We are using the TMDS442 in an application for A/B switching two Dual-Link DVI inputs (2560x1600 resolution). For testing, the output from the TMDS442 is connected via a short, high quality DVI cable to different computer monitors. No problems are observed on some monitors but on other monitors we observe a peculiar video artifact on switching from one active channel to the other that can best be described as pixel noise. The noise is not everywhere but localized in different specific areas of the video image.It is clear that the chip is trying to mitigate the situation by making some kind of internal "adjustment" which is sometimes successful in greatly reducing the noise or even eliminating it altogether. However, in other cases the mitigation attempt fails and the noise returns, sometimes to the extent that the entire video image starts blinking on and off . Sometimes using a longer DVI cable can help reduce the noise problem; other times not. The data sheet is vague on the issue of output loading and what happens under different situations. In all of the circuits I am familiar with that employ the presumably similar TMDS341, the output is connected directly to the LCD panel via a TMDS cable while in our case the output drives a DVI cable which connects to another computer monitor. We are confident the problem is not with the PCB layout which employs impedance and trace-length controlled lines on both input and output but rather with the nature of the interconnection. My question is has anybody seen and/or is familiar with the noise I try to describe and what might be the root cause.
Steve Sitko
