XTAL requirements for exact PAL/NTSC color frequency?

ManAtWork

Hello,

I'm trying to get a working TV screen display with my current board. Unfortunatelly, I decided to use the same crystal as for my other project. It doesn't have·5.0MHz but 4.91MHz instead. I selected it because "it was there", it's smaller than a standard HC49-SMD package and I already have·a reel·in a feeder of my pick&place machine.

The problem is, I get a nice rainbow across the screen but no true colors. If I try to adjust the system clock settings say from _XINFREQ = 4_910_000 to 4_911_000 or 4_909_000 the display changes to black/white. So I think the TV standards require a very accurate color frequency to work properly.

Is there a chance to get it working with a XTAL frequency other than exactly 5.000MHz? Are there any restrictions for the PLL ratio so that some special frequencies don't work? If required I'll change the layout and buy some new crystals, but I don't like to if not really necessary.

Thanks for your help
·

Dr_Acula

That would be NTSC I presume, 'Never Twice the Same Color'

Leaving aside technical jokes, timing is more critical for NTSC than PAL but it is still critical for both. Can you get hold of some 5Mhz xtals? If you want some 'right now' I guess that might cost a little more. If you can afford to wait www.taydaelectronics.com/servlet/the-272/5.000-MHz-5-MHz/Detail for $0.23

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
www.smarthome.viviti.com/propeller

Leon

I have a little LCD monitor that can be switched between NTSC and PAL, we use PAL here in the UK. I get reasonable results with both NTSC and PAL using a 5 MHz crystal.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Leon Heller
Amateur radio callsign: G1HSM

Phil Pilgrim (PhiPi)

Try adjusting _xinfreq in smaller steps. Your crystal is no worse than a 5 MHz one for TV generation. The colorburst frequency is derived from a PLL in either case.

-Phil

ManAtWork

Hello,

now, I tuned the _XINFREQ in 1kHz increments and, surprise, at 4_858_000 it works perfectly. This means that my crystal is about 1% off what's completely out of the tolerance spec.· I can't measure the frequency directly because of the 10+pF load of the probe. It changes the frequency and the screnn turns black/white as soon as I touch the pin. Maybe it has something to do with the missing load capacitors but the propeller manual says the propeller can do without.

Anyway, I'll buy new crystals. Money is not the problem. As Dr_acula said, they cost only around $0.23 for small quanties. I just don't like the bigger package of the HC49/S crystals and the smaller ones are quite expensive. And I can wait, I could test the colors on the demo board.

BTW, this is what I do: another DRO for use with glass scales or incremental encoders.

Thanks for your comments

Phil Pilgrim (PhiPi)

If your frequency is off by that much, you could be using a series-resonant crystal, which will oscillate at a different frequency from that spec'd when used in a parallel-resonant circuit like the Prop's. Be sure to specify a parallel-resonant crystal with about 20pF load capacitance when you go to buy more of them.

-Phil

Toby Seckshund

You may be slightly confused as to the "freq being wrong". Studio spec was for the sub-carrier frequency to be within 1Hz ( the BBC used Rhubidium locked sources), OBs were allowed 5Hz (because of temperature variations etc) as they were re-syncronized anyway. TV sets·had fairly wide tollerances of a 100Hz or so. So your tweeking of the frequencies may be just matching up your Prop's Xtal with the TV's Xtal. I do it also, my homemade "DemoBoard" gives a better background (less beating) at 4.999500MHz ie 500Hz down on a Panasonic 9" monitor.

The spec of 4.43.....MHz was derived after long and difficult choices. The use of (1135/4 x H freq + 1/2 V freq) was a best fit kludge to put a sysyem as invisibly as possible over the existing monochrome one. It had to be as high as posible and yet be within the 5MHz bandwidth limitations ( in NTSC it has to even lower). The system was designed to give a minimally visible artifacts on a normal content, changing picture. Static text and other fine detailed lines were never catered for.

The formula gives a defined subcarrier to horizontal relationship that repeats itself over 8 frames. If you look at full broadcast spec colour bars the at the centre line the green to magenta transition (180 degree phase change) the chroma crawl will show itself, one you lock onto it. If you place a finger at the top and sweep it down over four seconds to the bottom you will see that you travel the same as the crawl. Back on your chair with a large area of colour this would not be seen.

The phase alternation was a slight improvement so that instead of phase variations giving colour variations (NTSC) by averaging over two adjacent lines this would result in colour saturation variations which the eye is less able to see. This does reduce the vertical resolution though, again back in your chair on a moving picture this would not be seen so much.

All of this was done back in the 1950s and 1960s, on a 20-26 inch screen with poor focus and size regulation, people were happy with it. Now with 50-60 inch screens (in smaller rooms) the problems are literally "in your face"

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Style and grace : Nil point