Post Go back to editing

ADV7842 Data Enable signal instability with 1280x1024 DVI signal

Category: Hardware
Product Number: ADV7842

We have encountered a particular DVI signal that causes the DE (DataEnable) signal from the ADV7842 to be erratic. We have not seen this error on any other DVI or HDMI signal we have encountered, at any resolution. The captured image moves up and down and some scan lines will shift horizontally. The signal is 1280x1024, HF=63982Hz, VF=60.02Hz, Pixels Per Line=1688, Lines Per Frame=1066, PF=108.001616MHz. The signal looks like a standard DMT1260G format from a Quantum generator with a minor adjustment to the horizontal frequency. I have tried other video capture products with no problems, including one of our own that uses the ADV7441. I cannot reproduce the problem with any video generator, even by changing timing parameters. On the problem signal I have tried different cables, buffering with splitters and transmitting with an optical HDMI cable, with no change to the errors, so I don't think it's an equalization or TMDS signal integrity problem. If I put a 1:1 video scaler device between the source and our input, the problem goes away. So, receiving and re-transmitting does not pass the error through. The signal comes from a medical imaging system in a hospital so access to the signal for debugging has been difficult. There's something about the signal the 7842 doesn't like. The DE regeneration and vertical filters are locked according to the status registers. The odd thing is it seems to only affect the DE output, the vertical and horizontal sync outputs look stable.

I have attached scope pictures of the DE output (blue) with respect to VS (yellow) as the trigger. The first one shows where sometimes there is no DE after vertical sync (faint blue trace). The second one shows where sometimes DE occurs inside vertical sync (faint blue trace). The scope inputs come directly from the 7842 outputs so no contribution from any other on-board logic.

Any suggestions as to what could be causing this and if there is any possible solution, would be greatly appreciated. Since there is no documentation showing any of the HDMI PLL etc. registers I was not able to experiment with any settings.

Thanks.

  • Hi,

    My first suggestion is to try capture the overlap image using single trigger only, not auto.  Maybe have the scope trigger on the rising edge of VS and scale it so you can see one full frame on the scope,  See if the overlap still exist.

    I the only quick reason I can think of DE overlapping VS is the incoming HDMI data encoding is bad.  Very unlikely.

  • Hi Guenter,

    Thanks for the fast response.  If I connect an error free signal, with the same scope setup, the traces are clean.  I don't think the auto trigger is the reason for the faint traces, I think they are displayed that way because of the display persistence and the lower frequency of the errors. The trigger is VS at 60Hz so the auto trigger shouldn't fire.  Anyway, I think the pictures are a good view of good vs bad.

    You're right about the encoding, I believe it is correct since our capture card with the 7842 seems to be the only device that shows the problem.  The power supplies look clean, and I would expect if there was a problem there, I'd see it with a generator set to the same format.  We have run many different signals and formats into the card with no problems. I was hoping there might be a magic undocumented register setting for a frequency range like I've seen for analog mode.  This is a strange one.

    Thanks.

  • As long as you configure the ADV7842 using our scripts for the incoming format it should all be good.  I've never seen a problem with DE overlapping VS.  The stock response now would be "Can you reproduce the issue on our eval. board".  I don't like this answer but I can't think of any magic register to solve this.

    I'd still like to see one complete frame capture with the problem.  The only other thing is run your source into and HDMI analyzer to see if any errors are reported.

    Yes, this is strange. 

  • We do use your scripts, and until now we've been able to reliably capture every signal we've connected to.  It will be a while before I can try those suggestions since site access is difficult.  Thanks for your help.