How does digital TV broadcast prevent ghosting effects?

[MRW]

Good morning world!

I've been reading about the new digital TV standards (8-VSB, DVB-T,
ISDB-T) and they all mention that the standards are more robust
compared to the analog standard. For example, all state that
practically ghosting effects are eliminated with DVB-T and ISDB-T
providing more robust response to mulitpath effects than 8-VSB.

How do these standards eliminate ghosting effects?

I also read a similar scheme for analog TV using Ghost Canceling
Reference (GCR) signals from both transmitter and receiver. Is this
pretty much the same idea?

Thanks!

 

[Joel Koltner]

How do these standards eliminate ghosting effects?

The basic idea is that, since they're transmitting a digital signal, if the
reflections (ghosts) aren't strong enough to start "flipping bits" (or only
flip a few), the resultant signal is still "digitally perfect" (at least once
error correction is performed) and thus the picture displayed is exactly what
the transmitter started with.

An improvement can be had with a so-called "rake receiver" (see:
http://en.wikipedia.org/wiki/Rake_receiver): Assuming the ghosting is
relatively constant, you're just getting delayed copies of the original signal
at various points in time. If you start sampling at those various points in
time and summing up the result, you can faitfully reconstruct the original
signal. (Of course, finding the correct "various points" is not so
trivial...) Hence you're you're "raking in" all the copies of the original
signals to build up the result.

Supposedly first-generation ATSC receiver chipsets didn't do any of this
active ghost cancellation, whereas second- and (the current) third-generation
chips do.

I also read a similar scheme for analog TV using Ghost Canceling
Reference (GCR) signals from both transmitter and receiver. Is this
pretty much the same idea?

Somewhat, yes, although with a rake receiver you don't (necessarily) need a
reference signal -- you just try out the various sample points until you get
one that produces the "best" result. Unlike an analog system (where all you
really have to measure are signal to noise ratios) in a digital system usually
there are plenty of synchronization and test data patterns at known locations
within the signal, so it's usually easy to determine how well the system's
performing.

---Joel

 

[MRW]

The basic idea is that, since they're transmitting a digital signal, if the
reflections (ghosts) aren't strong enough to start "flipping bits" (or only
flip a few), the resultant signal is still "digitally perfect" (at least once
error correction is performed) and thus the picture displayed is exactly what
the transmitter started with.

An improvement can be had with a so-called "rake receiver" (see:http://en.wikipedia.org/wiki/Rake_receiver):Assuming the ghosting is
relatively constant, you're just getting delayed copies of the original signal
at various points in time. If you start sampling at those various points in
time and summing up the result, you can faitfully reconstruct the original
signal. (Of course, finding the correct "various points" is not so
trivial...) Hence you're you're "raking in" all the copies of the original
signals to build up the result.

Supposedly first-generation ATSC receiver chipsets didn't do any of this
active ghost cancellation, whereas second- and (the current) third-generation
chips do.


Somewhat, yes, although with a rake receiver you don't (necessarily) need a
reference signal -- you just try out the various sample points until you get
one that produces the "best" result. Unlike an analog system (where all you
really have to measure are signal to noise ratios) in a digital system usually
there are plenty of synchronization and test data patterns at known locations
within the signal, so it's usually easy to determine how well the system's
performing.

---Joel

Hi Joel, Thanks for the response.

I'm assuming that the error correction portion is dependent on the
modulation scheme. So, am I right in assuming that the COFDM technique
being describe as more robust to multipath effects than 8-VSB plays a
part in implementing the error correction scheme?

You know, it's a bit tricky for me to patch up all the stuff I've read
from various references. My main surprise is the fact that DVB-T and
ISDB-T are optimized for single frequency networks (only one carrier
frequency for many transmitters) compared to 8-VSB. Although, they are
supposed to be testing 8-VSB in single frequency topology.

 

[Joerg]

The basic idea is that, since they're transmitting a digital signal, if the
reflections (ghosts) aren't strong enough to start "flipping bits" (or only
flip a few), the resultant signal is still "digitally perfect" (at least once
error correction is performed) and thus the picture displayed is exactly what
the transmitter started with.

An improvement can be had with a so-called "rake receiver" (see:
http://en.wikipedia.org/wiki/Rake_receiver): Assuming the ghosting is
relatively constant, you're just getting delayed copies of the original signal
at various points in time. If you start sampling at those various points in
time and summing up the result, you can faitfully reconstruct the original
signal. (Of course, finding the correct "various points" is not so
trivial...) Hence you're you're "raking in" all the copies of the original
signals to build up the result.

Supposedly first-generation ATSC receiver chipsets didn't do any of this
active ghost cancellation, whereas second- and (the current) third-generation
chips do.


Somewhat, yes, although with a rake receiver you don't (necessarily) need a
reference signal -- you just try out the various sample points until you get
one that produces the "best" result. Unlike an analog system (where all you
really have to measure are signal to noise ratios) in a digital system usually
there are plenty of synchronization and test data patterns at known locations
within the signal, so it's usually easy to determine how well the system's
performing.

I just bought a digital TV but I have to head to a client so I can't
post any experiences until later. The lab here is in rather extreme
multipath conditions. So it'll be interesting. Bottomline is we'll see
whether or not we will still have TV when they turn off analog in 2009.
It'll be used as a monitor anyhow and for TV we could probably use the
web. The new sets usually have VGA and HDMI inputs.

 

[Joerg]

The basic idea is that, since they're transmitting a digital signal, if the
reflections (ghosts) aren't strong enough to start "flipping bits" (or only
flip a few), the resultant signal is still "digitally perfect" (at least once
error correction is performed) and thus the picture displayed is exactly what
the transmitter started with.

An improvement can be had with a so-called "rake receiver" (see:
http://en.wikipedia.org/wiki/Rake_receiver): Assuming the ghosting is
relatively constant, you're just getting delayed copies of the original signal
at various points in time. If you start sampling at those various points in
time and summing up the result, you can faitfully reconstruct the original
signal. (Of course, finding the correct "various points" is not so
trivial...) Hence you're you're "raking in" all the copies of the original
signals to build up the result.

Supposedly first-generation ATSC receiver chipsets didn't do any of this
active ghost cancellation, whereas second- and (the current) third-generation
chips do.


Somewhat, yes, although with a rake receiver you don't (necessarily) need a
reference signal -- you just try out the various sample points until you get
one that produces the "best" result. Unlike an analog system (where all you
really have to measure are signal to noise ratios) in a digital system usually
there are plenty of synchronization and test data patterns at known locations
within the signal, so it's usually easy to determine how well the system's
performing.

Oh, forgot one thing: This new set has an anti-ghosting comb filter.
Finally! I've used that technique since the 80's and I wonder what took
them so long.

 

[Jan Panteltje]

Good morning world!

I've been reading about the new digital TV standards (8-VSB, DVB-T,
ISDB-T) and they all mention that the standards are more robust
compared to the analog standard. For example, all state that
practically ghosting effects are eliminated with DVB-T and ISDB-T
providing more robust response to mulitpath effects than 8-VSB.

How do these standards eliminate ghosting effects?

Well in essence because digital will grab onto the strongest signal.
Weaker interfering signals produce no picture.

Some modulation systems use energy from each received path.

 

[Jim Thompson]

Oh, forgot one thing: This new set has an anti-ghosting comb filter.
Finally! I've used that technique since the 80's and I wonder what took
them so long.

What brand?

...Jim Thompson

 

[Joerg]

What brand?

Vizio VX37L, which is a 37 incher, at Costco.

 

[StickThatInYourPipeAndSmokeIt]

Hi Joel, Thanks for the response.

I'm assuming that the error correction portion is dependent on the
modulation scheme. So, am I right in assuming that the COFDM technique
being describe as more robust to multipath effects than 8-VSB plays a
part in implementing the error correction scheme?

You know, it's a bit tricky for me to patch up all the stuff I've read
from various references. My main surprise is the fact that DVB-T and
ISDB-T are optimized for single frequency networks (only one carrier
frequency for many transmitters) compared to 8-VSB. Although, they are
supposed to be testing 8-VSB in single frequency topology.

Think of it like this...

An audio CD (analog WORM style track) can have scratches and still play
as the ECC allows the playback hardware to reconstruct the analog audio
stream, but it has no rules that it all be there, and if it isn't, it
will even make pieces up.

In digital realm, ie MP3, etc the ENTIRE datagram has to be there, so
the ECC must PROPERLY reconstruct all of it in order for the decoder to
pass it on to the D/A analog amps.

An missing portions result in a drop out of playback.

So, for digital TV broadcasts, the FEC which is included in the signal
ensures that the datagram is all there and 100% correct so the MPEG
decoder can call it good data and "render" it. If the bit-error-rate
gets much above 10% one will see picture artifacts, lost frames, or the
picture and sound completely drop out until the decoder retrains on the
stream.

Since it is digital, you will NEVER see any multipath, because each
datagram must be completed by the FEC segment before it gets passed to
the MPEG decoder. Any incomplete segments get tossed, and the decoder
picks up on the stream at the next correct datagram.

So what you see is either a 100% complete and perfect picture, or a
frame or two with SOME artifacts (bit blt blocks) OR frames that are
skipped completely, which are usually replaced by blank frames that carry
the color the hardware was programmed to display during missing frame
segments, which is usually black.

 

[ChairmanOfTheBored]

Oh, forgot one thing: This new set has an anti-ghosting comb filter.
Finally! I've used that technique since the 80's and I wonder what took
them so long.

That's probably for the analog tuner section.

 

[Jan Panteltje]

Vizio VX37L, which is a 37 incher, at Costco.

1366x768?

Ar you kidding me?
I was under the impression HDTV was 1920x1080 these days,
and 50 and 60 Hz capable too (for Euope the 50).
The cannot even sell it here as HDTV with that few pixels.
Even my monitor has more.

;-)

 

[ChairmanOfTheBored]

Well in essence because digital will grab onto the strongest signal.
Weaker interfering signals produce no picture.

WRONG! That is called "diversity".

Digital TV is 100% corrected by the FEC which is also sent in the
stream. It isn't that it ignores all but the strongest signal, it IS
that it constructs perfect datagrams ALL the time, and ANY missing data
is reconstructed from the included FEC. If it cannot reconstruct a
perfect datagram, the frame or frames carried in that broken datagram get
skipped.

It really is THAT SIMPLE.

Some modulation systems use energy from each received path.

You are still not talking about the same thing.

 

[ChairmanOfTheBored]

Vizio VX37L, which is a 37 incher, at Costco.

Jeez. Vizio SUCKS. You should have flipped for the extra bucks for one
of the big players ion the industry.

 

[Spehro Pefhany]

1366x768?

Ar you kidding me?
I was under the impression HDTV was 1920x1080 these days,
and 50 and 60 Hz capable too (for Euope the 50).
The cannot even sell it here as HDTV with that few pixels.
Even my monitor has more.

;-)

http://www.costco.com/Browse/Produc...=0&Ntx=mode+matchallpartial&Nty=1&topnav=&s=1


If you don't have a really high quality signal and don't get too close
to it, it doesn't matter much. We just bought a little 23" Toshiba
TV/DVD with that resolution for the bedroom. It showed a great deal of
dislike for the DVD I put in there- I ejected it and then shut off the
power. The drive sucked in the DVD as it was shutting down, and then
*flung* it across the room (at least 4'), as its final act (presumably
some kind of spring post broke inside). Hopefully the replacement unit
will fare better.

http://www.buy.com/prod/toshiba-23h...-768-resolution-divx/q/loc/111/204056501.html

One thing that really ticks me off is that our cable supplier now
encrypts the digital TV signals so you can only use their 'official'
box rather than the QAM digital tuner built into the TV. I guess
unless they're forced to stop it by legislation they'll continue to do
so. They do leave a few music only and on-demand channels (hey, I can
watch what the neighbors are watching, starting from whenever they
started) unencrypted.

Best regards,
Spehro Pefhany

 

[Joerg]

1366x768?

Ar you kidding me?
I was under the impression HDTV was 1920x1080 these days,
and 50 and 60 Hz capable too (for Euope the 50).
The cannot even sell it here as HDTV with that few pixels.
Even my monitor has more.

;-)

I don't need an HDTV, just a monitor

 

[Jan Panteltje]

If you don't have a really high quality signal and don't get too close
to it, it doesn't matter much.

We have quite a few HDTV transmissions now via satellite, in Europe.
Indeed much of the material simply is not in high resolution, not even
in their demo channel.
But there is a screenshot from BBC HD that comes close:
ftp://panteltje.com/pub/00000300.gif
It comes close because it is made at a point where the camera stops panning (almost),
and I disabled de-interlace.
When using deinterlace on LCD the resolution drops to less then half indeed.

We just bought a little 23" Toshiba
TV/DVD with that resolution for the bedroom. It showed a great deal of
dislike for the DVD I put in there- I ejected it and then shut off the
power. The drive sucked in the DVD as it was shutting down, and then
*flung* it across the room (at least 4'), as its final act (presumably
some kind of spring post broke inside). Hopefully the replacement unit
will fare better.

http://www.buy.com/prod/toshiba-23h...-768-resolution-divx/q/loc/111/204056501.html

I had a DVD player like that, others bought the same after they saw me with it,
the mechanism went bad, they had it returned, I put it with the trash.
DVDs are, in my view, best played from the PC, as you can select any decoder / processing,
then connect to the monitor with DVI or whatever.

One thing that really ticks me off is that our cable supplier now
encrypts the digital TV signals so you can only use their 'official'
box rather than the QAM digital tuner built into the TV.

I no longer watch encrypted channels, in the past I did lots of decoding stuff,
but now it is illegal, death penalty I think even ;-) ),
so *if they do not want me to see their commercials, then it is their problem with the advertisers*.
There are really hundreds of free to air channels on satellite here, from all over the world,
from China to Russia, to South America (relayed), except of course US (no way).
But we have CNN Europe free.

I guess
unless they're forced to stop it by legislation they'll continue to do
so. They do leave a few music only and on-demand channels (hey, I can
watch what the neighbors are watching, starting from whenever they
started) unencrypted.

I dunno how many radio channels ere on satellite, but many hundreds....
http://en.kingofsat.net/
Just type a name...

Cable was here, dug a hole in my garden, although I told them I wanted no cable.
I protested, they filled up the hole again, and I have not missed it ever.

I had cable in Amsterdam, late at night the guys went home and turned of the cable....
Whole political fights what should be on cable and what not, satellite is freedom.

 

[Joerg]

http://www.costco.com/Browse/Produc...=0&Ntx=mode+matchallpartial&Nty=1&topnav=&s=1


If you don't have a really high quality signal and don't get too close
to it, it doesn't matter much. We just bought a little 23" Toshiba
TV/DVD with that resolution for the bedroom. It showed a great deal of
dislike for the DVD I put in there- I ejected it and then shut off the
power. The drive sucked in the DVD as it was shutting down, and then
*flung* it across the room (at least 4'), as its final act (presumably
some kind of spring post broke inside). Hopefully the replacement unit
will fare better.

http://www.buy.com/prod/toshiba-23h...-768-resolution-divx/q/loc/111/204056501.html

One thing that really ticks me off is that our cable supplier now
encrypts the digital TV signals so you can only use their 'official'
box rather than the QAM digital tuner built into the TV. I guess
unless they're forced to stop it by legislation they'll continue to do
so. They do leave a few music only and on-demand channels (hey, I can
watch what the neighbors are watching, starting from whenever they
started) unencrypted.

One reason why we don't have cable TV. Just ye old erriall...

 

[Joerg]

Jeez. Vizio SUCKS. You should have flipped for the extra bucks for one
of the big players ion the industry.

I know you prefer the latest and greatest stuff. However, this one's
good enough for here. Of course I could have sunk >$2k into a more
expensive one but 37" for around $750 is a pretty good deal.

 

[StickThatInYourPipeAndSmokeIt]

1366x768?

Ar you kidding me?

He went with the cheap shit. I hope he didn't pay the good shit price
though.

I was under the impression HDTV was 1920x1080 these days,
Yes.

and 50 and 60 Hz capable too (for Euope the 50).

No. Modern HDTVs are double rate 120Hz devices. Notice also that that
rate also kills the need for the 3:2 pull down requisite.

The cannot even sell it here as HDTV with that few pixels.
Even my monitor has more.

HD starts as 720, so it qualifies, regardless of what you, or your
lawmakers do "over there".

 

[ChairmanOfTheBored]

I know you prefer the latest and greatest stuff. However, this one's
good enough for here. Of course I could have sunk >$2k into a more
expensive one but 37" for around $750 is a pretty good deal.

At 37" your minimum requisite should have at least been 1080p.

1368x768 looks pretty grainy.

You essentially bought three year old technology at a closeout price.

That is aside from the fact that Vizio doesn't use the best panel
vendors around, so the end product ain't too hot. Hell, you should have
researched a bit. You can get a Viewsonic at that size and spec for the
same price, since you are buying the old, marked down stuff.