[mythtv-users] Video Noise and VHS-to-DVD conversion (long - only read if interested)

Brian Wood beww at beww.org
Sun Jan 7 22:44:46 UTC 2007

On Jan 7, 2007, at 1:45 PM, David Brodbeck wrote:

> Brian Wood wrote:
>> The best connectors for video are actually the PL-259 "UHF"
>> connectors (don't know why they call them that, they are not rated
>> for actual UHF frequencies (defined as 300-3000 Mhz.)). They are
>> hardly used at all anymore though, because they are a pain to install
>> and use, and take up a lot of space.
> They're called "UHF" connectors because they were invented in the  
> 1930s,
> when "ultra-high frequencies" were what we call VHF now.  They're  
> pretty
> much obsolete now, except for shortwave-band stuff like amateur radio
> and CB.  They don't work well above 300 MHz and they don't present a
> constant impedance, like more modern connector types.

Very correct, these connectors are usually only seen in RF systems at  
the HF and VHF ranges these days. They were used for video in the  
"old days". "N" connectors are used at UHF and higher frequencies,  
and these are really just weatherproof BNCs. Then there are "TNC"  
connectors ("threaded" vs. "bayonet" "N" connectors). For smaller  
connectors you see a lot of the "SMA" type connector these days,  
especially on 802.11 devices. You can get away with a lot  if your  
cables are short :-)

RCA connectors make for lousy audio as well. They are unbalanced for  
one thing. "Real" analog audio systems use 600-ohm balanced cables  
with "XLR" connectors (so-named for the Canon part #s). The  
Switchcraft part #s are A3M and A3F for "audio", "3 (conductor)" and  
M or F for the gender of the connector. These are also known as "QG"  
connectors for "quick ground" (study the female end of these and you  
will see that the connector is made so that the ground "makes" first,  
to eliminate hum and other crap in the system when you connect the  

> Fortunately, baseband composite video (like what comes out of your  
> VCR's
> "video out" jack) only has frequency components up to 10 MHz, so  
> none of
> that really matters.  As you point out, professional gear generally  
> uses
> BNC connectors.  Besides having better electrical characteristics than
> RCA connectors, they lock in place, which makes them a lot more  
> reliable.

> By the way, cabling problems tend to show up in three ways:
> 1. Ghosting (caused by improper impedance, which lets some of the  
> signal
> reflect off the far end of the cable)
> 2. Hum bars (caused by poor shielding)
> 3. Poor contrast (caused by excessive losses)
> If you make your own video cables, try not to use regular cable TV  
> coax.
>  It's meant for high frequencies and the shielding is inadequate for
> baseband video.  You want cable with at least 95% braid coverage.   
> This
> is available from places that cater to security camera installers.   
> It's
> more expensive, but will pick up far less hum on long runs.  You
> probably won't see a visible difference on a cable that's only a few
> feet long, but in my job I regularly deal with runs of over 100 feet,
> and this can make the difference between video that's usable and video
> that won't even sync properly.

Quite correct again. Cable designed for video is very different from  
that designed for RF, although with cable modems and STBs using the  
5-30Mhz. band for "return" the two are starting to converge a bit.  
Belden 88281 (AKA "Video Waveguide") works well but not many  
consumers will spend what it costs for the connectors, not to mention  
the cable.
>> You might have to violate some electric codes by lifting the third-
>> wire ground from one or more power cables to eliminate AC hum. The
>> best solution is an isolation transformer but few consumers have one
>> of those. "Humbuckers" and optical isolation units can break up
>> ground loops, but again not many consumers have access to those.
> If the incoming cable TV line is the culprit -- and it often is, since
> they're usually grounded separately from the power circuit -- you can
> sometimes isolate it by wiring two 75-to-300 ohm baluns together to  
> make
> a 75 ohm isolation transformer.  You *will* lose a fair bit of signal
> across an arrangement like that, though, so it depends on how much
> signal strength you have to spare.  You can't isolate baseband video
> this way because it has DC components which get lost when transformers
> are used.

That's one solution, but there are other problems such as the 300-ohm  
"twinlead" portion being subject to ingress of outside-world signals  
into the system, and radiating signal to the outside world as well.  
Since cable systems utilize frequencies that are used for other  
things in the atmosphere this can and does cause problems.

> Even when all the external noise issues are cleaned up, video from  
> VCRs
> tends to capture to digital poorly because it's not very stable.   
> Due to
> tape stretch and other mechanical issues there tends to be a lot of  
> sync
> jitter which drives some capture cards up the wall.  (The professional
> solution to this is a time-base corrector, but as you would say, "not
> many consumers have access to those.")  Slow-speed VHS recordings are
> also full of random color noise, which really messes with video
> compression.  With a variable-bitrate codec, I've seen a capture from
> videotape take up *ten times* the space of "clean" video from a  
> CCTV camera.

A true "TBC" requires a video machine that can accept "advance sync"  
information, not many consumer devices will do that. With memory  
costs dropping a "frame-sync" type device is probably a better way to  
go, but even that is not within most consumer budgets.

One problem is that compression algorithms can't tell the difference  
between "noise" and "detail", but there are some solutions once you  
get into the digital domain.

Nice to hear from somebody who obviously knows what's going on :-)  
Obviously we have to make some sort of compromise between what's  
practical for consumers and what's the "best" engineering solution.

More information about the mythtv-users mailing list