• State Resolved
  • Locked Locked
  • Replies 10 replies
  • Subscribers 32 subscribers
  • Views 300 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

Original question:

DS90UB635-Q1: DS90UB635-Q1 DS90UB638-Q1

DS90UB635-Q1: Trying to bring up 635/638 pair. Cannot establish backchannel

Barry W
Intellectual 850 points
Part Number: DS90UB635-Q1
Other Parts Discussed in Thread: DS90UB953-Q1EVM, DS90UB954-Q1EVM

It seems to me the first step of getting the FPD-Link III devices running is to establish the backchannel.  I cannot seem to get that far.

I have two custom prototype boards, one with the DS90UB635, one with the DS90UB638.  They have provisions for power over coax, but I have removed the first filter component on each side to eliminate the possiblity of power noise corruption.  The 635 is powered separately.

I do not currently have a CSI sensor attached to the 635 device.  My plan is to verify functionality using the BIST feature.  I can read and write I2C commands to each device using the I2C pins.

When I connect them together, the 635 device does not produce a CLK_OUT and the LINK_DET bit in register 0x52 = 0;  Register 0x03 reports 0x48 indicating it is in CSI-2 Synchronous Mode.

When I connect the 638 coax link directly to a low speed scope I see a reasonable encoded clock signal running at 50Mhz.

When I connect the scope (with a coax splitter) to the cables between the 635 and 638 I see these odd pulses every ~850µS.  It sure seems like they are generated by the 635.  I do not see any corresponding glitch on the 1.8V supply rail.

Does any of that provide any clues to why the back channel will not connect?

Are there other debugging steps I can take to find this backchannel issue?

  • 0
    Intellectual 850 points

    Now I see that if I continuously read the 635 register 0x52, 10-20% of the time the LINK_DET bit will report 1.

    I still never see any clock on the CLK_OUT/IDX pin.

  • 0 in reply to Barry W
    Intellectual 850 points

    Now I see that LINK_DET still reports 1 10-20% of the time even with the coax cable disconnected.

  • 0 in reply to Barry W
    Intellectual 850 points

    Here is a list of register values read from the 635 and 638 when both devices are powered and connected via coax.

    serdes6.txt 
    DS90UB635 Register Read Results
01.00 = 0x30
01.01 = 0x00
01.02 = 0x33
01.03 = 0x48
01.04 = 0x00
01.05 = 0x03
01.06 = 0x41
01.07 = 0x28
01.08 = 0xFE
01.09 = 0x1E
01.0A = 0x10
01.0B = 0x7F
01.0C = 0x7F
01.0D = 0xF0
01.0E = 0x0F
01.0F = 0x00
01.10 = 0x00
01.11 = 0x00
01.12 = 0x00
01.13 = 0x00
01.14 = 0x00
01.15 = 0x20
01.16 = 0x18
01.17 = 0x3C
01.18 = 0x80
01.19 = 0x62
01.1A = 0x62
01.1B = 0x62
01.1C = 0x00
01.1D = 0x00
01.1E = 0x00
01.1F = 0x00
01.20 = 0x00
01.21 = 0x00
01.22 = 0x00
01.23 = 0x00
01.24 = 0x00
01.25 = 0x02
01.26 = 0x00
01.27 = 0x00
01.28 = 0x67
01.29 = 0x33
01.2A = 0x01
01.2B = 0x00
01.2C = 0x00
01.2D = 0x00
01.2E = 0x00
01.2F = 0x00
01.30 = 0x00
01.31 = 0x20
01.32 = 0x09
01.33 = 0x04
01.34 = 0x00
01.35 = 0x00
01.36 = 0x00
01.37 = 0x7A
01.38 = 0x00
01.39 = 0x00
01.3A = 0x00
01.3B = 0x00
01.3C = 0x00
01.3D = 0x00
01.3E = 0x00
01.3F = 0x00
01.40 = 0x00
01.41 = 0x00
01.42 = 0x00
01.43 = 0x00
01.44 = 0x00
01.45 = 0x00
01.46 = 0x00
01.47 = 0x00
01.48 = 0x00
01.49 = 0x00
01.4A = 0x00
01.4B = 0x00
01.4C = 0x00
01.4D = 0x00
01.4E = 0x00
01.4F = 0x00
01.50 = 0x20
01.51 = 0xC0
01.52 = 0x20
01.53 = 0x01
01.54 = 0x00
01.55 = 0x00
01.56 = 0x00
01.57 = 0x00
01.58 = 0x07
01.59 = 0x07
01.5A = 0x07
01.5B = 0x00
01.5C = 0x00
01.5D = 0x00
01.5E = 0x00
01.5F = 0x00
01.60 = 0x00
01.61 = 0x00
01.62 = 0x00
01.63 = 0x00
01.64 = 0x00
01.B0 = 0x04
01.B1 = 0x4A
01.B2 = 0x3F
01.F0 = 0x5F
01.F1 = 0x55
01.F2 = 0x42
01.F3 = 0x39
01.F4 = 0x35
01.F5 = 0x33
DS90UB638 Register Read Results
00.00 = 0x60
00.01 = 0x00
00.02 = 0x1E
00.03 = 0x20
00.04 = 0xD3
00.05 = 0x01
00.06 = 0x00
00.07 = 0xFE
00.08 = 0x1C
00.09 = 0x10
00.0A = 0x7A
00.0B = 0x7A
00.0C = 0x83
00.0D = 0xB9
00.0E = 0x08
00.0F = 0x7F
00.10 = 0x00
00.11 = 0x00
00.12 = 0x00
00.13 = 0x00
00.14 = 0x00
00.15 = 0x00
00.16 = 0x00
00.17 = 0x00
00.18 = 0x00
00.19 = 0x00
00.1A = 0x00
00.1B = 0x00
00.1C = 0x00
00.1D = 0x00
00.1E = 0x04
00.1F = 0x02
00.20 = 0x30
00.21 = 0x01
00.22 = 0x00
00.23 = 0x00
00.24 = 0x00
00.25 = 0x00
00.26 = 0x00
00.27 = 0x00
00.28 = 0x00
00.29 = 0x00
00.2A = 0x00
00.2B = 0x00
00.2C = 0x00
00.2D = 0x00
00.2E = 0x00
00.2F = 0x00
00.30 = 0x00
00.31 = 0x00
00.32 = 0x00
00.33 = 0x00
00.34 = 0x40
00.35 = 0x00
00.36 = 0x00
00.37 = 0x00
00.38 = 0x00
00.39 = 0x00
00.3A = 0x00
00.3B = 0x01
00.3C = 0x14
00.3D = 0x6F
00.3E = 0x00
00.3F = 0x40
00.40 = 0x00
00.41 = 0xA7
00.42 = 0x71
00.43 = 0x01
00.44 = 0x00
00.45 = 0x00
00.46 = 0x00
00.47 = 0x00
00.48 = 0x00
00.49 = 0x00
00.4A = 0x00
00.4B = 0x12
00.4C = 0x00
00.4D = 0x00
00.4E = 0x02
00.4F = 0x00
00.50 = 0x00
00.51 = 0x00
00.52 = 0x00
00.53 = 0x00
00.54 = 0x00
00.55 = 0x00
00.56 = 0x00
00.57 = 0x00
00.58 = 0x1E
00.59 = 0x00
00.5A = 0x00
00.5B = 0x00
00.5C = 0x00
00.5D = 0x00
00.5E = 0x00
00.5F = 0x00
00.60 = 0x00
00.61 = 0x00
00.62 = 0x00
00.63 = 0x00
00.64 = 0x00
00.65 = 0x00
00.66 = 0x00
00.67 = 0x00
00.68 = 0x00
00.69 = 0x00
00.6A = 0x00
00.6B = 0x00
00.6C = 0x00
00.6D = 0x7C
00.6E = 0x88
00.6F = 0x88
00.70 = 0x2B
00.71 = 0x2C
00.72 = 0xE4
00.73 = 0x00
00.74 = 0x00
00.75 = 0x00
00.76 = 0x00
00.77 = 0xC5
00.78 = 0x00
00.79 = 0x01
00.7A = 0x00
00.7B = 0x00
00.7C = 0x20
00.7D = 0x00
00.7E = 0x00
00.7F = 0x00
00.B0 = 0x08
00.B1 = 0x14
00.B2 = 0x3F
00.B3 = 0x08
00.B4 = 0x25
00.B5 = 0x00
00.B6 = 0x18
00.B7 = 0x00
00.B8 = 0x8C
00.B9 = 0x33
00.BA = 0x83
00.BB = 0x74
00.BC = 0x80
00.BD = 0x00
00.BE = 0x00
00.BF = 0x00
00.C0 = 0x00
00.C1 = 0x00
00.C2 = 0x00
00.C3 = 0x00
00.C4 = 0x00
00.C5 = 0x00
00.C6 = 0x00
00.C7 = 0x00
00.C8 = 0x00
00.C9 = 0x00
00.CA = 0x00
00.CB = 0x00
00.CC = 0x00
00.CD = 0x00
00.CE = 0x00
00.CF = 0x00
00.D0 = 0x00
00.D2 = 0x94
00.D3 = 0x17
00.D4 = 0x60
00.D5 = 0xF2
00.F0 = 0x5F
00.F1 = 0x55
00.F2 = 0x42
00.F3 = 0x39
00.F4 = 0x35
00.F5 = 0x34
00.F8 = 0x00

  • 0 in reply to Barry W
    Intellectual 850 points

    I'm reviewing the DS90UB954-Q1EVM, DS90UB953-Q1EVM schematics and comparing them to our prototype PCBs.  I have not found any issues.  Most of the selectable options on the EVMs are now more static on our boards.

    Can you take the DS90UB954-Q1EVM and just replace the deserializer with the DS90UB938?  Are other changes required?

    Can you take the DS90UB953-Q1EVM and just replace the serializer with the DS90UB635?  Are other changes required?

    Will your software tools account for these changes in the test setups?

  • 0 in reply to Barry W
    TI__Genius 17715 points

    Hi Barry, 

    The DS90UB635 and DS90UB953 are pin to pin compatible.

    The DS90UB638 and DS90UB954 differ as the DS90UB638 is a one channel deserializer whereas the DS90UB954 is dual channel. Software will be compatible between the two devices excluding port specific registers. At startup, we recommend writing register 0x0C=0x81 for proper operation. Currently, this is set to 0x83. 

    Would you be able to share the schematics for these custom boards? Have different cables been tested between the 638 and 635?

    Best,

    Zoe

  • 0 in reply to Zoe Bison
    Intellectual 850 points

    Hello, Zoe,

    Thank you so much for looking at my problem.

    I am very embarrassed that I did not notice either of the notes in the 638 data sheet requiring that 0x0C be set to 0x81.  I cleared bit 1 (and also set bit 6 in 0x21 for good measure).  Unfortunately it did not fix my issue.  The 638 still indicates not locked and the 635 still mostly reports LINK_DET false.

    I have only tested 2 cables so far a 3' low loss cable and a 10' RG-178 cable.  We use SMA connectors for our boards.  I also tested both cables in series with a splitter for a scope input.  The connection status is always the same.

    635 Schematic:

    638 Schematic

    I think the only thing not shown here is the PowerEn18 signal source.  It comes from the power good signal of the 1.8V regulator.  I measured it at 1.5V.

  • +1 in reply to Barry W
    TI__Genius 17715 points

    Hi Barry,

    Thanks for the update. Is there a reason the 2kOhm termination resistor was selected? We recommend a value of 50Ohms for the termination resistor. 

    Best,

    Zoe

  • 0 in reply to Zoe Bison
    Intellectual 850 points

    Thanks for the thorough review. That resistor sounds like a mistake. We will replace both of them and retest first thing tomorrow. 

  • 0 in reply to Barry W
    Intellectual 850 points

    Hello, Zoe,

    Replacing the termination resistors with the correct value worked!  Imagine that, impedance mismatches are bad.  I looked at that 100 times and even changed the termination capacitor values as a test, but never noticed the resistor value.

    I made a much longer test cable and it is still connecting reliably.  This is a very important project for our small company and seeing these boards connect is a major milestone for us.

    Sorry to drag you into debugging my silly mistake.  I can only say how much I appreciate your help.  My goal as an engineer/person is to come to work and try to help people with their problems.  If your goals are similar you have so succeeded today.

    Please have a great weekend.  Don't hesitate to ask if we can return the favor.

  • 0 in reply to Barry W
    TI__Genius 17715 points

    Hi Barry, 

    Glad the system is up and running on your end! Please feel free to let us know if there are any remaining questions, happy to help. 

    Best,

    Zoe