Comment by gardarh | Hacker News Reader

Comment by gardarh | Hacker News Reader

Parent Full thread

I don't feel very confident about the article after reading the following: "Before sending the signal out via the HDMI output, the 1s and 0s are rearranged to minimize how many transitions there are. So instead of 10101010, the transmission may look like 11110000. If you really like math, how it does this is cool [wikipedia link to 8b/10b encoding] ..."

It's the other way around, the encoding makes sure there is a bit transition every now and then (details in wiki article); this is done in order to preserve the clock in the signal.

When someone writes an article on such a technical subject and messes up a technicality it makes me wonder whether there are more serious factual errors in there.

I agree with his point though, my $10 HDMI cable works perfectly :)

His description of how differential signaling is used is totally whacked too. His link to Wikipedia's "Phase Difference" page has "phase" in the title, but is irrelevant to how and why differential signaling is used ("puts the out-of-phase signal back in phase, then compares it to the "real" signal" Huh? Not.). He should have linked to "Differential Signaling" instead. http://en.wikipedia.org/wiki/Differential_signaling

The "control" characters are chosen to have a high number of transitions, but the "data" characters are chosen to have fewer transitions. https://secure.wikimedia.org/wikipedia/en/wiki/Transition_Mi...

DVI/HDMI uses it's own encoding (which together with CML-based physical layer is called TMDS - Transition Minimised Differential Signaling), that encode 8bit symbols as 10bit codewords, not the original IBM's 8B10B encoding. In contrast to most line codes, TMDS encoding is really designed to reduce number of bit transitions and thus output signal bandwidth (in the MHz sense). Clock recovery is different from most high speed serial interfaces, as HDMI/DVI provides separate signal with reference clock (which essentially eliminates problem with receiver clock drift that would otherwise necessitate inserting additional transitions into signal).

Reply on news.ycombinator.com