still drop is there.

    Tried feeding the Voltage right before FB5 using external bench power supply then the Freeze issue is not seen.

    Thanks 

    Rohith

  • Hi Rohith,

    I am sorry this did not solve the issue. However, I need to know if there were any improvements and by how much. Could you please provide a screenshot of the voltage ripple across C55?

  • Hi Luis,

    Please find the waveform.

    Thanks 

    Rohith

  • Thanks Rohith. Was there a difference in POC voltage between the screenshots you have provided? If so or not, could you please tell me what the POC voltage is?
  • Hi Luis,

    Right now the POC voltage is 11V.

    Rohith
  • Hi Rohith,

    What was the PoC voltage in the previous screenshots you provide?
  • Hi Luis,

    Earlier voltage was 5.8V

    Rohith
  • Hi Rohith,

    Thank you. Could you provide the same voltage ripple screenshot across C55 with the added 100 uH inductors and 200 uF caps and a VPOC of 5.8 V?
  • Hi Luis,

    Could you provide the same voltage ripple screenshot across C55 with the added 100 uH inductors and 200 uF caps and a VPOC of 5.8 V?
    One question, did you want me to replace inductor 4.7uH with 100uH or swap between the 100uH and 4.7uH?(L6 and L8)

    Thanks
    Rohith

  • Hi Rohith,

    Keep L6 and L8 in the same order, this is critical. For L3 and L4, please increase these inductances.

    I want to compare the voltage ripple across C55 with the original recommended filter (plus 0.1 uF decoupling after L8), against the voltage ripple across C55 with the new added inductance (100 uH x 2) and 220 uF capacitance on C69, both at the same VPOC. This will help me understand if there are improvements and what caused them.

    I would expect a reduction in voltage ripple across C55 with the extra inductance and capacitance.
  • Hi Luis,

    Please find the waveform.

    NOTE: This is measured on the board with 100uHs and 220uF and all other 3 cameras, which are unchanged(4.7uH), are connected.

     

    NOTE: This is measured on the board with 100uHs and 220uF and all other 3 cameras are not connected.

     Rohith

  • Hi Luis,

    Please note the TIDA-00421 evaluation kit has a ripple of 200mV when measured on C1 capacitor.

  • Hi Rohith,

    Thanks for the info.

    I want to refer back to an earlier post of your from April 17, where you mentioned feeding in power externally before FB5 not having freeze issues. Does this mean power was not delivered over coax, but externally, so only data was transmitted over the coax cable? This may be an important piece to this puzzle.

  • Hi Luis,

    Yes, that's right. This is done by placing one end ferrite bead FB5 on single pad and powering it from other end. Isolating power and signal, otherwise the signal in coax was filtered by bench power supply filter.

  • Hi Rohith,

    Thank you for confirming.

    It sounds like a complete fix may require new HW. All my recommendations apply, specifically having LOCK signal access to trigger on a scope, extra inductive and bulk capacitive (>470 uF) filtering and higher poc voltage (>12 V).

    If poc is required, I highly recommend an LED driver as this will provide slew rate control and a ripple reduction. It is difficult to control this using just a switch. Turning the switch on/off with little external control is causing the sharp voltage drop and sharp voltage rise. Having access to the LOCK signal will enable accurate timing of link loss (if it is lost) to best optimize your design. The higher poc voltage will reduce the magnitude of ripple on the line, further reducing the likelihood of link loss.

    If poc is not a requirement, it may be easier to use a shielded twisted quad STQ cable, or separate poc and LED power cables.

  • Rohith,

    Could you measure the slew rate on the VPOC when the LED is turned on?
  • Hi Luis,

    Do you want me to measure the slew rate of voltage after Load switch?

    If yes, We can actually control the slew rate of the load switch using C67. We tried increasing the slew rate of load switch. Now the rise time is 820ms with C67 -10uF.

    Blue is the Voltage after Load switch

    Pink is Vpoc

  • Hi Rohith,

    Thanks for the info. I am concerned about the slew rate on VPOC. Could you provide a zoomed in waveform of the initial drop on VPOC after LED ON, with slew rate measurement?

    By increasing the slew rate on the load switch, do you see an improvement?

  • Hi Luis,

    The slew rate is 900mV/420ms.

    Please find the image.

    By increasing the slew rate on the load switch, do you see an improvement?

        The slew rate of both IR voltage and Vpoc increased. But freeze issue is still there.

  • Hi Rohith,

    I am interested in the sharper drop, which occurs on the order of microseconds.

    Could you provide the slew rate of this voltage drop?

  • Hi Luis,

    Thanks for your support.

    We got solution for the freeze issue. The capacitor on Vpoc rail was increased to 2X 470uF. All the other components are revert to initially what it was. 

    Thanks 

    Rohith