Digital TV: Charles W. Rhodes
The Plotting Thickens
In January, Gary Sgrignoli of Zenith and I co-authored a paper presented
at the International Conference on Consumer Electronics (ICCE),
the ideal forum to communicate with engineers designing future DTV
receivers. We are concerned over the extent of DTV-DTV interference
when all DTV transmitters soon go to maximum authorized power.
The following summarizes our paper, "Interference Mitigation
Strategies for Improved Terrestrial DTV Receiver Designs."
Strong TV signals up to -8 dBm at the receiver input will be present
five miles from a full-power transmitting facility. The maximum
symbols of the 8-VSB signal are transmitted 4.8 dB above average
power, so the DTV signal maximum symbol power is 34.8 dB above 1
kW. For analog TV, the peak power is 37 dBK, only slightly higher
from the interference perspective. Spectrum plots where undesired
signals are -19.59 dBm display third-order intermodulation in blue.
I calculated the third-order intercept power (IP3) of the amplifier
under test at 0 dBm, so the maximum signal power was about 20 dB
below IP3. Note that I used 8-VSB DTV signals, not unmodulated carriers
as in the classical two-tone testing of intermodulation.
Fig. 1 shows the spectrum of an 8-VSB signal from a slightly overloaded
amplifier. Note how it resembles the spectrum of a DTV transmitter
before the RF mask filter. The signal power output is shown at -19.67
dBm.
Fig. 2 shows the third-order intermodulation products (IM3) in
the lower adjacent channel (-58.67 dBm) for this signal output power.
The IP3 can be determined from this data to be 0 dBm. IP3 is normally
determined using a twotone test signal, but this technique is of
much greater interest to broadcasters than is two-tone testing because
we can predict DTV-DTV interference with actual DTV signals.
Fig. 3 shows two 8-VSB signals each at -19.59 dBm on adjacent channels
with IM3 in the channel above the signals at -53.32 dBm. The noise
IM3 in this channel is 6.35 dB higher due to the second 8-VSB signal
of equal power. That is, a 3 dB increase in power input results
in a 6 dB increase in distortion (noise). Theory suggests this increase
would be 9 dB for a two-tone test signal, but it's not so with DTV
signals because of spectrum spreading.
Fig. 4 also shows two distorted 8- VSB signals, but this time,
they are separated by one channel. In this case, they are on the
channel pair n-4 and n- 2. The noise (IM3) in the desired channel-
shown in light blue (n)-is due to IM3 generated in the receiver
and measures -53.53 dBm. Note the noise spreads above the desired
channel. Five channels will have significant noise with two undesired
signals spaced by one channel. Having done that, the minimum usable
DTV signal power is determined by adding 15.2 dB (the SNR of our
8-VSB signal) to the total noise in the channel. This demonstrates
why the receiver should not generate IM3, as radiated sideband splatter
is significant without any receiver-generated IM3.
Fig. 5 also shows two 8-VSB signals of equal power but these are
the n-6, n- 3 channel pair with n-6 not shown. The IM3 in the desired
channel (n) is -52.97 dBm. With these -20 dBm undesired signals
present, the minimum usable desired signal power is > -38 dBm.
Fig. 6 shows the n+3 and n+6 (though it is not shown) channel pair
whose IM3 in the blue-tinted channel (n) is -53.71 dBm. Note that
there is significant noise on both n-1 and n+1. Spectrum spreading
is even greater with increased spacing between two undesired signals.
Fig. 7 gets more interesting. Three undesired DTV signals on Channels
35, 36 and 37 are shown. Each alone would have been at -19.67 dBm,
but the three together have increased the total signal power and
hence overloaded the amplifier under test.. These signals suffer
compression because they are at -20.46 dBm instead of -19.59 dBm.
The effect of signal compression never has been fully studied as
it applies to 8-VSB, but I believe that 1 dB of signal compression
is too much. That is, this amplifier works well with -19.59 dBm
power output, but it couldn't handle more.
Fig. 8 shows that the three 8-VSB signals produce IM3 in an adjacent
channel at -45.01 dB, a big increase over the IM3 with two 8-VSB
signals. The desired signals not only suffer from IM3 noise, but
also from signal compression. This topic will be the subject of
future studies.
Fig. 9 shows the IM3 in the other adjacent channel, -43.91 dBm.
Finally, fig. 10 shows the spectrum spreading when two or more
undesired DTV signals of comparable power are present and spaced
by one channel. The noise power in the blue-tinted (desired) channel
is -53.15 dBm. It's about the same two and three channels above
the higher undesired channel, and of course for two channels below
the lower undesired channel. This is a "blockbuster,"
with five channels subject to jamming.
Having demonstrated that a receiver with an IP3 of 0 dBm will be
in serious trouble, we demonstrated that by attenuating the antenna
signal by even 5 dB, the IM3 would decrease by 15 dB and the minimum
usable desired signal power would decrease by 7.7 dB. More RF attenuation
brings even lower minimum desired signal power levels and this will
be needed.
This is too cumbersome for most viewers, so RF attenuation should
be automatically switched by the receiver without viewers having
to do something as they channel surf. We then showed how this can
be done with wideband RF automatic gain-control (AGC) circuitry.
We then compared this to what could be done with higher IP3 mixers
(these require excessive LO power and can be costly) and high IP3
RF amplifiers, which are needed unless wideband RF AGC circuit is
used.
Now you know a bit of what the ICCE attendees learned (I hope)
and you didn't have to fly to Las Vegas to find out.
I want to acknowledge the extensive assistance rendered by Mark
Aitken and Harvey Arnold of Sinclair Broadcast Group for the use
of their spectrum analyzer and their help in producing the spectrum
plots. Stay tuned.
Charlie Rhodes is a consultant in the field of television broadcast
technologies and planning. He can be reached via e-mail at charleswrhodes@worldnet.att.net
|
