Re: [Phys-l] Digital TV signal delay

[John Denker <jsd@av8n.com>]

On 09/22/2008 09:14 PM, Steve Highland wrote:
> Tonight one of the fellows on my bowling team asked me why the digital TV
> signal he receives is substantially delayed relative to the radio broadcast
> of the same live football game.

For _live events_ the signal is delayed (usually about 7 seconds)
to give censor time to react if there is a cussword or a "wardrobe
malfunction".

If the radio station is using a 5 second delay and the TV is 
using a 7 second delay, you would observe a difference of 2
seconds.

If the two delays are the same, you don't notice anything ... 
unless you take a portable radio to the actual stadium.

Still, I stand by my previous answer:  I would expect the radio
to be synchronized with plain old (*) analog TV, and the digital 
TV to be delayed by the compression algorithm.

  (*) Beware that nowadays some analog TV stations are just decoded
   versions of the digital station, so they have just as much 
   compression delay.

> He says he used to be able to watch live TV and listen to the radio together
> and they were nearly simultaneous. Now he says the digital TV is about two
> seconds behind the radio.

I've seen that.  It's amusing when the radio commentator announces
the result of the play before it is (according to the TV) over.

On 09/22/2008 09:30 PM, Michael Edmiston wrote:
> > I am under the impression that digital reception is buffered in the receiver 
> > so there is time to apply correction schemes to the data when the 
> > error-checking process detects a glitch in the received signal.


Sorry, no, that's not the right way to think about it.

Yes, there is some "buffering" going on in the receiver, but it
would be grossly unfair to blame the receiver for failing to 
decode a signal that has not yet been fully transmitted.  The
serious delay is in the encoder.

Decoding is incomparably easier than encoding.  Partly that's by
design, since there are thousands or millions of decoders for
each encoder ... but also it's analogous to getting out of a
parallel parking space:  getting out takes a lot less skill than
getting in.  All the creativity and heuristics are in the encoder;
the decoder just does what it's told.

And yes, there is some delay due to FEC (forward error correction)
but that is on the order of 10 milliseconds and cannot possibly
explain the observed two second delay.  And furthermore that is
only for OTA (over the air) signals;  cable signals do not need
(and do not bother with) much in the way of FEC.