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.

DP83867CR: DP83867CRRGZT temperature testing issue with full load

Part Number: DP83867CR

Hi TI team,

We are using DP83867CRRGZT consumer grade (0DC to +70DC) part for our project intended for residential application (as emergency calling device over Ethernet).

We have conducted temperature testing (from 0DC to +55DC) on open setup (without enclosure) with full load (Ethernet, DMIC, speaker, LEDs, etc.

The board works fine from +1DC to +55DC with full load. Between 0DC to +1DC, Eth link fails (8 out of 10 times) and we lose access of the board. I have attached setup images for your reference.

Ethernet PHY is on the bottom side (opposite side of what we can see in image) of white colored board.

We just kept board in a thick plastic box and closed the box. Power and Ethernet cables are taken out from hole of plastic box as seen in the image. Please suggest what could be the issue between 0DC to +1DC when Ethernet is not working most of the time.

Let me know if you need any other details.

Regards,

Nija Mankodi

+91 7405533797

  • Hello Nija,

    1. Do you power down and power up the board before checking for link at each temperature? Or you just establish the link at room temperature and then sweep the temperature to see if the link drops? (Trying to understand the sequence tested when you see the link drop)

    2. Do you have access to PHY registers? If so can you check if the link drop recovers after writting register 0x001F = 0x8000 to the PHY (software reset register).

    --

    Regards,

    Vikram

  • Hi Vikram,

    Sorry for being late on this.

    1) For your query# 1, we you establish the link at room temperature and then sweep the temperature (room temperature to +55DC to room temperature to 0DC and then again to room temperature. In a nutshell, we do temperature sweep in our product's operating temperature range.

    2) We will check your poinr#2 and revert ASAP.

    Again, along with open-board, we have also tested our board with enclosure but we are getting same issue. Attached enclosed device images for your reference.

  • Hi Vikram,

    Any update on this?

  • Hello Nija,

    PHY have been tested til -40C so I dont expect it to be PHY issue as such.

    - Is it related to some water freeze issue in your setup?

    -There must be few components on your board which are there to support the PHY. Can you please check that those components have sufficient temperature range support : example crystal or oscillator, LDOs?

    -How much time does it take in your setup to go from room to 0C? Though as you are using chamber, so I assume that it will be in a few minutes.

    - Lets try that reset using register write (if possible) to see if restarting a part of PHY recovers your issue.

    --

    Regards,

    Vikram

  • Hi Vikram,

    We are using DP83867CRRGZT PHY part which is of consumer grade only and supports 0DC to +70DC. I do not think, this particular part would have been testing till -40DC. Please confirm my understanding.

  • Hi Vikram,

    I would like to share an update on below point which you suggested to try.

    --Do you have access to PHY registers? If so can you check if the link drop recovers after writing register 0x001F = 0x8000 to the PHY (software reset register)

    When device loses Ethernet connectivity, we tried to access device using USB and read register 0x001F as 0x0000. We modified its value to 0x8000 and then ethernet link got re-established and ethernet started working. However, in this working condition when we again tried to read this register, it was 0x0000 and not 0x8000 which is strange. Ethernet link was re-established but register value did not change to 0x8000.

    Please share you valuable comments on this and above understanding.

  • Hello Nija,

    Bit 15 of 0x001F is self-clearing bit. It resets to 0 automatically after reseting the PHY. Is resetting the PHY during link-loss a workable solution for your system?

    --

    Regards,

    Vikram

  • Hi Vikram,

    Our product will be ON 24*7 as it is emergency device. Can you please guide on how can we implement this solution after deployment ? This will help us to identify whether it is workable or not.

    Also, do you see any other option to resolve this issue if resetting the PHY is not feasible?

  • Hi Vikram,

    Please provide update on this.

  • Hi  and TI Team,

    I had conducted the thermal tests on these units with Nija. And as mentioned previously that ethernet connection is getting lost when tested in temperature range from 0DegC to +1DegC

    Ethernet connection was lost twice when I tested in the temperature range, 0DegC to 1DegC continuously for duration of 2 hours but in the rest of this duration data packet was continuously transferred at rate of 900-980Mbps (approx. 1Gbps).

    As you already know ours is the Emergency calling device and this will be installed where the ambient will be low as 1DegC so pls suggest solution to over come this issue.

    Right now we already have assembled units with TI PHY part# DP83867CRRGZ so how can we work around on these assembled units so that ethernet PHY works even in 0DegC to 1DegC temp. range.

    By resetting the PHY with this s/w register 0X100F is the only solution or if any alternate method that we can implement to work on these assembled units.

    Waiting for your prompt support.

    Thanks,

    Dhara

  • Hi  and TI Team,

    We can try to work with resetting the PHY during link-loss scenario. But for that we will need to know when to reset the bit in this software register.

    Can you suggest us what register we can check to see if the PHY link is lost or not?

    Is there a register bit which can tell us that the PHY connection is lost due to the temperature issue?
    Like wise we can implement to monitor that register and then keep checking it in the SW driver for DP83867 and in case of connection loss we will reset the PHY in the driver.
    But pls advice on which register to monitor to identify link loss event.
    Awaiting your prompt support to resolve this issue ASAP.
    Thanks,
    Dhara
  • Hi Vikram,

    Can we please have an update on this?

    This is highly critical for us now to identify the root-cause of this issue and resolve the same with workable solution.

  • Hello Dhara,

    Bit reg<0x0001>[2] is the link-status : 0 = invalid link; 1 = valid link.

    If because of any out of boundary condition at pins of PHY, PHY goes into bad state then it can be recovered by applying reset to the PHY.

    1. Pin reset : By toggling reset_n pin low.

    2. Software reset : register<0x001F> = x8000

    --

    Regards,

    Vikram

  • Hi ,

    I am the SW engineer supporting Nija and Dhara with this issue. Please help to answer below queries:

    1. Will the register bit <0x0001>[2] be in set condition when a Ethernet cable is not connected? If yes, then in case we implement a logic to read this bit continuously and we have not connected the cable, the driver will keep resetting the PHY. This may have unnecessary (maybe even detrimental) effect on the system.

    Does DP83867 have a register status bit that suggest connection lost only due to temperature issue? So that we reset the PHY only when we face this corner case of temperature issue.

    2. Or do you have a patch or code sample which implements this kind of logic for checking the link status bit? We want to make sure that if we implement such logic then we do not cause any issue in other system of PHY which are perfectly working.

    We have further questions regarding this implementation. Can we have a joint call to discuss them?

    Regards,
    Naman Thaker

  • Hello Dhara and Naman,

    I am closing the thread for now as we have discussed the next steps over the call. You may reach us again through FAE for any further help.

    --

    Regards,

    Vikram