'Simple' standards conversion - UK Vintage Radio Repair and Restoration Discussion Forum

Karen O · 23rd Jul 2011, 9:47 am

I'm given to understand that some 'simple' 625-405 standards converters (and possibly including the early BBC ones) work with limited line buffers and convert on a three-lines-in / two-lines-out basis.

That results in output lines that are about right but not spot on. In fact it yields a line frequency of 10.417kHz which is probably within the adjustment range of most TVs. But it aint standard.

Also, a 3-to-2 conversion produces slightly too many output lines, although with the higher line rate and the fact that the active lines of 625 are a smaller proportion of the frame time than 405, the picture is probably smaller in height than should be for a full spec 405 line picture.

More surprisingly, this 3-to-2 approach results in 416.67 lines not 405 lines (this is assuming the line frequency is not varied over the frame). At 208.33 lines per frame sync, that would result in a very strange triple interlaced output!

How do these simple converters achieve anything close to standard? Do they?

Of course, the 'proper' way to do it, and the way the Aurora does it I'm sure, is to convert on a 3.06-to-2 basis, meaning that the phase of the interpolation shifts from line to line. But that can be handled by changing the FIR coefficients according to the line. I guess a table of such coefficients can be held, indexed by output line number...?

Ray Cooper · 23rd Jul 2011, 10:31 am

Quote:

Originally Posted by Karen O

...How do these simple converters achieve anything close to standard? Do they?...

Well, for starters, they were not quite as simpla as a 3 to 2 ratio might suggest, and in fact weren't tied down to any particular ratio. Yes, their output was standard. Get one of the early reports on the subject:-

http://downloads.bbc.co.uk/rd/pubs/reports/1964-06.pdf

- and, if you've the stamina to wade through that lot, you'll know exactly how they did work...

ppppenguin · 23rd Jul 2011, 11:35 am

I don't think any of the 625>405 standards converters that have ever been available have used this simplified 3:2 method. The BBC CO6/501 and CO6/509 certainly didn't. If you haven't done so already, my first standards conversion article (in the publications section of my website) has a summary of conversion techniques together with links to most of the relevant BBC research papers.

Sample rate conversion, for this is what is required, can be done in various ways. http://en.wikipedia.org/wiki/Sample_rate_conversion Remember that this is done on the vertical axis. Most (all?) 625>405 converters leave the horizontal and time axes well alone. You can regard the 125:81 conversion ratio (yes I know the ratio of active lines is slightly different) as either as a proper fraction or an irrational number. Makes little difference. Ultimately it's a polyphase filter, the number of ocefficient sets is determined by how accurate you need the answer to be. For 625>405 conversion it's actually not too critical. BBC research found that beyond a 3 line interpolator (3 tap FIR) the improvement was negligible. The actual aperture shape isn't too critical either.

You can get really quite good results by a simple 2 line interpolator that takes either the whole of one line or a 50:50 mix of the 2 nearest lines. Works remarkably well, I think the Pineapple converter did this. The improvement over simply omitting approximately every 3rd line is huge.

tubesrule · 23rd Jul 2011, 1:38 pm

Karen,
When doing standards conversion, only the active portion of the video is considered. All of the VBI area in the input is ignored and the VBI in the output is generated from scratch. This means the output will always have the correct number of lines and timing. So for a simple 625>405 you are converting 288 lines to 189. 2/3 of 288 is 192 so you would need to cheat and drop a few lines off.

The other problem with this simple converter is 3 lines of 625 takes 192us while 2 lines of 405 takes 197.53us. This means the data is coming in slightly faster than its going out. At the end of a field you will have over 8 lines of input data stored up. This would be the minimum amount of memory required for such a system.

Converters like my Aurora do use a table of coefficients for the filter to achieve any arbitrary scaling ratio.

Darryl

Ed_Dinning · 23rd Jul 2011, 7:25 pm

Hi Karen, if you PM Fernseh he may be able to give you a primer on this method as he built his own converter many years ago, well prior to microprocessor control systems.
Ed

Karen O · 24th Jul 2011, 2:28 am

I am certain that 625-to-405 has been done in this simple fashion. I recall something about a discontinuity in the line syncs prior to the frame sync so as to avoid the wierd interlacing.

I guess it worked but there are much better ways of doing it these days.

It's quite staggering when you think about it - a conversion process that took racks and racks now fits in the palm of your hand. Amazing.

23rd Jul 2011, 9:47 am	#1
Karen O Rest in Peace Join Date: Jul 2011 Location: Bridgnorth, Shropshire, UK. Posts: 787	'Simple' standards conversion I'm given to understand that some 'simple' 625-405 standards converters (and possibly including the early BBC ones) work with limited line buffers and convert on a three-lines-in / two-lines-out basis. That results in output lines that are about right but not spot on. In fact it yields a line frequency of 10.417kHz which is probably within the adjustment range of most TVs. But it aint standard. Also, a 3-to-2 conversion produces slightly too many output lines, although with the higher line rate and the fact that the active lines of 625 are a smaller proportion of the frame time than 405, the picture is probably smaller in height than should be for a full spec 405 line picture. More surprisingly, this 3-to-2 approach results in 416.67 lines not 405 lines (this is assuming the line frequency is not varied over the frame). At 208.33 lines per frame sync, that would result in a very strange triple interlaced output! How do these simple converters achieve anything close to standard? Do they? Of course, the 'proper' way to do it, and the way the Aurora does it I'm sure, is to convert on a 3.06-to-2 basis, meaning that the phase of the interpolation shifts from line to line. But that can be handled by changing the FIR coefficients according to the line. I guess a table of such coefficients can be held, indexed by output line number...?

23rd Jul 2011, 11:35 am	#3
ppppenguin Retired Dormant Member Join Date: Dec 2003 Location: North London, UK. Posts: 6,168	Re: 'Simple' standards conversion I don't think any of the 625>405 standards converters that have ever been available have used this simplified 3:2 method. The BBC CO6/501 and CO6/509 certainly didn't. If you haven't done so already, my first standards conversion article (in the publications section of my website) has a summary of conversion techniques together with links to most of the relevant BBC research papers. Sample rate conversion, for this is what is required, can be done in various ways. http://en.wikipedia.org/wiki/Sample_rate_conversion Remember that this is done on the vertical axis. Most (all?) 625>405 converters leave the horizontal and time axes well alone. You can regard the 125:81 conversion ratio (yes I know the ratio of active lines is slightly different) as either as a proper fraction or an irrational number. Makes little difference. Ultimately it's a polyphase filter, the number of ocefficient sets is determined by how accurate you need the answer to be. For 625>405 conversion it's actually not too critical. BBC research found that beyond a 3 line interpolator (3 tap FIR) the improvement was negligible. The actual aperture shape isn't too critical either. You can get really quite good results by a simple 2 line interpolator that takes either the whole of one line or a 50:50 mix of the 2 nearest lines. Works remarkably well, I think the Pineapple converter did this. The improvement over simply omitting approximately every 3rd line is huge.

23rd Jul 2011, 1:38 pm	#4
tubesrule Hexode Join Date: Jul 2004 Location: Michigan USA Posts: 325	Re: 'Simple' standards conversion Karen, When doing standards conversion, only the active portion of the video is considered. All of the VBI area in the input is ignored and the VBI in the output is generated from scratch. This means the output will always have the correct number of lines and timing. So for a simple 625>405 you are converting 288 lines to 189. 2/3 of 288 is 192 so you would need to cheat and drop a few lines off. The other problem with this simple converter is 3 lines of 625 takes 192us while 2 lines of 405 takes 197.53us. This means the data is coming in slightly faster than its going out. At the end of a field you will have over 8 lines of input data stored up. This would be the minimum amount of memory required for such a system. Converters like my Aurora do use a table of coefficients for the filter to achieve any arbitrary scaling ratio. Darryl __________________ Aurora video standards converters: http://www.tech-retro.com/Aurora_Design/Video_Home.html

23rd Jul 2011, 7:25 pm	#5
Ed_Dinning Dekatron Join Date: Jan 2004 Location: Newcastle upon Tyne, Tyne & Wear, UK. Posts: 8,194	Re: 'Simple' standards conversion Hi Karen, if you PM Fernseh he may be able to give you a primer on this method as he built his own converter many years ago, well prior to microprocessor control systems. Ed

24th Jul 2011, 2:28 am	#6
Karen O Rest in Peace Join Date: Jul 2011 Location: Bridgnorth, Shropshire, UK. Posts: 787	Re: 'Simple' standards conversion I am certain that 625-to-405 has been done in this simple fashion. I recall something about a discontinuity in the line syncs prior to the frame sync so as to avoid the wierd interlacing. I guess it worked but there are much better ways of doing it these days. It's quite staggering when you think about it - a conversion process that took racks and racks now fits in the palm of your hand. Amazing.