Post Go back to editing

AD8194

Hi everyone,

I've a PCB with an AD8194 and AD9880. The AD8194 mux is connected to a DVI and HDMI connector.

I've a trouble with the AD8194. The AD9880 doesn't detect signal with some DVI sources like personal computers.

If I remove the AD8194 and I do a bypass with some wires, the AD9880 detects the video source. Please, could anybody help me?

Thank you

Best regards

Jose

  • Hi Jose,

    Are you able to confirm if signal is present at the output of the AD8194? 

    Do you have any other indication what is specifically different about the DVI sources that are creating the issue?

    Dave

  • Hi daverowe,

    I've been testing with three kind of monitors:

    One of them has a DVI input (it has an AD9887), it works fine

    Other with DVI input and AD9880, it works fine

    Other with DVI and HDMI inputs (AD8194 and AD9880), The DVI input doesn't work fine with two PC

     

    The PCs have the same HW

     

    They have a main board with an Intel G33/31 express graphic board integrated (DVI, HDMI and VGA outputs) and an extra graphic board (Gigabyte 7G5818A Geoforce 8500 GT)

     

    With Intel I don't have any problem and all monitors work fine in HDMI and DVI

     

    With Gigabyte (7G5818A) I can't see anything on the monitor with AD8194, only sometimes the image appears for 1 or 2 seconds (with wrong colours) but quickly disappears. The rest of the monitors work fine

     

    If I remove the AD8194 I don't have problem

     

    Thank you very much

     

    Regards

     

    Jose

  • Hi Jose,

    Is there any difference if you change resolution or cable (short vs. long or thin vs. thick gauge)?

    You mentioned that AD9880 is not detecting a signal in the failing cases.  Can you probe the TMDS CLK signal before and after the AD8194 in both the failing and working cases? 

    Dave

  • Hi Dave,

    I have tested with some cables, I've tried with different length and with a DVI to HDMI cable too

    The result is the same, it doesn't work fine

    I'm going to probe the CLK before and after of AD8194. I'll tell you the news

    Thank you

    Jose

  • Hi Dave,

    I've probed the CLK signal before and after of AD8194

    Before: CLK- has a signal with 2.25v of DC and 350mv of AC over the DC

    CLK+ has a DC of 1.6v (more less)

    After: CLK- has a DC signal of 3v

    CLK+ has a DC signal of 2.5v

    The AD8194 doesn't have clock in its output

    If I do a bypass of AD8194 the AD9880 works fine with that signal

    The AD8194 has internal equalization, could it be the problem?

    The AD8193 doesn't have equalization and is pin to pin compatible with AD8194, could it be a solution?

    Thank you

    Regards

    Jose

  • Hi Jose,

    The dc common-mode voltage of the CLK + and - lanes should be the same and should be AVcc - 250mV  The minimum logic low level for DVI should be AVcc - 600mV.  It would appear as though the DVI sources are non-compliant or damaged.   What are the CLK signal levels when connected directly from DVI source to AD9880?   It is possible that the AD8194 might have been damaged by and ESD event.  Have you tried replacing the AD8194?

    I do not think your issue is related to the AD8194 equalizer.

    Best Regards,

    Dave

  • Hi Dave,

    I have replaced the AD8194 and the result is the same.

    In fact I've probed DVI with several tft monitors and the result is the same in all of them. I can't see image on the screen. I only see image if I remove the AD8194

    I think you're right, the sources of DVI are damaged

    I've measured the source of DVI in other PC with problems and the results are:

    CLK+ it's ok (The AVcc and Vswing are right)

    CLK- it's ok

    TMDS 0+ it's ok

    TMDS 0- it's ok

    TMDS 1+ it's ok

    TMDS 1- it's ok

    TMDS 2+ it's ok

    TMDS 2- it's wrong. I have a countinuous 3.3v more less

    I attach some images of the measures before and after of AD8194

    I don't understand why the AD9880 works fine if I remove the AD8194? It's a bit strange

    Thank you

    Regards

    Jose

  • Hi Jose,

    We do try very hard to make our solutions robust even when interfacing to non-compliant sources and sinks.  There's always a limit to how for out of spec a source can be and still work.  I am not a AD9880 expert, but it would certainly seem the AD9880 is more tolerant in this scenario.  The AD9880 is deserializing and processing the TMDS data.  The AD8194 is essentially an analog differential comparator.  So if one side of the differential pair is at the rail, there's very little gain and not surprising that bit errors are occuring. 

    It's common for this type of problem to be a mechanical (broken pin) problem with the connector.  If the graphics card is removable, perhaps you can check and if necessary replace the DVI connector.

    Dave

  • Hi Dave,

    Yes, you're right, the AD9880 must be more robust than 8194

    I'm going to probe the AD8193, it doesn't have equalizator and it's possible that its behaviour could be different

    I'll replace the DVI connector of the graphic board too

    Thank you very much for your support, it has been a great help

    Regards

    Jose