[opendtv] Analog or Digital: Ghosts Plague Reception
- From: "Manfredi, Albert E" <albert.e.manfredi@xxxxxxxxxx>
- To: "opendtv@xxxxxxxxxxxxx" <opendtv@xxxxxxxxxxxxx>
- Date: Fri, 17 Sep 2010 19:54:52 -0500
Figure 1 in the Charles Rhodes' article looks like this:
Bisector between
T7 and T2
|
|
T7--|--T2
|
T6 T1 T3 (Radius = 10 miles for each tower)
T5 T4
The radius of each transmitter's coverage is 10 miles, providing an overall max
diameter of 60 miles for the SFN, but with a 22 percent loss in coverage area
caused by non-overlap in coverage circles. He talks about increasing power to
mitigate that loss, which of course also exacerbates the potential for
interference that can't be overcome.
Here's my take on his SFN assumption. BTW, I think the towers of this SFN are
too far apart, even for practical DVB-T.
1. If the transmitters are synchronized to emit their symbols at the same time,
then IDEALLY, you should never see a "leading echo." Meaning that the strongest
symbols would ideally always be those that arrive first, and the other towers'
symbols would always be weaker and arrive later. In some locations, like along
that bisector line between T7 and T2, a receiver would see symbols from T2 and
T7 as if they were from one transmitter. They would be equal strength and
arrive simultaneously.
2. This ideal condition should be good news for ATSC, since basically all Gen 2
and newer receivers do quite well with lagging echo. Even gen 2 could manage
+44 usec or so of loud echoes. However in practice, with obstacles in the way
or weird atmospheric conditions, some distant tower could provide stronger
signals following that of the closer tower, or some more distant towers will
provide "strong enough" signals in addition to the closest two. Bad news. So
this is not what I would deem reliable.
3. Charles talks about degraded performance beyond 10 usec echo, for ATSC
receivers. That is only for leading echo, and some are much better than others.
I think that if the FCC really wanted to deploy SFNs, they should have mandated
echo tolerance requirements. It seemed obvious even way back when. And a good
choice from 2005 would have been the Samsung Gemini chip, with total +/- 60
usec echo tolerance. Or COFDM.
4. To be more specific about COFDM, in the 8K version, an SFN like the one he
postulates would require a GI of 1/4, if the towers are synchronized. Which is
a substantial hit in spectrum efficiency (25 percent capacity lost). You need
1/4 GI because the towers are up to 40 miles apart. Disregarding echoes from
sources other than the towers, this creates a possibility of echoes out to ~213
usec. In 6 MHz channels, 1/4 GI is good for 298 usec. So this should be safe.
If the towers are passive retransmitters, all bets are off.
The next issue, of course, is what this actually buys you, given that we want
single-market SFNs. The safe COFDM configuration costs you 1/4 of capacity. And
the next market over will need different frequencies anyway. It could buy ease
of reception, in the ideal case, as long as you aren't at the edges of
coverage. In the outer perimeter, and in areas of weak signal inside the SFN,
reception could be more difficult than with a big stick. That 22 percent
coverage loss problem.
Bert
-------------------------------------
http://www.tvtechnology.com/article/106586
Analog or Digital: Ghosts Plague Reception
by Charles W. Rhodes, 09.17.2010.
It is a little early for ghost stories, but let's hope single frequency
networks don't come to haunt broadcasters.
Ghosts plagued the reception of radio signals long before there were TV
signals. At night radio signals reflected from the ionosphere return to earth
sometimes thousands of miles from the station. At such distances, the ground
wave is nonexistent, so the station wasn't heard, but its ionospheric
reflections were heard. You heard ghosts.
Early TV experimental broadcasts used AM radio transmitters around 1,500 kHz.
Any night, their pictures could be received, but with a ghost or two below the
image-not to its right side as with analog TV, but below because the radio wave
traveled hundreds of miles to the ionosphere and back.
When commercial TV broadcasting commenced just before World War II, ghosts due
to reflections from hills and manmade structures appeared to the right of the
image. The only way to remove them was with a rooftop, directional antenna
aimed to reduce ghosting.
We never see ghosts with DTV, but we do have echoes. You will never see a ghost
on a DTV screen, but because of echoes of the DTV signal you may see and hear
nothing at all if the ATSC receiver's Adaptive Channel Equalizer (ACE) cannot
cancel the echo or echoes present without lowering the signal-to-noise power
ratio (SNR) to or below the threshold SNR of the ATSC System, 15.2 dB.
Echoes upset the flatness of the spectrum of all DTV signals, be they ATSC,
with 8-VSB modulation or DVB-T or ISDB-T with COFDM modulation.
With COFDM modulation, there are thousands of carriers spaced a few kilohertz
apart across the channel. Each carries a very low data rate as the Symbol Time
is in milliseconds, not nanoseconds as is the case for our single-carrier 8-VSB
modulation scheme.
Since COFDM receivers do not have an ACE, they get rid of echoes by shutting
off the receiver for the first few tens of microseconds (the Guard Interval) of
each Symbol Time. This costs them some reduction in data rate, but it is highly
effective in killing ghosts. Any echo, which falls outside the Guard Interval
acts as noise lowering the received SNR.
About 2005, excellent designs for the ACE of ATSC receivers were introduced. In
the June 23 issue of TV Technology, spectrum plots of ATSC signals and DVB-T
(with COFDM) afflicted by a single strong echo were published. You can still
see these on www.tvtechnology.com under Digital TV.
Leading echoes, those which arrive before the stronger signal, are usually
cancelled by a Finite Impulse Filter (FIR) in the ACE, and those echoes
arriving after the stronger signal, lagging echoes, are always cancelled by an
Infinite Impulse Response Filter.
This means while some receivers may behave in the same way for lagging echoes,
they may react differently to leading echoes. What it also means is that
different receivers behave differently as the design of these filters is
proprietary and is not subject to performance specifications by the FCC.
In countries using COFDM, the administrating governmental entity for their
transmitting facilities determines the width of the Guard Interval and all
receivers behave in the same way. In mountainous terrain, larger Guard
Intervals are chosen.
We have more than one hundred million DTV receiving devices whose echo
cancelling is not standardized. It varies with the chipset used by the receiver
manufacturer, and it varies model year to model year.
What we do know is the velocity of all electromagnetic radiation is 300 meters
per microsecond or about 1000 feet per usec or 0.19 mile per usec. The echo
delay is 5.28 microseconds per mile.
This means that, the ACE in the best 2005 ATSC receiver shown in Fig. 1 of my
June 23 column performs well in locations with echo delays of less than two
miles. However, its performance degrades with echos of 10 µsec or greater. For
an echo at +/- 60 µsec, it can do nothing. Such echoes act like noise in the
receiver. A convenient estimate of signal field strength vs. distance is 0.9 dB
per km or 1.4 dB/mile.
INTERFERENCE BETWEEN TRANSMITTERS
Fig. 1 shows an SFN composed of seven transmitters, each of which radiates
enough power to provide reception for say 10 miles around the tower. The
circles around the transmitter sites correspond to the minimum usable ATSC
field strength, which the FCC says, for a directional, rooftop antenna feeding
one TV set is 41 dBuV/m. Field experience suggests that it may be as high as 50
dB µV/m.
The large outer circle represents the minimum field strength from a single
transmitter, which provides the noise-limited coverage the station expects, per
the FCC. I chose a 60-mile radius typical of a medium-sized market. Each small
circle (10-mile radius) represents that same field strength, but from one of
the seven smaller transmitters of this SFN.
The shaded areas of Fig. 1 indicate the lost coverage of this SFN, which is
about 22 percent of the former coverage area. By increasing the Effective
Radiated power of each Tx by 1.1 dB much of the former coverage area could be
regained, as the coverage area is no greater than the present FCC coverage
allocation. It is not a quite circular.
Some will still lose reception where the field strength is too weak at their
home unless they install a well-designed low noise amplifier at their rooftop
directional antenna. There are six echoes from the other transmitters, plus
whatever reflections are present at that site.
Fig. 1 shows the echo path lengths in microseconds to an arbitrary receiving
site. The nearest transmitter of an SFN does not always provide the strongest
signal because the direct path from that Tx to some sites may be blocked either
by terrain features or manmade structures. This is especially true for an SFN
where the transmitter antennas will be on short towers, perhaps 250 feet
compared to the average height of 1250 feet for present towers.
In Fig. 1, a bisector A-B is shown perpendicular to the baseline between T1 and
T2. Along this bisector the paths from T1 and T2 are of equal length, so in
theory, these signals are of equal power and equally delayed. This gives rise
to the notion of a zero dB echo. Anywhere along this bisector, many modern ATSC
receivers can be expected to work as they can handle one zero dB echo.
In practice, points where these signal powers are equal will be distributed
around the bisector because the path losses from the transmitters are unequal.
Moreover, the directional rooftop antenna should be pointed towards one of
these transmitters and therefore away from the other. The FCC assumes the DTV
receiving antenna provides 10 dB gain and a front-to-back ratio of 14 dB in the
UHF band. This would shift the equal power line far from where it is in Fig. 1.
One mile from the bisector in Fig. 1, one echo arrives 10.3 microseconds before
the other. If the early signal is weaker than the later signal, we have a
leading echo of -10.6 µsec. If the early signal is stronger than the other, we
have a lagging echo of +10.6 µsec.
The adaptive channel equalizers of some unknown number of ATSC receivers and/or
down-converters have limited capabilities to reject a leading echo. These may
fail within one mile from the bisector line. Next month, we will continue this
analysis.
Charles Rhodes is a consultant in the field of television broadcast
technologies and planning. He can be reached via e-mail at cwr@xxxxxxxxxxx
----------------------------------------------------------------------
You can UNSUBSCRIBE from the OpenDTV list in two ways:
- Using the UNSUBSCRIBE command in your user configuration settings at
FreeLists.org
- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word
unsubscribe in the subject line.
Other related posts: