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.

TLK10232: Communication error

Part Number: TLK10232

Hi team

Could you let me know your advice for following situation?

Customer's condition
1. TLK10232CTR is being used for 10 GB Ethernet to XAUI, but there is problem that seems to be caused by this IC.
2. In the communication IF using TLK1023, a 10GB Ethernet communication error occurs.
3. When send 5,000 to 6,000 PING commands, it detects about 5 to 50 communication errors.
4. An error occurs only Ach. No error is detected in Bch.
5. Error frequency is high at low temperature start (error about 50 times / 6,000 Ping), error frequency is low at high temperature start (about 5 errors / 6,000 Ping).
6. Even if the temperature changes after startup, there is no change in subsequent error frequency. The error frequency depends on the PHY temperature at startup.
7. Confirmed communication error with 2 out of 8 PHYs. The remaining 6 are normal.
8. Since the error frequency is caused by the PHY temperature at startup, it is specified that there is a factor in the PHY.
9. Confirmed that the XAUI communication waveform, power supply waveform, clock waveform (jitter) was not problem.

Customer's question
1. Is there any trouble similar to the above in the product? Does TLK10233 have any issue?
2. The error frequency depends on the temperature at startup. Does the IC calibrate at startup?
3. Please let me know if there are any design factors.
4. If you need schematic and waveform, I will send offline.

Best Regards,
Takanori Yuzawa

  • Takanori-san,

    Thank you for all the information. Has the customer tried to tune HS_SERDES_CONTROL_x during this test to eliminate error? Is this a optical application? Just to confirm you have tried a "swap" test, where you replace a functioning part in a functioning system with a failing part part to confirm that the problem "follows" the device? Please see answers/questions to your questions below.

    1. Is there any trouble similar to the above in the product? Does TLK10233 have any issue?

    • There is no know issue with TLK10232 across temperature. What temperatures are the errors seen at?

    2. The error frequency depends on the temperature at startup. Does the IC calibrate at startup?

    • At startup the IC will have internal PLLs lock on to input reference clocks. There is also a calibration done to compensate for input voltage offset at the receiver after the internal PLLs have locked and the receiver is enabled. The customer can disable this calibration via the HS_AZCAL register to see if this is what is causing issue?
  • Hi Malik-san

    Please see the customer's feedback below and let me know your advice.

    Has the customer tried to tune HS_SERDES_CONTROL_x during this test to eliminate error?
    => No. They use default value.

    Is this a optical application?
    => Yes

    Just to confirm you have tried a "swap" test, where you replace a functioning part in a functioning system with a failing part part to confirm that the problem "follows" the device?
    => No. It's because they have only a few devices and it might have risk to replace.

    What temperatures are the errors seen at?
    => They used point cooler for low temp test and heat gun for high temp test.

    The customer can disable this calibration via the HS_AZCAL register to see if this is what is causing issue?
    => They will try it.

    The register value to change from default below.

    #1
    PHY Add:0x00
    Dev Add:0x1E
    Reg Add:0x0000
    Write Data:0x8610

    #2
    PHY Add:0x00
    Dev Add:0x1E
    Reg Add:0x0000
    Write Data:0x0E10

    #3
    PHY Add:0x00
    Dev Add:0x07
    Reg Add:0x0000
    Write Data:0x2000

    #4
    PHY Add:0x00
    Dev Add:0x01
    Reg Add:0x0096
    Write Data:0x0000

    #5
    PHY Add:0x00
    Dev Add:0x1E
    Reg Add:0x0004
    Write Data:0xD500

    #6
    PHY Add:0x00
    Dev Add:0x1E
    Reg Add:0x000E
    Write Data:0x0008

    The result after confirming the waveform

    1. The timing for MDIO and MDC are not violated on the datasheet.
    2. The quality of waveform for XAUI is no problem. Input CLK is stable when RST on PHY de-assert.
    3. The each supply voltage is stable.

    Best Regards,
    Takanori Yuzawa

  • Hi Malik-san

    The customer is considering that the error is related to incorrect AGC(Auto Gain Controller) behavior because of the customer's evaluation.
    What is the reason for incorrect AGC behavior? And also, does AGC restart by DATAPATH_RESET setting?

    Please see the customer's evaluation result below.

    #1: PHY Add:0x00 Dev Add:0x1E Reg Add:0x0009 Write Data:0x0388
    Detected error when set HS_AGCLOCK_OVERLAY and monitor LOSA and then the PHY is cooled with a point cooler.
    If it keeps to cool then LOSA keeps high. If not keeps to cool then LOSA keeps low.
    It does not change the level when the value of HS_AGCCTRL(HS_SERDES_CONTROL_2) change.
    Only if it set Force the attenuator off, LOSA keeps high.

    #2: PHY Add:0x00 Dev Add:0x1E Reg Add:0x0009 Write Data:0x0390
    When set HS_INVALID_CODE_OVERLAY and monitor LOSA then the error often occur.
    When cooled PHY and set DATAPATH_RESET then the error increased.
    When warmed PHY and set DATAPATH_RESET then the error decreased.

    #3: PHY Add:0x00 Dev Add:0x1E Reg Add:0x0003 Write Data:0xA848 or 0xA808 or 0xA888 or 0xA8C8
    When set all HS_AZCAL then the behavior did not change.

    Best Regards,
    Takanori Yuzawa

  • Takanori-san,

    I am continuing to investigate the issues here, does the customer know the specific temperatures they are cooling and heating the device to while testing? This information is important to understanding what may be happening. Also is the customer saying that the AGC never locks (LOSx = low) during testing  or becomes unlocked when cooled regardless of how long it has been running or is this test only conducted at startup? Customer could also monitor HS_AGC_LOCKED bit to determine when the AGC becomes unlocked during their test. 

    Is HS_AGCCTRL[1:0] = 00 is device setup? Does the customer still see issues if they force the attenuator on (HS_AGCCTRL[1:0] = 11)?

  • Thread taken offline, closing. Marking as "TI Thinks Resolved".