dsi@unccvax.UUCP (Dataspan Inc) (02/15/85)
This month's SMPTE journal just arrived, and there is an excellent
article by Alexander G. Day of the Canadian Association of Broadcasters.
Interestingly enough, he points the finger at "receivers" being the entire
problem, although he feels that 6 mHz (4.2 mHz luminance BW) is wholly
inadequate.
I'll post his "Table 1" as an indication of degradation (in response
to my remark about playing a CD player through an AM transmitter....being
representative of video quality)
"At output of" SNR (dB) "Kell factor" Resolution
Studio camera 55 1 500-600 H
VTR 52-54 1 500
Studio vision chain 50-55 2 500
Microwave/Satellite 48-52 2-3 450-500
Transmitter 46-50 3-5 335
Cable system 40-50 4-6 300-320
Receiver accessories 38-48 5-7 275-300
Receiver 35-50 5-20 200-250
This resolution stuff is sticky business. I suppose he really means
horizontal resolution, because the vertical K-factor goes up as the scanning
spot size goes down. Nevertheless, the degradation due to the receiver is
impressive.
Although I do not share his "hopelessness" view of NTSC (because it
isn't -- we use NTSC for diagnostic radiology with fully diagnostic results)
the article makes for interesting reading. Elsewhere in this month's Journal,
Wendland and Schroder make a more reasonable case for HQTV using the PAL-B
system as well as MAC (multiplexed analog components). PAL-B is roughly
equivalent to NTSC (625 lines, 50 frames, 2:1 interlace) but the improvement
is not really dramatic. Special techniques such as preprocessing by dithered
sampling, and the use of frame stores, adaptive notch/comb chrominance
decoding at the receiver (comb filters alone degrade vertical chrominance
resolution) and progressive scanning techniques could satisfy your demand
for excellent TV pictures.
* * *
Right now, though, the main problems are that the present receivers are
terribly poor. Bad nonlinear-phase filtering, deliberate demodulation of the
colour along axes (and basis transformation matrices) which limit the "palette"
of colours; poor high voltage regulation and slew rate limiting in the final
video amplifier...this is what ails you.
The RCA ColourTrak 2000 suffers from very poor high voltage regulation,
compared to a Conrac 5700. No, the components used in the Conrac don't cost
appreciably more. They use a funny high voltage tripler, a handful of jelly-
bean parts, and an ordinary flyback transformer. You can see poor HV regulation
when graphics are on the screen and the raster is wider at that point due to
the lowered velocity of the electron beam. The main problem with consumer
recevier design has been to get power consumption down by skimping on the
high voltage section. (Come to think of it, poor power supply regulation is
what ails most audio equipment, too)
An argument can be made for getting the performance of NTSC up to snuff.
The tecniques exist for eliminating the channel-to-channel hue shift problem
(such as VIR and a proposal about 10 years ago called ART-advanced reference
transmission). The fine-pitch CRT screens that would be required for 1125-line
HDTV (although HDTV is impressive) limit brightness because of the fragility
of the shadow mask. In order to support dot pitch of 0.31 mm or less, the
shadow mask has very low rigidity, and actually WARPS at high beam current
due to local heating. This, in turn, causes the "purity" to shift. (Put
a white square on such a monitor and crank the brightness way up...it will
usually turn "pink" in a matter of minutes.) The warp can be, uh, permanent,
and you have to junk the CRT. This would be socially unacceptable.
A second unhappy consequence is lowered brightness. Those huge dot
pitches in present use allow one to crank the brightness to the point of
pain. In high school I repaired consumer sets, and was continually amazed
at the amount of brightness (and especially colour level) people would dial
in. A reeducation program (Sterno Review, are you listening) could help here.
.
.
.
Well, there I go again. If you own an RCA ColourTrak, you can fix
the demodulation problem with a 1000 ohm resistor in most cases. Nothing will
fix 20 % regulation. Not that I'm picking on RCA (everyone else has these
problems) . . . it is just the brand where the good part of the learning
curve is.
dya
.