Video Bandwidth
Here is a description of almost everything videophiles may want to know about analog video bandwidth. Bandwidth applies to all circuits, all cables, and even all switches used with audio and video and other electronic signals.
In a Nutshell
Definition: Bandwidth is defined as the frequency range that the circuit, component, cable, or complete system can pass where the frequencies reproduced , transmitted, or amplified worst come out at least half as strong as those coming out best.
Bandwidth is important when shopping for HDTV or progressive scan video equipment.
System Bandwidth Requirements for Analog Video:
NTSC Broadcast and VHS: 4.2 MHzLaser Disk (now obsolete): 5.3 MHz
Regular NTSC DVD: 7 (6.8) MHz
Progressive Scan NTSC DVD and 480p DTV: 13.5 MHz
1080i HDTV: 37 MHz; in practice with 22 MHz the picture is still very superb.
720p HDTV; 37 MHz.
1080p, 74 MHz.
4K and UHD, Theoretically 150 MHz; we do not know of any in analog form.
Digital video also requires a certain bandwidth, which is greater for the higher resolutions (such as 1080p) compared with lower resolutions (such as 480i). Equipment for digital video is rated for the resolution (e.g. 1080p) instead of bandwidth (e.g. 37 MHz).
Why not just call it "frequency response"? Click here.
"Digital quality" really does not mean much. HDTV quality cables should be those with higher bandwidth.
Unfortunately video bandwidth is not easy to determine. Salesmen often don't know it, manufacturers are often reluctant to disclose it, and it is difficult to measure without test equipment.
Currently there is no easy to understand rule for selecting equipment, each piece with a given bandwidth, so that your complete system has the bandwidth you desire. I suggest twice the amount you are thinking of (100% more) for each component. Some experts suggest ten times the desired system bandwidth.
Consumers assume with varying degrees of success that the more expensive equipment is better, and bandwidth often makes the difference.
Limitations in video bandwidth show up first as side to side softening of small or narrow details; as loss of horizontal resolution.
Because the final bandwidth of several components connected together is less than the bandwidth of the "worst" component, we suggest purchasing equipment with a more bandwidth than you think you need.
Lack of Knowledge and Misinformation
Unfortunately many video salesmen knew little or nothing about bandwidth. Unfortunately the publishing of bandwidth figures had been a slow trend. Unfortunately some manufacturers publish the bandwidth for the "strongest link" as opposed to the "weakest link" in the "chain" of circuits in the video signal path.
With the advent of digital video, the rules for relating bandwidth to resolution and quality changed completely. Bandwidth per se affected only cables, that is, transmission paths. Bandwidth for digital video cables is expressed as bitrate rather than just as megahertz.
One example of the "strongest link" advertising is for the horizontal resolution of analog TV sets. Some manufacturers take the bandwidth of the main video amplifier, apply a formula (79 lines of resolution for every megahertz for interlaced NTSC) and publish that figure. Never mind that the video pre-amplifier just inside the S-video jack might have a lesser bandwidth or the screen dot pitch is too coarse to reproduce such fine details.
Even by calling the manufacturer, it is not easy to determine the video bandwidth of equipment or cables.
Using resolution test patterns the consumer can get an approximate measurement of the bandwidth of equipment. So far this is limited to DVD or SDTV (standard, not high, definition TV) since the only commonly available test patterns available (as of 2007) don't go high enough. These test patterns are on the Video Essentials and AVIA video disks. The most important patterns are the frequency sweep and resolution test wedge (broom) patterns.
The wedge patterns are usually on a frame showing a large circle inthe middle and four smaller circles at the corners. Look for the converging lines in the upright wedge remaining distinct all the way down to the narrow end.
The frequency sweep pattern has alternating black and white upright lines getting closer together at the right side of the screen. At the left (wide) end it is OK if the stripes blend from black to white as opposed to having crisp black to white edges. Adjust the sharpness control of the TV so no part of this pattern stands out as being more strident. If the thinner lines at the right side are blurred together, the bandwidth is barely enough for DVD and far short of HDTV requirements.
For digital video, insufficient bandwidth (or bitrate) for cables results in breakup of the entire picture.
"Digital Quality" Means Little
You cannot be assured of the bandwidth of a cable unless the bandwidth or the bitrate is truthfully advertised or the cable is tested. Too often, cables are labeled as being of "digital quality" where the only difference is a higher price than the same cable in a different package. "Digital" cables should at least be suitable for HDTV but then we don't know whether they would barely serve for 1080i (22 MHz) versus do a good job of handling 720p (37+ MHz).
Originally, "digital quality" was applied to analog video cables as digital broadcasting began but the connection such as from a cable box to a TV set was still via analog cable.
Today (2018) most digital cables (HDMI cables and similar) are made with a sufficient bandwidth to assure proper performance with the video resolution (e.g. 4K, 1080i\p) advertised. If there is a shortcoming with the cable, the picture will break up into checkerboards or snow from time to time.
No Good Rules
I suggested buying individual pieces of equipment with at least twice the bandwidth you want to achieve for the whole system.
There was no easily understood rule for selecting analog video equipment, each piece with a given video bandwidth so that your complete system had the desired 22-25 MHz for 1080i or necessary 37 MHz for 720p. In particular, if you connect two pieces of equipment each with the same bandwidth together, the net bandwidth of the system will likely be less.
Let's go back to the 1960's when consumers started moving into high quality audio. Manufacturers started qualifying the frequency responses with plus or minus so many decibels (dB). The word "bandwidth" wasn't used in consumer advertising until at least the late 1960's, and it was a compromise description for the maximum power output to loudspeakers. In my opinion, use of the term "bandwidth", beginning with "power bandwidth" back then, was just a way to let manufacturers advertise audio amplifiers as having so much power, but at the lower end (bass requires more power output) the power was only guaranteed to be half of that (-3 dB).
Audio degradation caused by cables and switchers was negligible. Therefore from what was recorded on the record or tape, the typical worst case audio signal path took into account the phono cartridge, preamp, power amp, and speaker system. Since frequency response advertised was typically plus or minus one dB or less for the electronics, no one worried about the cumulative degradation of separate preamp and power amp. No speaker had perfectly flat response so consumers just accepted the fact that each make and model "colored" the sound differently.
Back to the present day when consumers are evaluating video equipment. Switchers and cables, too, have to be rated in terms of frequency response and impedance (the latter affecting freedom from ghosting). Again, "bandwidth" seems easy to talk about but we don't know where the 1 dB down point is, where the 2 dB down point is, and so on. If a video frequency within the bandwidth of one piece of equipment suffers a 2 dB degradation in each of two pieces of equipment, it will be more than 3 dB down (less than half remaining) after going through the entire system. So we don't know for sure what bandwidth we need to shoot for in each component so that the complete video system has so much bandwidth.
For the time being, advertising will continue to consist of bandwidth recommendations seemingly pulled out of thin air. It may require a return to frequency responses with actual plus or minus dB qualifications, and a corresponding education of the public, for meaningful video performance to be described and understood. For now I am suggesting at least twice the bandwidth (earlier I suggested just 20% more) for each component, for example 74MHz bandwidth for each component to get a system bandwidth close to 37 MHz for 720p HDTV.
Loss of Horizontal Resolution
For analog video, horizontal resolution is the first to suffer when bandwidth is insufficient. The waveform represents the actual changing from dark to light and vice versa as the electron beam sweeps across the screen. If the waveform cannot change fast enough, picture details cannot be made as narrow. The 1080i HDTV theoretically requires a 37 MHz video bandwidth. If the net bandwidth of all the components and cables in the signal path is half that, the smallest picture detail reproduceable is about 1/960'th the screen width as opposed to 1/1920'th the screen width. This is unnoticeable on most of today's TV sets. Other factors, such as electron beam spot size and screen dot pitch also make it difficult to create a dot that small. The 1920 x 1080 format and the smoothness of diagonal lines and edges is still maintained in that picture details may still line up with any of the 1920 positions across the screen.
But if 720p video, which also requires 37MHz, is limited to half the bandwidth, the narrowest picture detail is 1/640'th the screen width as opposed to the theoretical 1/1280'th. Thus softening of the 720p picture due to insufficient bandwidth is more pronounced.
Analog Component Video Switching
It has been suggested that an "audio-video" switch box can be used for component (Y,Pb,Pr) video switching. More specifically, the bandwidth of its internal circuitry is the significant factor.
On average, videophiles attempting to use less expensive audio switch boxes have found the manually switched boxes (with pushbuttons) to be better for video than those with remote control and electronic switching.
There is no currently available easy-to-use test equipment or video disk other than Video Essentials or AVIA. Using one of the latter test disks, a reasonable test can be done as follows:
(If the box is really a component video switcher, use just the component video jacks and not the audio jacks for video cables.)
Test 1. Connect the DVD player component video Y output to a switch box yellow or green input. Connect the switch box yellow (green) output to the corresponding red input. Connect the red output to the corresponding white or blue input. Connect the white (blue) output to the TV component video Y input. If you are using progressive scan, set the DVD player to progressive scan. This is a torture test that puts one video signal through all three signal paths in the switch box.
Play the resolution test pattern with the large circle in the middle and four small circles at the corners. For AVIA use the pattern titled 200 TVL. There should be no degradation of horizontal resolution compared with connecting the DVD player directly to the TV. Also there should not be more echoes or ghosting (ringing) of upright grid lines. Although the AVIA disk test goes up to 540 TVL which is really not enough for HDTV, passing Test 1 probably means the equipment will work acceptably for HDTV..
If Test 1 passes, skip Test 2. (Inferior cables can also cause Test 1 to fail.)
Test 2. Connect up the DVD player using just the Y jacks, using just the switch box yellow (or green) jacks (Test 2a), just the red jacks (Test 2b), and just the white (or blue) jacks (Test 2c) going to the TV. These three tests will also show the test patterns in black and white. If you are using progressive scan, conduct the tests using progressive scan.
Play the resolution test pattern with the large circle in the middle and four small circles at the corners (same as for Test 1). For AVIA use the pattern titled 200 TVL. For DVD the yellow channel should show no degradation of horizontal resolution, the red and white channels should show resolution to at least 300 lines. For HDTV (your DVD player may not have enough resolution) the horizontal resolution should go all the way on all three channels, 540 lines on the AVIA pattern. 540 TVL is really not enough for HDTV but this Test 2 will weed out the equipment or cables that don't even make it that high.
Test 3. Next connect up the switch box with all three component video cables. For an A/V switch box, use the yellow channel for Y, the red channel for Pr, and the white channel for Pb. Play several of the test patterns including the color bars, checking for dot crawl or other distortions. This test is to verify freedom from crosstalk (leakage) between the three video channels. Also view the chroma delay test (vertical red line on a yellow background if your test disk has it.
Be sure to compare what you see on the test patterns while using the switch box against what you see when connecting the DVD player directly to the TV. It is the amount of degradation we want to look for. If a test "fails" without using the switch box, this means the DVD player and/or TV you are using are not good enough to conduct the tests.
Also perform these tests with a prospective A/V receiver you are considering buying. (Omit Test 1 with "upconverting" A/V receivers, namely those that convert 480i to 480p or do similar conversions.
Examples of Bandwidth Calculations
1. NTSC Broadcast Video Horizontal Resolution.
Because the sound is modulated at a subcarrier at 4.5 MHz, the video information must remain below this point and 4.2 MHz was defined as the maximum video frequency.
Although NTSC has about 480 (usually 483) scan lines containing picture information, all 525 scan lines per video frame must be considered when calculating bandwidth.
For NTSC broadcasts the 525 scan lines are repeated 29.97 times per second, 30 frames per second is close enough for our calculations. 30 times 525 is 15750 so you need a frequency of 15750 Hz just to paint the scan lines.
Then, to make pcture details horizontally, we subdivide each scan line. Limited to 4.2 MHz, the video signal can put at most 266-2/3 cycles (4.2 million divided by 15750), each containing one black and one white pixel, on any one scan line, for a total of 533 pixels. In a simple waveform one cycle has one "up" and one "down" and in video "up" means dark and "down" means light, therefore one simple cycle represents two pixels, one dark and one light.
To give the electron beam time to get back to the left side of the screen, and also to provide a place to put synchronizing pulses, NTSC discounts 17% of the scan line. This leaves 83% of the scan line to hold picture information, which spans 442 of the smallest possible details.
So the effective pixel dimensions of NTSC broadcasting are about 442 horizontally by about 480 vertically. In the horizontal direction, the maximum pixel count may vary, for example some systems may "roll off" (start losing) the higher frequencies by more than 50 percent at 3.9 or 4.0 MHz to make sure that there is no video signal above 4.4 Mhz where it may interfere with the audio.
The commonly published figure of 330 lines of resolution for broadcast video comes from the fact that the largest circle fitting in a 4:3 screen spans about 330 of the 442 possible details across one scan line.
Progressive scan needs twice the bandwidth of the corresponding interlaced scan video.
Example 2: 1080 HDTV
We will work this example in the opposite direction, starting off with the pixel count.
1080i HDTV has 1080 scan lines containing picture information (and 45 scan lines for retrace and synchroniziing for a total of 1125 scan lines). Horizontally there are 960 dark pixels and 960 light pixels (1920 visible pixels in all) occupying the first 87.3% of the scan line and we must pretend that there are 140 similarly sized sync. pulses with 140 gaps in between, filling the rest of the scan line. (The total pixel count is 2200 per scan line.)
The scan rate is 30 (29.97) full video frames per second.
Multiply 1125 x 2200 x 30 per second and we get about 74 million pixels per second. Since one cycle consists of one black and one white pixel, the bandwidth needed to display the smallest picture details is 37 MHz.
Coincidentally, 720p also requires 37 MHz.. 720p has has 1280 visible pixels and room for 1650 total pixels per scan line. It has 720 visible and 750 total scan lines repainted approx. 60 times a second. 750 x 1650 x 60 is about 74 million black/white pixel pairs or about 37 million cycles per second.
Laserdisk allows video frequencies of up to 5.3 MHz recorded as composite video. The approximate total maximum pixels is 680, the visible pixels is 565, the EIA* horizontal resolution specification is 425 (TV lines per picture height @ 4:3).
The 480i and 480p DTV formats and DVD provide for 858 total pixels per scan line, 704 to 720 of them visible, with an EIA horizontal resolution specification of about 540 for a 4:3 aspect ratio.
We can get away with equipment having around 25 MHz (a little over half the requirement) for 1080i HDTV because almost no analog TV set can produce a spot smaller than 1/1000'th the screen width (two pixels horizontally). At normal viewing distances the human eye can't distinguish adjacent details that small. Insufficient bandwidth does not make diagonal lines more jagged; picture details can still be put in any of the 1920 x 1080 possible pixel positions.
*EIA -- Electronic Industries Association, a consortium of electronic equipment manufacturers; also refers to the standards they have adopted.
Bandwidth vs. Frequency Response
In analog electronics, ability to reproduce high frequencies or fine video detail does not suddenly drop from clear black on white details, say, 1/720'th the screen width to a total blur if the details are 1/721'st the screen width, rather the distinction gradually fades to gray as the details or dots get narrower. Assuming manufacturers and advertisers are truthful, the term "bandwidth" is used to specify the frequency range with some known amount of allowable degradation, namely 50%.
The term "frequency response" for electronic equipment requires three numbers: a lower bound (for example 20 Hz), an upper bound (e.g. 20,000 Hz), and a tolerance (e.g. plus or minus 3 dB). Bandwidth as a standard or as a term implies a tolerance, or reasonable flatness, of plus or minus 1.5 dB (total 3 dB), or +0, -3 dB, which equals a maximum degradation of 50% from the best reproduced frequency within the range, measured electronically. (It is not easy to "eyeball" the screen and point out a 50% degradation optically.) The lower bound is also omitted when speaking of bandwidth and is assumed to be below any critical value (one scan line or about 16 KHz is about right for video) although not zero (direct current).
If two pieces of equipment each have a bandwidth of, say, 4 MHz and within each the frequency of, say, 4 MHz came through with a relative loss of 50%, the frequency of 4 MHz going through both pieces of equipment is cut in half twice, leaving 25% of its strength. Since the frequency response of equipment or cables tends to roll off at the high end, the bandwidth of this equipment combination is somewhat less than 4 MHz. To reasonably assure that your complete system has close to the bandwidth you want, you need to get individual components each with somewhat more bandwidth. Unfortunately there is no magic formula to help you here but I suggest doubling it.
Digital video is a bit tricky. Some video components process the incoming signal; others just pass it through. If the component processes the signal then the choice of cable for the video output to the next component is not dependent on the quality of the video cable that brought the video signal from the previous component. But if the video component just passes the sighal through, perhaps with switching (some A/V receivers), then a marginal quality cable connected to the input together with a marginal quality cable connected to the output could result in a video signal path that degrades the quality.
What is a Decibel?
A common unit of measure for signal strength in electronics, or audible loudness of sounds, is the "bel", named after Alexander Graham Bell. One bel stands for a factor of ten. Minus one bel (or ten decibels) means cut 90%; minus 3 dB means cut 50% (halved); plus 3 dB means doubled. One decibel is the smallest change in loudness that the typical person can detect.
It so happens that psychologically, sounds rated in terms of loudness on a scale of 1, 2, 3 required energy rated on a scale of 1, 10, 100. (We are referring to electrical energy if the sounds were produced using electronics.) This corresponds to a scale of 1, 2, 3 in bels..
If two cables or pieces of passive (non-powered) equipment each have a bandwidth of 4 MHz and both happen to attenuate the frequency of 4 MHz by the full allowed 3 dB, the net loss at that frequency after the signal goes through both is minus 6 dB (minus "three plus three" dB), or a net loss of 75%.
Decibels are also used in measuring the performance of video circuits although the human eye does not interpret changes in brightness in terms of factors of ten.
All parts (c) copyright 2001-4, 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.