Why there is 625 lines [ PAL ] in T.V. scanning or odd lines?

I can't get why there are 625 lines in a PAL scanning system. I read R.R.Gulatti's standard book. In that there is a reason like:

"Suppose there are x lines in 1 field. Then for two field its 2x.For 1 field there are 292.5 scanning lines, so for 2 fields there are 2x=292.5*2=585 & 40 other lines ( because of merging effect and all).So 585+40=625."

I searched this on net but I could not find a proper reason for this. Also in Wikipedia there is no satisfactory answer.

I can't understand why there are 625 lines or odd lines in T.V. scanning. There is a sentence that mentions that "Scanning lines are always odd" but there is no reason given.

I got this solution is it right?

If you had an even number of lines then all lines must start at the beginning of the frame. If you have an odd number of lines then you will achieve the same resolution on the screen but the half line (the odd line), split in half with half at the start of the second frame and half at the end of the second frame allows for an extra line period in order to flyback vertically.

This "appears" to the human eye to be a better picture. If you think about the lines, they are all sloping so if you use full lines then for a full line scan the area at the top right and bottom left have no data in for a wider area than with half line second frame scan. The missing data area remains the same but is now split into two parts, immediately before the half line starts and at the end of the first half line.

• I have added a diagram at the bottom of my answer that matches the description in your last paragraph re the half lines. – tcrosley Feb 17 '14 at 20:04

8 Answers

For PAL, used in Europe, part of Africa, part of South America, Asia and Australia, the number of scan lines is 625. For NTSC, used in most of the Americas and Japan, the number of scan lines is also an odd number, 525. This answer discusses the latter, as I cannot find a definitive answer why PAL uses an odd number of lines.

The National Television System Committee (NTSC, established in 1940) recommended a frame rate of 30 frames (images) per second, consisting of two interlaced fields per frame at 262.5 lines per field and 60 fields (30 frames) per second. Other standards in the final recommendation were an aspect ratio of 4:3, and frequency modulation (FM) for the sound signal (which was quite new at the time).

When the standard for color television was approved, there was a slight reduction of the frame rate from 30 frames per second to 30/1.001 (approximately 29.97) frames per second.

Each frame is composed of two fields, each consisting of 262.5 scan lines, for a total of 525 scan lines, but only 483 scan lines make up the visible raster. The remainder (the vertical blanking interval) allow for vertical synchronization and retrace. This blanking interval was originally designed to simply blank the receiver's CRT to allow for the simple analog circuits and slow vertical retrace of early TV receivers. However, some of these lines may now contain other data such as closed captioning.

The actual figure of 525 lines was chosen as a consequence of the limitations of the vacuum-tube-based technologies of the day. In early TV systems, a master voltage-controlled oscillator was run at twice the horizontal line frequency, and this frequency was divided down by the number of lines used (in this case 525) to give the field frequency (60 Hz in this case).

For interlaced scanning, an odd number of lines per frame was required in order to make the vertical retrace distance identical for the odd and even fields, which meant the master oscillator frequency had to be divided down by an odd number. At the time, the only practical method of frequency division was the use of a chain of vacuum tube multivibrators, the overall division ratio being the mathematical product of the division ratios of the chain. Since all the factors of an odd number also have to be odd numbers, it follows that all the dividers in the chain also had to divide by odd numbers, and these had to be relatively small due to the problems of thermal drift with vacuum tube devices. The closest practical sequence to 500 that meets these criteria was 3 × 5 × 5 × 7 = 525.

This diagram show both the visible lines and the horizontal and vertical retrace lines. I was not aware the latter zigzagged back and forth, but have seen that on a couple of independent diagrams.

Note the half-lines, starting at the top for the odd field and ending at the bottom for the event field.

Here is a good website that describes interlaced scanning in much more detail.

• Actually, NTSC may be 525 but PAL is 625 lines. – RedGrittyBrick Feb 6 '14 at 17:24
• PAL and other 50 Hz systems use 625 lines. Only 60 Hz NTSC uses 525 lines (there may be a 60 Hz PAL somewhere that is 525 lines, as well). Ancient British black-and-white TV was 405 lines, I think. – Peter Bennett Feb 6 '14 at 17:25
• @RedGrittyBrick Good point, I forgot completely about PAL. Of course there is a separate entry for PAL in Wikipedia also. I think my entry is still of use since it discusses why NTSC uses an odd number of lines. (I could not find any information in the PAL entry why the number of lines has to be odd.) The NTSC entry also discusses differences between its standard and PAL. – tcrosley Feb 6 '14 at 17:45
• PAL in the UK only displays 582 lines and of those not displayed I believe 16 were used for the predecessor of the internet called CEEFAX or teletext. If vertical hold went belly-up you could see these extra lines and witness the black and white patterns of the data. By the 1980s I bet most folk in the UK had seen this very early form of internet!! – Andy aka Feb 6 '14 at 22:52
• Why 30/1.001? That seems like asking for trouble. – pjc50 Feb 6 '14 at 23:18

Why 625 for a 50Hz PAL? (or 525 line 60Hz NSTC)

Short answer - an engineering compromise - someone (or more likely a committee) took a decision to standardise on that particular value because it worked with the current technology at that time.

Long answer (a lot of things needed to be considered in designing standards for consumer devices in many different industries world wide and it needs to be seen in an historical context)

n.b. Adjusting for different mains frequencies gives 50/60 x 625 = 520 (near enough 525)

This refers to the 625 50Hz PAL system (a similar argument also applies to 525 60 Hz NSTC)

(1) Transmission bandwith limitation:

There is a limited frequency space for each transmission. Depending on the type of PAL signal the channel bandwidth is between 6 to 8 MHz. Not a lot bandwidth to send all that analogue data for the picture, the audio, sync signals etc. Of the 625 lines only about 576 lines are actually visible. The others being used for frame syncing etc. In other words - there is no point in making a very high resolution colour display because you haven't got bandwidth/ current technology to send it high definition images (this was the 1960s after all).

(2) Mains frequency limitation (for synchronising frames and flicker reduction)

Mains frequency of 50Hz (UK) and given that the human flicker fusion threshold is usually taken as 16 hertz (Hz) someone decided that by interlacing two fields (ODD and EVEN) you could (a) reduce the demand on the bandwidth of the signal (you only need to transmit half the information of the picture in 1/50th sec and (b) get away with a slower phosphor decay) to give a perfectly adequate 'moving' image at 25 full frames per second (easy to divide frequency by 2) (Basic flicker frequency division arithmetic: 50/2 = 25Hz (no flicker), 50/4 = 10.25Hz (flicker))

(3) Line time and dot resolution

Look at the individual LINE output (typical 64uS for 625/50hz PAL). The first part of the signal uses up about 8uS (sync/colour burst) leaving 52uS for display. Manufacturers of TV tubes had to accurately lay down dots/bars of three colours of phosphor across the inside (of comparatively small) curved glass screens. So horizontal and vertical resolutions were limited to how small these dots (or bars) could be produced. Decreasing the size of dot or increasing the area of the screen increases the rejection rates (similar problem to IC manufacture.) Also add to that the repeatable production of some very chunky electro-magnets around the neck of the tube had to move an electro-statically focused beam of high speed electrons (three guns - one per colour) across and down in a very linear manner. Bigger screens meant either a deeper (and very heavy) set or tricky scan systems (Remember this was all done with old school analogue electronics). There was little point in have ultra high resolution - manufacturers couldn't build the display to a price that the public would pay.

(4) Aspect ratio

Back then the aspect ratio was 4:3. (almost square) If you displayed the horizontal line image with a substantially different resolution to the vertical image then your picture was going to look very strange. Roughly the vertical and horizontal resolutions had to be the same. Given that the horizontal resolution was limited by phosphor dot size then the number of vertical lines followed from that (How many groups of 3 phosphor dots/bars could you get vertically). Someone choose just 625 lines (remember it only displays 576) because the arithmetic works. I have no doubt that any number of lines somewhere around about that value would work just as well.

It is a pure Technical and Compatibility issue.

There is a local free run oscillator in the TV sets to ensure the operation even in absents of an input signal. In early TV days the only time base they can use (without much electronics) was the frequency of mains (50 or 60 Hz). But this low frequency it is useless for a reliable picture reproduction. The basic idea is to use this time base oscillator as a frequency divider/multiplier, locked with a new received signal (in much higher frequency) but very rapidly and with a lock range as wider as possible for obvious reasons.

But this harmonic locked oscillator (subharmonic in this case) has low speed, limited locking range and division ratios, and therefore it is not suitable for this application. On the other hand, the injected signal by definition sould be set as a multiplier of the basic local oscillator frequency.

The locking range enhancement technique of Divide-by-Odd is aiming at solving the problem of division ratios. Practically it is a “differential cascode” circuit, that can generate the proper mixing product and meets the desired timebase specifications to use in TV sets, thous achieving maximum stability.
Looking at the mixing components of the two signals, this topology corresponds well to a division ratio of any odd number (3,5,7..). So the multipliers 5*5*5*5, plus the phosphore afterglow persistance, plus the opportunities for bandwidth reduction, resulting the 625 lines (2x312.5). The same valid for any scanning upgrade.

When technology became available at low cost and phosphors response improved and LED etc introduced, the only remaining issue was the “backward compatibility”.

....then politics makes the differece between the TVsystems deeper and wider!!

A CRT requires horizontal sync frequency to be within a certain range, and likewise vertical sync, but conceptually a CRT has no reason to care about any relationship between the two frequencies. If the horizontal frequency is an exact multiple of the vertical frequency, the stripes laid down by every vertical sweep will coincide with those from the previous sweep. If it's not a rational multiple of the vertical frequency, every vertical sweep will lay down passes in a different place. Between those two extremes, if for relatively-prime numbers numbers H and V, the horizontal frequency is H/V times the vertical frequency, then the lines laid down on a vertical sweep will correspond with those laid down V sweeps ago, but not for any between that sweep and the present one. For the scenario where L is odd and F is 2, each line laid down by a vertical sweep will be halfway between two lines that were laid down by the previous sweep.

The reason the number of lines in a frame is odd is that the number of fields per frame is 2, and it's necessary for the number of lines per frame to be relatively prime to the number of frames per field. It would have been possible to design North American video to use e.g. five 120Hz fields per frame (on the assumption that the AC power waveform are likely symmetric); had that been done, the number of lines per frame could have been an even number, but would have had to have been something that wasn't a multiple of five (to minimize 'line crawl' effects, it should probably be a multiple of five plus two or three). That would have precluded the use of a 5x multiplier in the frequency-generating chain, but on the flip side would have allowed power-of-two multipliers. For example, a 648-line format could have used a master clock of 77,760Hz divided by five to yield a 15,552Hz horizontal scan, and by 3x3x3x3x2x2x2 to yield 24Hz vertical [the x3 and x2 factors could be combined in whatever groupings were convenient, e.g. 9x9x8 if one could reliably divide by nine, or 3x6x6x6 if one couldn't divide by nine but could by six, etc.]

The simple answer is that in the original PAL & NTSC TV systems of the 1940's an odd number of scan lines was required for two reasons:

A. The fast persistence of the CRT's of the day required "interleaving" to avoid flicker. Interleaving is where each complete displayed image (625 scan lines for PAL, 525 for NTSC) is split into two, alternating half-images. One half-image (or "field") contains the odd scan lines and the other the even lines. These fields are scanned onto the CRT screen in continuous odd-even-odd-even, ad infinitum order. The basic reason for flicker was that the lines at the top of the screen would start fading as the scanning process reached the bottom of the CRT screen due to the non-persistence of the phosphors which comprise the CRT screen, if the CRT were scanned from top-to-bottom in one complete pass of 525 or 625 lines.

B. In order to achieve interleaving with the hardware of the 1940's, the circuit designers incorporated a "trick" into the vertical scanning circuitry. This "trick" will only work if the total number of scan lines is odd. In order to fully understand how this "trick" worked, you would need to understand how the original vertical and horizontal delflection yokes and their vacuum tube driving circuitry worked. I'm sure that you are not interested in THAT!

I will get to what you actually asked in a moment. In order to produce a picture without too much flicker, it must be repeated about 50 times per second, if displayed on a cathode ray tube. This is because when the electron beam passes a point, it flashes up brightly, then quickly fades. The display has inherent flicker and must be repeated quickly so that the persistence of vision of the eye minimises it. This does not apply to flat screens, because the brightness remains almost the same from one refresh to the next, even at a low rate. Back in the mid 1930s when the original British 405 line system was being developed, the BBC asked for at least 300 lines and EMI decided to give them 405. But it needed a trick. To provide 405 lines, 50 times per second, with similar vertical and horizontal resolution, needs a bandwidth of about 6 MHz, which was impossible at the time. So EMI used interlace. 202 1/2 lines were transmitted with gaps between them starting in the top left. Then the spot returned to the top, half way along a line and filled in the second 202 1/2 lines, in between the first 202 1/2. Thus, points very close together were scanned 50 times per second and the whole picture 25 times per second. This was almost the same as scanning the whole 405 lines 50 times per second, because the eye mainly perceives flicker over large areas, but it only required a bandwidth of 3 MHz. Now to what you actually asked. All interlaced systems must have an odd number of lines so that the second half of them are drawn in between the first half and not on top of them, in which case there would be no interlace and only half the number of lines. The spacing comes from the timing and organisation of the sync pulses, not by deliberately adding an offset. This explanation has been simplified a bit because flyback takes time, but this is the essential reason. Interlace works after a fashion, but it has problems. There is a lot of power in the horizontal scanning circuits of a CRT TV and if it gets into the vertical scanning circuits, it can cause the lines to pair up. Also following objects moving up or down the screen with the eye, can have a similar visual effect. However since the scanning and display were almost simultaneous (real time), it did not cause a problem with horizontally moving objects. One picture of 202 1/2 lines could be drawn and the next one just to its right for example, as it came in from the camera. Interlace causes problems on modern displays, because both interlaced fields, e.g. 2 times 202 1/2 lines, or 2 times 312 1/2 or whatever, are refreshed at once. This creates unpleasant jagged edges on moving objects, because each refresh contains two different half pictures, taken at different times, but moving together as a pair from one refresh to the next. Deinterlacing is difficult and perfect deinterlacing is almost impossible, though there are many half measures.

The 625 lines of each picture or frame are divided into sets of 312.5. To achieve the horizontal sweep oscillator is made to work at a frequency of 15625Hz (312.5 * 50Hz = 15625Hz) to scan the same number of lines per frame (we know that there are 25 frames) so 15625 / 25 = 625. Means line frequency divide by total number of frames we get 625 number of lines.

To avoid flicker the interlaced system was used in order to mimick film. In film you usa a shutter so you can display the same film image twice giving a frequency of 48 instead of 24. It's cheaper. interleaved it's a gimmick that in the end does not work as planned. It's better a progressive system.