Do You Need Ten Times the Bandwidth?
Sometimes you hear people say you need 100 megahertz of video bandwidth.
Currently I recommend that you choose audio/visual receivers, switchers, video processors, and cables with bandwidth at least twice the maximum video frequency for ordinary analog component video connections.
Maximum actual video frequency content is: 7 Mhz for regular analog video (DVD), 14 Mhz for progressive scan, 37 Mhz for HDTV. So twice that in round numbers is 15 MHz for regular video, 30 MHz for progressive scan, and 75 MHz (almost 100 MHz) for HDTV.
In a Nutshell
Why do some experts say you need ten times as much analog bandwidth (370 MHz for HDTV) as the highest video frequency?
Or why do some experts suggest you buy ultra high grade video cables?
Reasons:
1. The per component versus overall system bandwidth issue.
When several components each with a given bandwidth are connected together the complete system has a bandwidth that is somewhat less and not easily predicted. Experts therefore arbitrarily suggest selecting components each with two to three times the net bandwidth desired just to overcome this issue.
2. The sine-square issue (analog video).
When the system has three to five times the bandwidth than one may otherwise think necessary, tiny details will be reproduced more crisply. But experts disagree as to whether this makes a noticeable difference in picture quality.
3. The sine-square issue (digital video).
If the components are rated in megahertz, we suggest choosing cables, switchers, and other equipment so the entire system has a bandwidth in megahertz five times the desired megabits per second figure. This assures that the pulses in the digital signal waveform will keep more of their squared off appearance. Some TV sets may be less demanding but we don't know which ones.
Three times three.
Taking the factor (multiplier) of two to three from reason 1 and the factor of three to five from reason 2, we should at least vaguely understand why experts came up with a bandwidth figures of six to fifteen times the highest video frequency.
1. The per component versus system bandwidth issue
Insufficient bandwidth here starts to show up as loss of the finest picture details in analog video, and can lead to intermittent, partial, or total loss of the picture in digital video.
Bandwidth with no other qualification means the frequency range over which the frequency reproduced worst (or transmitted worst) comes out at least half as strong as the frequency that is reproduced best. The technical term for this is "not more than 3 decibels down".
If we have two components (each cable counts as a component) with a given bandwidth connected together, some frequencies may suffer the 50% loss twice which in terms of decibels is six dB down. Usually the greatest loss is at the top of the frequency range in question. So we might want to figure out what is now the highest frequency that still comes out at least half as strong as the best reproduced frequency. This is not an exact science and is not easy to measure. So experts simply pick a safety margin such as twice or three times the bandwidth. The more components in the video signal path, the greater the potential loss. For a typical home theater setup, there are two sets of cables and the audio visual receiver between the source device such as a DVD player, and the TV. I suggest choosing components with at least twice the bandwidth needed (up from a previous recommendation of 20% more bandwidth.). Some experts say you need three times the bandwidth for each component to almost guarantee a system frequency response no more than one dB down (less than 20% loss) at the highest video frequency desired.
For HDTV, which requires 37 Mhz system bandwidth, choose each piece of equipment to have at least 74 Mhz bandwidth if you are using component video connections.
2. The sine-square issue (analog video including component video)
The effects upon the picture caused by the sine-square issue are not highly noticeable, unless you are comparing two systems side by side where one system has a much greater bandwidth than the other. We actually believe that you can ignore the sine-square issue completely, for analog component video.
A pure tone or a single frequency is represented by a sine wave. Any other kind of waveform with the same frequency consists of that sine wave together with harmonics (overtones; multiples) of that frequency. When we said that a system has a given bandwidth, all of the harmonics of the highest frequency of interest are going to be twice that frequency or more and most likely little or none of those harmonics will make it through the system. Deprived of its harmonics, a repeating waveform of a given frequency turns into a sine wave of that frequency.
Meanwhile when we think of fine picture details (or pixels) on a screen we think of a crisp transition from one pixel to the next. The crisper the transition, the more like a square wave is the waveform that represents the left to right motion of the electron beam drawing a scan line. The less crisp the transition, the less contrast the tiny detail will seem to have.
To get an idea of the visual difference between dark to light transitions represented by a sine wave versus a square wave video signal, stand back ten feet and judge the relative intensity of the rounded versus squared off red spots below. In a digital situation, the numeric value for a pixel is generally the average taking into account the rise and fall, so a quicker rise and a longer time before the fall for a square wave gives a larger average, a larger pixel value, and therefore a larger intensity. Remember that the sine-square issue only affects the tiniest details* so the test you are viewing is to explain the issue, not show you an actual size case of what you will see.
Alternatively, compare the two rows of black and white spots which correspond to how crisp light transitions might look when reproduced with a sine wave versus a near square wave. Again, since this problem only affects the finest details* (one pixel wide) you really should stand back from the computer screen fifty feet if you want to see what one pixel really looks like. Depending on the way your computer monitor represents the black to white gray scale, either the black spots or the white spots may appear larger.
* If the above diagrams represent single pixels, for wider, say two pixel wide, details the gray blend area or red curved area stays the same width (not becoming twice as wide), so there is more pure black, white, or red.
Comparing a square wave to a sine wave, the most prominent harmonics of the former are the third and the fifth. Some video experts say that not reproducing any of the harmonics is insufficient while reasonably reproducing the third harmonic is adequate. Other experts say that the system should pass the fifth harmonic. These cases mean having a system bandwidth of three or five times the highest video frequency desired, respectively.
The sine-square issue is less important because the source device such as your HDTV set top tuner box has already rounded off the pixels into sine waves because that box output stages did not have three times the bandwidth. If the rest of your system, namely your cables and A/V receiver had five times the desired bandwidth net, this is now overkill. If the rest of your system had two to three times the bandwidth, it won't worsen the picture any more. The system won't take away more of the fundamental waveform simply because the harmonics are already gone.
2. The sine-square issue (digital video)
The sine-square issue is one of the reasons why you cannot equate megabits per second to megahertz when shopping for cables. Another way of comparing a sine wave and a square wave is knowing that a sine wave has a slower rise time than a square wave. In the real world there is no such thing as an instant rise time as a perfect square wave would have, but there do exists specs. for cables and other equipment (we don't have such specs. available here).
If the bandwidth of the cable is insufficient, the rise times of the digital pulses will be lengthened. In an extreme case, the pulse will have ended before the rise time has elapsed with the result that the pulse comes through with insufficient amplitude to be treated correctly as a one or a zero.
It is not easy to predict what cable will not work, since it is not easy to determine how sensitive the rest of your equipment is to an insufficient rise time.
Where the experts got the factor of ten
So when I say you need twice the desired system bandwidth per component to overcome the losses of stringing components together and other experts say you need five times the bandwidth to pass the fifth harmonic, we are up to ten times the highest video frequency for the bandwidth of each component.
Or if someone says you need three times the desired system bandwidth per component to overcome the losses of stringing components together and you or he is satisfied with passing the third but not the fifth harmonic, we are up to nine times the highest video frequency for the bandwidth of each component.
In my opinion the sine-square issue is not important for analog connections. I still contend that if you choose equipment with bandwidth twice to three times the highest video frequency (75 to 110 Mhz for HDTV) you won't find anything to complain about. Most likely your TV itself doesn't have more than 25 Mhz bandwidth for 1080i HDTV but every bit of bandwidth for the other components will help.
All parts (c) copyright 2003-5, Allan W. Jayne, Jr. unless otherwise noted or other origin stated.
If you would like to contribute an idea for our web page, please send us an e-mail. Sorry, but due to the volume of e-mail we cannot reply personally to all inquiries.