[opendtv] Re: Toward digital TV

  • From: S J Birkill <sjb@xxxxxxxxx>
  • To: opendtv@xxxxxxxxxxxxx
  • Date: Mon, 19 Dec 2005 15:16:01 +0000

I'm sorry, but I must step in and inject a chord of reality into the
speculations in this list about European, and specifically UK, DTT. We may
be across an ocean from ATSC-land, but we're not on Mars, and the facts are
on record:

1. Analog TV in the UK is neither exclusively nor even predominantly low
power or local. The first level of coverage, the UHF main stations, have
typically a pair of 40kW PSP transmitters per service, combining with three
or four other services into a usually omnidirectional HRP antenna, giving
500kW or 1MW ERP per channel. The original (1960s) main stations are
located close to major population centres (London, Birmingham, Cardiff,
Manchester, Newcastle etc.) with an average spacing between stations of
some 80 miles. They serve by far the majority of the country's population.
The second layer fills the gaps: these transmitters are generally located
in rural areas rather than close to any one major city (Heathfield,
Hannington, Sudbury, Sandy Heath, Waltham, Bilsdale etc.) They are also in
the 500kW class, at intervals of about 50 miles. 

Only after that do we get down to the relay stations, with power levels
anywhere from a handful of watts up to 10kW. Unlike much of the USA, where
cities are on the plains or in wide basins, the UK's history has placed
settlements preferentially in valley locations, often shielded by steep
hillsides from geographically distant main stations. Western and northern
parts of England, and most of Wales and Scotland in particular, are
difficult to serve from main stations, and are littered with low power
transposer sites, each serving its own small pocket of population. Even
some sizeable UK cities (Sheffield, Bristol) are notoriously hilly and are
covered, according to local topography, by a combination of distant main
stations and local relays. Here in Sheffield for instance, antennas can be
seen pointing not only to the local high power station (Emley Moor, 20
miles) but also to those in adjacent regions (Belmont, 55 miles; Waltham,
45 miles), as well as to the local relays.

So it is not at all unusual for fixed (rooftop) antennas to receive their
TV over a path in excess of 40 miles, and in places as much as 70 miles. In
the low-rise city suburbs set-top antenna use is widespread, though 'main
set' installations in private houses generally use rooftop or loft-space
yagis, not so much because of low field strengths as to overcome multipath
or co-channel. Of course the UK differs from continental Europe in that
apartment-dwelling (and with it the communal antenna system) is not so
widespread here, outside the inner-city areas.

2. Digital TV uses the very same transmitting stations, but at an ERP (per
mux) nominally 20dB (pending analog switch-off) below the analog peak sync
value. So the signal paths and service areas are essentially the same.
Multipath isn't such an issue, but CCI is, and digital coverage is often
limited by the presence of co-channel analog signals from distant stations.
Where field strengths are low, impulse interference can be the principal
impairment to indoor-antenna reception. In addition, the comprehensive
nature of the original 4-channel UHF analog service plan has meant that
(again, pending ASO) in some areas one or more digital multiplexes has had
to be allocated a frequency outside the locally-grouped antenna sub-band,
so new wideband receiving antennas have been needed. 

3. Most of the London area is well served by the 1MW analog, 6.5kW digital
ERP from Crystal Palace in South London. The (Villiers Street?) ground
floor location described is a mere 6.1 miles from CP, though shielded from
the direct signal by the buildings across the Thames, and by the heavy
girderwork of the Hungerford rail bridge. Not an untypically difficult UK
city-centre location for indoor antenna use, and field strength should be
quite adequate even for a receiver of modest performance. So we needn't be
surprised if a retailer just up the road from there can demonstrate set-top
DTT reception, and a second-floor apartment-dweller get by quite happily
with a Silver Sensor.

4. But... those legacy 'professional' rooftop antenna installations out in
suburbia are not all they may appear. Many date from the end of the 1960s
when UHF PAL colour took over from VHF 405-line black and white. It's not
at all unusual to see broken or missing elements, drooping booms and
quarter-inch open-braid single-shielded co-ax. The DTG and the
Confederation of Aerial Industries are working hard to educate installers
and the public on the importance of a suitable antenna, correctly installed
and with good quality feeder. The public is resistant, and often incredibly
tolerant of the slow decline in their analog picture quality. But they're
not so happy to accept glitches in their digital reception, especially when
the upgrade coincides with the purchase of a large plasma screen. So it
says a lot that the COFDM system is so robust that in the majority of cases
it delivers customer satisfaction without an antenna upgrade. That this can
happen is surely at the root of Freeview's success in the Sky-sceptic,
technophobe majority of middle England. 

5. The suggestion that first-generation COFDM receivers had better tuners
is belied by even the most cursory test of the products themselves, and by
studies carried out by Digital Television Group, the UK industry's
independent and confidential testing organisation. The first COFDM STBs
were built for the failed On-Digital (later ITV Digital) pay service; most
used a basic analog TV tuner, with outboard IF filtering and amplification
prior to the ADC. Intermodulation performance at both RF and IF was poor,
AGC was not optimized for digital, and typical noise figures were in the
region of 9 to 12 dB. Demods were 2-chip or first-generation single-chip
jobs, and frankly a little rough. These units had poor echo tolerance and
struggled, mostly failed, to meet the DTG's target sensitivity of -78dBm
(QEF BER with a 2k 64QAM 2/3 1/32 signal in a Gaussian channel) and analog
ACI protection ratio of -35dB. And they just threw their toys out the pram
when presented with interfering analog signals above about -25dBm -- a bag
of boosters and another of attenuators was part of every installer's kit.

The first boxes sold for Freeview were modified (or in some cases
unmodified) On-Digital products -- the card slot wasn't needed. But with
the first generation of Freeview-specific STBs in 2002 came an
engineering-led drive for improved performance, as typified by my own
SetPal design, long championed here by Dermot. With purpose-designed tuners
these achieved 3.5dB typical noise figures, and combined with the latest
COFDM demods they pushed QEF sensitivity to -83dBm (64QAM, equal to -87dBm
with 16QAM) and ACI PRs through the -40dB mark -- in some cases as high as

The conventional tuner-can manufacturers had to sharpen their act, and
newer designs from the likes of Philips and Thomson were soon up there with
the original SetPal, giving even the cheapest boxes (now <$50 retail)
almost the same performance margins. Despite the cut-throat price
competition rampant in UK consumer electronics, most current Freeview
receivers are so good there is very little to choose among them in terms of
RF performance -- their technology has evolved to be the perfect fit for
the Freeview plug-and-play DTT model.

6. I'd also like to demolish the old recurring myths about COFDM's alleged
power budget inferiority (2dB, was it?) relative to 8VSB, for equivalent
payload and error protection, and the mystical belief that somehow the
European system degrades faster than the US ouside a 'sweet spot' of range
from the transmitter. The first is based upon DVB/ETSI specification
implementation margin minima rather than the real sensitivities achieved by
almost all demods since 2002, and in an unrealistic Gaussian scenario at
that. Any real differences are so small they'd be absorbed in cable length
or antenna gain variations, rather than demanding transmitter power
increase. The second has nothing to do with modulation; it originates from
one flawed field study and has been perpetuated within this list to the
level of dogma, by those who surely know better. But I haven't the time to
present my case right now.

Season's Greetings to all!


You can UNSUBSCRIBE from the OpenDTV list in two ways:

- Using the UNSUBSCRIBE command in your user configuration settings at 

- By sending a message to: opendtv-request@xxxxxxxxxxxxx with the word 
unsubscribe in the subject line.

Other related posts: