240 volts

I was a junior engineer at Bell Labs at about the time all this was happening, but not directly involved with TV. Thanks for the insight!

Aren't we glad that the CBS method of using a whirling "color wheel", didn't become the standard. I actually built a working model in 1949. The DLP TV sets have gone back to this way of getting color!

It would be interesting to attempt to do your analysis on PAL or SECAM, (or even the new digital standards).

[snip]

Fascinating. That also explains the mystery why the vertical synch was defined to be 59.94 Hz when it "should have been" 60 Hz to avoid power line interference effects.

On Wed, 05 Mar 2008 12:59:13 -0500 Michael A. Terrell wrote: | snipped-for-privacy@ipal.net wrote: |> |> On Wed, 05 Mar 2008 02:18:39 -0500 Michael A. Terrell wrote: |> |> | snipped-for-privacy@ipal.net wrote: |> |>

|> |> How many degrees of phase drift do you think is OK? |> | |> | You still haven't told us WHICH burst circuit you are using. There |> | are a lot of variables, and no single answer to your vague question. |> |> I'm not talking about a specific circuit. | | | In other words, you have no clue how rthe different circuits work. | That's no surprise.

If that is your conclusion, you have no clue of logic.

|> You can talk about any circuit you wish. | | | You started this crap, so it is YOUR problem to find a circuit and | analyze it. Do all the math you want, but you won't get an answer | without a prototype.

I clearly didn't need any "answer" you offered.

On Wed, 5 Mar 2008 19:57:13 +0000 (UTC) Dave Martindale wrote: | snipped-for-privacy@ipal.net writes: | |>There's no ignorance in what I say. If there was, anyone who wanted to |>point it out would have been specific and said exactly what was wrong. |>No one did. | | Has it ever occurred to you that people may avoid pointing out errors in what | you say because of your combative debating style, not because they agree with | what you said? I expect to be attacked just for saying anything that | disagrees with you, no matter how correct or well-supported what I write is.

Are you making that determination about debating style based on how I answer posts that fail to point out specific errors? You cannot conclude what the style would be for specific topical challenges when what you see is how I respond to personal attacks from people who have a history of doing so.

| For the record, here is something I wrote (a long time ago) to explain where | the magic numbers in NTSC come from. The ratio 63/88 does not appear | anywhere in the original standard that I could see. There are a number of | other ratios that do appear, and a particular product of them can be reduced | to 63/88. So that value is theoretically exact - but knowing it doesn't tell | you anything about where it came from. The note below does.

I got that expression from the FCC rules. I came up with something else when I worked backwards from the original number. That something else is equivalent to 5000000*63/88.

| ----------------------------------------------------------------------------- | | This is based mostly on the NTSC committee's own report, with a little bit of | guessing on my part (in the section about prime factors of divisors). | | Original B&W standard: | | * 60 Hz vertical frequency, so "hum bars" from poor power supply rejection | are stationary on screen | | * Horizontal frequency is 525*60 Hz. Odd number gives interlaced image, to | give off better vertical spatial resolution in a fixed bandwidth | | * Channel spacing is 6 MHz, with 4.5 MHz offset between sound and video | carrier | | * Video is transmitted vestigial sideband, with 4.2 MHz video bandwidth. | | The new color standard needed to be compatible with existing B&W receivers: | | * Colour information would be encoded on subcarrier; subcarrier would be | visible on B&W receivers in areas of saturated color. | | * To minimize visibility of subcarrier, lock subcarrier to H sync so any | resulting pattern is stationary, not moving. | | * Use odd multiple of half line frequency, so subcarrier forms a fine | "checkerboard" instead of lines - less visible. | | * The higher the subcarrier frequency, the less visible on B&W sets, but the | less bandwidth available for carrying color information. Tests showed the | best compromise frequency to be around 3.6 MHz. | | * The two constraints above mean that subcarrier should be approximately | 457/2 times horizontal frequency. But 457 is a prime number, and dividing | by 457 is hard - there are no cheap digital dividers available in 1950. | | * Looking at nearby odd numbers, 453 = 3*151, 455 = 5*7*13, 459 = 3*3*3*17, | and 461 is prime. 455 is the easiest divisor to generate - all its | prime factors are 13 or less. So subcarrier is set to Fh * 455/2. | | * So, at this point, the magic numbers are: | Fv = 60 | Fh = 60 * 525/2 = 15750 | Fsc = Fh * 455/2 = 3583125 | | | But there's a problem: to minimize visibility of any beat frequency between | color subcarrier and sound carrier, it is desirable to have the difference | between the two be an odd multiple of half the line frequency. | | * With numbers above, offset is 916875 Hz. 916875/Fh = 58.21 = 116.4/2. | So nearest odd multiple of Fh/2 is 117/2. | | * Thus new sound carrier offset should be Fh*(455 + 117)/2 = 4504500 Hz. | This is (exactly) 1001/1000 times the old sound offset. | | * But (in those days) TV sound used a separate FM transmitter and possibly | a separate antenna; changing sound offset means retuning the sound | transmitter. | | * To avoid this, the NTSC moved all the *video* frequencies down by a factor | of 1000/1001 instead, giving the desired relationship between subcarrier | and sound carrier. | | * So subcarrier becomes 3583125 * 1000/1001 = 3579545.4545 (rounded to | 3579545 in the original standard). | | * New frequencies (without intermediate rounding) | Fsc = 3579545.4545 | Fh = Fsc * 2/455 = 15734.266 | Fv = Fh * 2/525 = 59.94 | | * The tolerance on these is 3 PPM, so the range of permitted values is | entirely within the looser tolerances of the old B&W frequencies, so | B&W TVs should continue to work at the new frequencies (though hum bars | will now roll slowly). | | * If you happen to have a precise 5 MHz frequency standard, to derive Fsc | from it your need a multiplier of | | (60 * 525/2 * 455/2 * 1000/1001) / 5000000 = 63/88 (exactly) | | So the numbers 63 and 88 never appear in the NTSC standard; they are just the | rational number defined by all those *other* numbers above, reduced to | simplest form.

Excellent summary. I had read most of that in various places before. But this sums it up succinctly. Thanks.

On Wed, 5 Mar 2008 22:01:09 +0000 (UTC) Michael Moroney wrote: | snipped-for-privacy@cs.ubc.ca (Dave Martindale) writes: | |>For the record, here is something I wrote (a long time ago) to explain where |>the magic numbers in NTSC come from. The ratio 63/88 does not appear |>anywhere in the original standard that I could see. There are a number of |>other ratios that do appear, and a particular product of them can be reduced |>to 63/88. So that value is theoretically exact - but knowing it doesn't tell |>you anything about where it came from. The note below does. | | [snip] | | Fascinating. That also explains the mystery why the vertical synch was | defined to be 59.94 Hz when it "should have been" 60 Hz to avoid power | line interference effects.

And it's a good thing they did design it the way they did. Otherwise the backronym for NTSC would have become (N)o (T)elevision (S)ince (C)olor.

Right. that's why you keep spouting your inanae crap. You have no clue as to what works in the real world.

Thanks.

I'd be interested as well, but it would probably have to be done by someone with better access to PAL or SECAM documentation. I just happen to have a copy of the NTSC committee's original report, but nothing similar for the other standards.

What I do know:

PAL started from a monochrome system with different frame and field rates, wider channel spacing, and thus more video bandwidth. This allowed them to use a higher subcarrier frequency, reducing its visibility. Despite that, there was room in the channel to transmit both colour components double sideband, so they could encode (R-Y) and (B-Y) directly instead of having to generate I and Q.

But they did a couple of weird things. They specified different RGB colour primaries (which are probably more realistic than the NTSC ones). But instead of recalculating the RGB to YUV transform matrix for these new primaries, they re-used exactly the same matrix as NTSC. And the relationship between subcarrier and H sync is more complex, with a 90 degree phase shift per frame that I don't understand at all.

Dave

It actually was 60 Hz in B&W days.

In retrospect, it seems like changing all of the video frequencies to avoid touching sound was a bad decision. If they'd just left sound alone, it wouldn't be quite an odd multiple of half the line frequency, but it's FM anyway - it's not a fixed frequency. Or they could have move the sound carrier up a bit - a 4500 Hz shift in sound frequency in a system with 25 kHz deviation should not have screwed up sound receiption in the old B&W TV sets.

Either way, we could have had exactly 60 Hz frame rate, and drop-frame time code would never have been invented, and the slow time error accumulation of even drop-frame code would not worry anyone. And 24 FPS films could be run at exactly 24 FPS instead of 23.98. Many potential headaches would disappear.

On the other hand, ironically, the multiplier to generate subcarrier from a 5 MHz reference source would become 5733/8000 instead of the value of 63/88 we have now. By pure blind luck, the 1000/1001 multiplier used to shift the video frequencies just happens to cancel most of the prime factors of 455:

455/1001 = (5*7*13)/(7*11*13) = 5/11

and that's mostly why you get the simpler 63/88.

Dave

I was remembering a discussion from a few weeks ago about 3 phase power on utility poles. However, I now see that I confused you with "krw". It was him who was being unpleasant to people (including me), *not* you.

So my comments about your debating style are unjustified, and I apologize for them. It's my fault for not checking who said what before writing.

On the other hand, my more general point still stands. Because someone (not you) was throwing around unjustified personal insults in a previous discussion, I was reluctant to join this one. I imagine there are other people with technical knowledge that also avoid contributing because of the tone. Thus, it's naive to suggest that anyone who disagreed with you would have posted an article saying that, and to take the fact that no one disagreed with you publically as evidence that no one disagrees. It only takes one unreasonable person to create an unpleasant atmosphere.

Dave

On Thu, 6 Mar 2008 21:18:42 +0000 (UTC) Dave Martindale wrote: | snipped-for-privacy@world.std.spaamtrap.com (Michael Moroney) writes: | |>Fascinating. That also explains the mystery why the vertical synch was |>defined to be 59.94 Hz when it "should have been" 60 Hz to avoid power |>line interference effects. | | It actually was 60 Hz in B&W days. | | In retrospect, it seems like changing all of the video frequencies to | avoid touching sound was a bad decision. If they'd just left sound | alone, it wouldn't be quite an odd multiple of half the line frequency, | but it's FM anyway - it's not a fixed frequency. Or they could have | move the sound carrier up a bit - a 4500 Hz shift in sound frequency | in a system with 25 kHz deviation should not have screwed up sound | receiption in the old B&W TV sets. | | Either way, we could have had exactly 60 Hz frame rate, and drop-frame | time code would never have been invented, and the slow time error | accumulation of even drop-frame code would not worry anyone. And 24 FPS | films could be run at exactly 24 FPS instead of 23.98. Many potential | headaches would disappear.

ATSC has real 60 Hz (and 30 Hz and 24 Hz) as optional modes, in addition to those that are 1000/1001 lower.

| On the other hand, ironically, the multiplier to generate subcarrier | from a 5 MHz reference source would become 5733/8000 instead of the | value of 63/88 we have now. By pure blind luck, the 1000/1001 | multiplier used to shift the video frequencies just happens to cancel | most of the prime factors of 455: | | 455/1001 = (5*7*13)/(7*11*13) = 5/11 | | and that's mostly why you get the simpler 63/88.

A fixed number of audio samples per field would be nice, too. That is not the case with 59.94 Hz and either 44100 Hz or 48000 Hz. With 44100 Hz, we get exactly 735.735 audio samples per field (see how 1001 affects that). With 48000 Hz we get exactly 800.8 audio samples per field (see how 1001 also affects that, too. If we had exactly 60 Hz, we would have exactly

735 or 800 samples per field. OTOH, exactly 120000 Hz as an audio sample rate will give a whole number of audio samples per field for all field rates (5005 at 23.976 fps, 5000 at 24 fps, 4800 at 25 fps, 4004 at 29.97 fps, 4000 at 30 fps, 2400 at 50 fps, 2002 at 59.94 fps, 2000 at 60 fps, as well as 1001 at 119.88 fps and 1000 at 120 fps for those shooting video in those special high frame rate cameras).

On Thu, 6 Mar 2008 21:48:14 +0000 (UTC) Dave Martindale wrote: | snipped-for-privacy@ipal.net writes: | |>Are you making that determination about debating style based on how I answer |>posts that fail to point out specific errors? You cannot conclude what the |>style would be for specific topical challenges when what you see is how I |>respond to personal attacks from people who have a history of doing so. | | I was remembering a discussion from a few weeks ago about 3 phase power | on utility poles. However, I now see that I confused you with "krw". | It was him who was being unpleasant to people (including me), *not* you. | | So my comments about your debating style are unjustified, and I | apologize for them. It's my fault for not checking who said what | before writing.

Accepted. N/P

| On the other hand, my more general point still stands. Because someone | (not you) was throwing around unjustified personal insults in a | previous discussion, I was reluctant to join this one. I imagine there | are other people with technical knowledge that also avoid contributing | because of the tone. Thus, it's naive to suggest that anyone who | disagreed with you would have posted an article saying that, and to | take the fact that no one disagreed with you publically as evidence | that no one disagrees. It only takes one unreasonable person to create | an unpleasant atmosphere.

What I guess I need to learn is to just "let them win" and move on. I would prefer that someone who finds something wrong in what I say go ahead and point it out. So finding some way to end the flame fests (that Usenet is infamous for) early would help.

Headaches? No. Minor problems, at best. You have to realize how primitive electronics was when NTSC was developed. Or they could have just kept TV monochrome, until digital technology was perfected. The development of NTSC video mandated compatibility with existing monochrome TV sets, and introducing a sync buzz wasn't acceptable. There was also a problem with chroma noise crawling up the screen if they kept the original timing.

Considering that frequency division was done with vacuum tube multi-vibrators, they had to keep the ratios as simple as possible. We are talking tubes like the 6SN7, octal based tubes. There were no fancy phase locked loop synthesizers back then.

| Considering that frequency division was done with vacuum tube | multi-vibrators, they had to keep the ratios as simple as possible. We | are talking tubes like the 6SN7, octal based tubes. There were no fancy | phase locked loop synthesizers back then.

OTOH, 63/88 would not have been hard to do, given the range of frequencies.

I don't know if they knew of this method, or not. But they most certainly did know the pre-requisites, which were to generate harmonics, filter a narrow band, do phase comparison, and work at UHF frequencies. What could be done is take the 63rd harmonic of 5 MHz (or 21st harmonic of 15 Mhz if that would be easier), and take the 88th harmonic of 3.579545 MHz (or the

22nd harmonic of 14.318182 MHz if they wanted that), bandpass filter them at 315 MHz, and do the phase comparison there. Then the phase error would be used to tweak the derived frequency oscillator.

An actual circuit design using tubes of the era is left as an exercise for the readers that really care.