Hacker News Reader

Comment by mseebach | Hacker News Reader

I think "It is digital. It either works or it doesn't" in the context of A/V refers to the quality of the converted analog signal - what comes out of the speakers or is visible on the screen. And the signal getting through can either work or not, no matter how well engineered the cable is, isn't that correct?

View on HN

But the reason why your digital TV shows the correct screen is because the cable transmitted the "digital" bits between your DVD player and the TV correctly. If you have a poorly made cable you will get bit errors. Electromagnetic interference from other A/V equipment could also cause bit errors. A well made cable with twisted pairs, proper shielding (few people do this correctly), and well formed connector crimps will prevent those bit errors. With bit errors, what your awesome Blu-Ray player decodes off of the Blue-Ray disc will not be what is shown on the screen (assuming there isn't bit error correction in the TV, etc.).

And your speakers are playing an analog signal. Cable quality matters even more there.

> But the reason why your digital TV shows the correct screen is because the cable transmitted the "digital" bits between your DVD player and the TV correctly. If you have a poorly made cable you will get bit errors.

Yes. Which is why the poorly made cable can be said to not work.

> And your speakers are playing an analog signal.

Huh? I've never heard of HDMI cables capable of carrying analog sound.

Yup, the correct signal might not get through if you don't have perfect shielding. And a speck of dust on a CD could theoretically ruin a whole song.

That's why CDs (and cellphones, modems, and countless other digital devices) use channel encoding.[1] That way you don't have to have a perfectly noise-free signal to reconstruct the original information.

I don't know that many specifics of the DVD or Blu-Ray standards, but I'd put money on them using Reed-Solomon or something similar.

[1] http://en.wikipedia.org/wiki/Reed-Solomon_error_correction#D...

(Edit: forgot footnote)

I think that "it either work or doesn't" only if the receiver has some kind of cyclic redundancy check before processing the signal.

My guess is that very few digital A/V systems implement this. Proof of this is the weird sound you may hear on a damaged compact disc, it half works.

There probably are CRCs run during the decoding process, but CRCs only tell you "yes this is correct" or "no, something went wrong."

There are additional protection schemes that can detect and correct bit errors. Those are probably also run during the decoding process. The problem is that you can only detect so many wrong bits in a given word, and beyond that no correction can be made. That is probably why the damaged disc half works.

Yes it depends on the way you handle the response of the CRC, you could say "give me the data I don't care" or "please try to read it again it's wrong".

In my opinion that is just the proof that it's not as simple as "it just works or not" but rather : depending on muptiple factors such as error correction and the way your algorithm works on the receiver it may "half work" no matter what kind of cable you plug in.

Compact discs use something far better (in that context), a combination of two interleaved Reed–Solomon codes called CIRC, which is error-correcting (and not just detecting like a CRC). All digital A/V systems have to implement this in order to read a CD. It's made to be strong against burst errors which is why it can handle scratches on a CD pretty well. Samples can also be interpolated if something's really unreadable.

Even with an error-detecting code, there comes a point where there's too many errors to know if you received a valid code word or if the errors just "canceled each other". If 000 and 111 are your only valid words, it's still possible that a 111 gets turned into a 000 (3 consecutive errors) and there's no way to know about it...

Reply on news.ycombinator.com