Hello,
We're using the DS92LV2421 serializer, DS25BR100 repeater buffer, DS25BR110 receive buffer, and DS92LV0422 deserializer. The components are located on three different boards: the 2421 is located on a graphical processor board, the BR100 is located on a backplane board (the processor board plugs into the backplane board with 2mm hard metric connector set), the BR110 and 0422 are located on a display interface board, and the serial data stream between the BR100 and BR110 pass through a 3 foot custom cable with 100 Ohm STP. The serial data stream is AC-coupled with 0.1uf coupling caps, per the data sheets for the serializer and deserializer. All the power pins of the serializer and the deserializer are filtered with ferrite beads. Our pixel clock is at 66.4MHz. The 24-bit RGB data, control signals and clock being presented to the 2421 are not violating the 2421's input requirements. The issue we're having is intermittent flickering of the display. The flickering comes for a short period of time and then dissipates and the display operates just fine for about an hours and then the flickering comes back for a couple of minutes and then goes away again. We can count about 300 loss of lock during the flickering time. I can provide a schematic of the link if desired. I have tried replacing the coupling caps with 0.01uf and that helped, but didn't solve the problem. Please help.
Thanks,
Walid
Hi Walid,
You may have already tried these but please let us know responses to the questions below. Hopefully these will help us to pin point where these:
1). Would you observe the same behavior if you use short cable ?
2). Do you happen to have eye diagram at different points?
3). Would you see this problem if you reduce the data rate or change data pattern?
4). If you use DS92LV0422 evaluation board do you observe loss of lock intermittently as well?
5). Is it possible to measure clock in and clock out of the serializer and de-serializer?
As you can see, these questions help us to localize the problem. Also, it would help to attach the relevant schematics as well.
Regards,,nasser
Hi Nasser,
Thanks for getting back so quickly. The schematic is attached. As for the questions, here are the replies:
1.) The three foot cable is the shortest cable I can use. My alternate cable is over three meters long. With the longer cable the loss of lock happens about 500 times in a matter of two minutes.
2.) I don't have an eye diagram that is of any quantitative value. I have very limited test equipment access right now.
3.) I have no control over the data rate. The data pattern is different colored and sized sqaures that pop up on top of previous displayed squares at random locations in the display area.
4.) I haven't tried to use any eval boards.
5.) The clocks both measure 66.4MHz.
I should add, the loss of lock happens only about three minutes after power up and lasts about two minutes. Then the video is fine and there are no more issues. Hope this helps.
I work very closely with Mike Wolfe and he mentioned he is working on this as well.
The last two sentences that you mentioned could be the key to solving this issue. Based on my understanding loss of lock happens about 3 minutes after the initial power up and then afterward, i.e in steady state mode, the video runs error free for ever. If indeed this is the case, please do this, once the video is running error free in steady state mode power off the receiver, wait for 7 seconds or so, and then power on the receiver. Does the video comes up without the flicker after 2-3 minutes or there is still flicker? We should do a similar procedure for transmiter and the buffer in order to identify the unti that causes loss of lock.
I currently don't have the software capability to power down the deserializer while the unit is powered up to test what you suggest. The reason the display runs flicker-free indefinitely after the short flicker 2-3 minutes after power up is that I replaced the 0.1uf coupling caps with 0.01uf caps everywhere except between the BR110 and the 0422. What I have tested, however, is to place the 2421 and 0422 in BIST mode and the PASS pin on the 0422 stays high and the video comes back flicker free after I deassert BISTEN on both chips.
When I had the 0.1uf coupling caps in all the AC coupled sections of the data stream, the video would flicker intermittently every now and then. But, I could make the flickering go away when I touched or probed the AC coupled serial data stream. Does this sound to you like a charge build-up that would push the common mode voltage to either extreme of the input voltage range of the 0422, or decrease the noise margin of the system to the breaking point?
First of all, by changing the AC coupling Cap value from0.1uF to 0.01uF you are increasing the high frequency constant of the video pattern. You can achieve the same result by enabling the de-emphasis on DS25BR100. If you haven't done this, you can try this.
Secondly, you had mentioned once you power up the device video goes through for 3 minutes or so, then for two minutes you notice loss of lock, then finally afterward everything works fine and video goes through with no error. As far as we know, there is nothing internal to the part that could have such a long time(i.e in order of minutes). My suspicion is that this could be a mechanical issue since this sort of mechanical behavior could have such a long time constant.