15.7 KHz RGBs to Component (YPbPr) schematic/IC?

[Jerry Lynds]

I am trying to find a schematic, or even a data sheet on an IC, that will
convert analog 15.7 KHz RGBs (Composite Sync) to component video. I find a
lot of IC's that will convert to composite video or Y/C but not toYPbPr.

There are a lot of transcoder/encoder boxes to buy, but I'd like to build my
own if it is cost effective and I can have some fun building it


Jerry Lynds
[email protected]

 

[[email protected]]

I am trying to find a schematic, or even a data sheet on an IC, that will
convert analog 15.7 KHz RGBs (Composite Sync) to component video. I find a
lot of IC's that will convert to composite video or Y/C but not toYPbPr.

There are a lot of transcoder/encoder boxes to buy, but I'd like to build my
own if it is cost effective and I can have some fun building it


Jerry Lynds
[email protected]

Hm, that's interesting. I am working on something very similar.

http://www.dfpresource.org/

This project takes 15.7 KHZ RGBI or CGA and outputs 31.5KHz analog VGA.
This is for old computers to use new monitors. One thing that came up
is that LCD TVs have component inputs and that they are compatible with
15KHz signals. This means it wouldn't take an external scan converter
anymore.

The operation to go from RGB to YUV is fairly simple to understand,
it's just a matrix. You need to look for colorspace converter chips but
the ones I've found are big, powerful and in crazy packages.

RGBI to YUV can be done in a handful of logic gates. RGB to YUV would
need a handful of video amps and mixers of some kind.

Hey why don't you email me and we'll see what we can come up with.

 

[Richard Crowley]

I am trying to find a schematic, or even a data sheet on an IC, that
will convert analog 15.7 KHz RGBs (Composite Sync) to component video.
I find a lot of IC's that will convert to composite video or Y/C but
not toYPbPr.

There are a lot of transcoder/encoder boxes to buy, but I'd like to
build my own if it is cost effective and I can have some fun building
it

Seems pretty simple to me. You aren't transcoding or
converting anything. You aren't changing the timings,
etc. You just need RGB to YPbPr.

You can look up the ratios of Y and Pb and Pr,
and do this in a simple matrix with a handful of
resistors and three op amps. Here is the formula:
http://en.wikipedia.org/wiki/YCbCr

I'd bet you a moderate-price dinner that you can
find the matrix schematic out there on the WWW.

 

[Jerry R]

I am trying to find a schematic, or even a data sheet on an IC, that will
convert analog 15.7 KHz RGBs (Composite Sync) to component video. I find a
lot of IC's that will convert to composite video or Y/C but not toYPbPr.

There are a lot of transcoder/encoder boxes to buy, but I'd like to build
my own if it is cost effective and I can have some fun building it


Jerry Lynds
[email protected]

If it's analog, you could just use some op-amp sum & difference circuits.
Calculate the resistor values from info on Charles Poynton's page:
http://www.poynton.com/ColorFAQ.html

Intersil has a DC restore chip, the EL4093... overkill for lo-res video, but
cheap, IIRC.

And the venerable old LM1881 sync separator for timing.

....jerry

 

[Jerry Lynds]

Thanks for your replies,



I have found a schematic here...

http://www.siliconchip.com.au/cms/A_101537/article.html


This takes RGB and converts it to YPbPr using op amps. However, this does
not use the composite sync as input...only the RGB...how does the video get
its timing? I know the R,G,B is combined to get a total luminance (Y),
which in svideo and component contains the sync as well...but how would
taking just R.G,B without sync work? Does each color component (R,G,B)
contain its own fragment of timing, so that when combined to form Y it also
contains all the sync too? The above link is for a pay and get schematic so
I can't just post it...but it does all make clear sense to me...just not how
it can make a Y that conatins sync without taking sync (composite in my
case)as input....

Jerry

 

[Jerry Lynds]

Sorry...wrong link sent before....


http://www.siliconchip.com.au/cms/access/purchaseRequired.html?buy=A&id=102648

 

[petrus bitbyter]

Thanks for your replies,



I have found a schematic here...

http://www.siliconchip.com.au/cms/A_101537/article.html


This takes RGB and converts it to YPbPr using op amps. However, this does
not use the composite sync as input...only the RGB...how does the video
get its timing? I know the R,G,B is combined to get a total luminance
(Y), which in svideo and component contains the sync as well...but how
would taking just R.G,B without sync work? Does each color component
(R,G,B) contain its own fragment of timing, so that when combined to form
Y it also contains all the sync too? The above link is for a pay and get
schematic so I can't just post it...but it does all make clear sense to
me...just not how it can make a Y that conatins sync without taking sync
(composite in my case)as input....

Jerry

Hmm... IMHO this circuit converts YPbPr to RGB. Apparently Y is supposed to
carry the sync. In this case the sync has is own output. I saw RGB signals
in which one of the colors (don't remember which) carries the sync. For that
way separating the sync from a composite video signal is the easiest part.
It can be done by some old fashioned discretes but the venerable old LM1881
has also been mentioned already.

petrus bitbyter

 

[Jerry Lynds]

That circuit I was looking at must have been for a sync on green (RGsB)
input...crap...

 

[Richard Crowley]

Thanks for your replies,
I have found a schematic here...

http://www.siliconchip.com.au/cms/A_101537/article.html


This takes RGB and converts it to YPbPr using op amps. However, this
does not use the composite sync as input...only the RGB...how does
the video get its timing? I know the R,G,B is combined to get a total
luminance (Y), which in svideo and component contains the sync as
well...but how would taking just R.G,B without sync work? Does each
color component (R,G,B) contain its own fragment of timing, so that
when combined to form Y it also contains all the sync too? The above
link is for a pay and get schematic so I can't just post it...but it
does all make clear sense to me...just not how it can make a Y that
conatins sync without taking sync (composite in my case)as input....

You only asked about the video signals, not the sync signals.
Impossible to answer your question without some real-world
context....
Does your source have separate H & V sync?
Composite sync? Sync on green?

Which type of sync does your destination want?
Any combination can be converted to any other,
but you have to specify exactly what you need.

 

[xray]

You only asked about the video signals, not the sync signals.
Impossible to answer your question without some real-world
context....
Does your source have separate H & V sync?
Composite sync? Sync on green?

Which type of sync does your destination want?
Any combination can be converted to any other,
but you have to specify exactly what you need.

Or at least what the source and target hardware are. I think there are a
few variations of RGB and YPbPr. How the sync is encoded is the most
likely variable thing.

 

[jasen]

Hm, that's interesting. I am working on something very similar.

http://www.dfpresource.org/

This project takes 15.7 KHZ RGBI or CGA and outputs 31.5KHz analog VGA.
This is for old computers to use new monitors. One thing that came up
is that LCD TVs have component inputs and that they are compatible with
15KHz signals. This means it wouldn't take an external scan converter
anymore.

some CGA cards had luminance/chrominance outputs, they could probably go
straight to a "svhs" input.

I once hooked the lumiance to the composite in of a VCR and got a (monochrome)
image on a TV.

but why would anyone want to keep a CGA card when a VGA card
is backwards compatible ?

I have one somewhere that can even drive a CGA monitor....

Bye.
Jasen

 

[Poxy]

Thanks for your replies,



I have found a schematic here...

http://www.siliconchip.com.au/cms/A_101537/article.html


This takes RGB and converts it to YPbPr using op amps. However, this does
not use the composite sync as input...only the RGB...how does the video get
its timing? I know the R,G,B is combined to get a total luminance (Y),
which in svideo and component contains the sync as well...but how would
taking just R.G,B without sync work? Does each color component (R,G,B)
contain its own fragment of timing, so that when combined to form Y it also
contains all the sync too? The above link is for a pay and get schematic so
I can't just post it...but it does all make clear sense to me...just not how
it can make a Y that conatins sync without taking sync (composite in my
case)as input....

I've never looked closely at TV RGB, but I think in the absence of separate
sync (ie. composite) sync is often in the green channel.

 

[Jerry Lynds]

I did mention in my first post I was using RGBs...analog 15.7 KHz rgb with
composite sync as output. This needs to be converted to YPbPr for connection
to a TV with component video. I pretty much just embarrassed myself with my
last post about not uderstanding how the sync was getting into the Y...the
circuit I was looking at must have been for a sync on green rgb input. That
circuit was here...

http://www.siliconchip.com.au/cms/access/purchaseRequired.html?buy=A&id=102648

But anyway...that circuit was helpfull in my understanding of how to use op
amps at matrix the rgb to get Y, R-Y and R-B....but I still need to
introduce sync...would another op amp do this to add sync to the green
before R, G, B get combined to create a Y = 0.3R + 0.59G + 0.11B signal? Or
even after?



Jerry

 

[Mark Robinson]

I did mention in my first post I was using RGBs...analog 15.7 KHz rgb with
composite sync as output. This needs to be converted to YPbPr for connection
to a TV with component video. I pretty much just embarrassed myself with my
last post about not uderstanding how the sync was getting into the Y...the
circuit I was looking at must have been for a sync on green rgb input. That
circuit was here...

http://www.siliconchip.com.au/cms/access/purchaseRequired.html?buy=A&id=102648

But anyway...that circuit was helpfull in my understanding of how to use op
amps at matrix the rgb to get Y, R-Y and R-B....but I still need to
introduce sync...would another op amp do this to add sync to the green
before R, G, B get combined to create a Y = 0.3R + 0.59G + 0.11B signal? Or
even after?

Elantec used to make some nice video sync separators, it seems Intersil have
bought them.

You could try looking at http://www.intersil.com/video/ but the site is broken
beyond usability in my browser. It does seem that they do samples though.

 

[Richard Crowley]

"Jerry Lynds" wrote ...

I did mention in my first post I was using RGBs...
analog 15.7 KHz rgb with composite sync as output.

OK, so you are saying that your source equipment has four
coax outputs: R,G,B, and composite (H+V) sync. Right?

This needs to be converted to YPbPr for connection
to a TV with component video.

Sorry, still not clear exactly what your destination
equipment wants. Are you saying that it wants sync
on Y? Separate composite sync? What, *exactly*?

How many connectors does it have? How are they
labeled? You are unlikely to get any useful specific
advice with only generic descriptions of your situation.

OTOH, you could identify the source/destination
equipment by make and model and make everything
much easier to decode here.

 

[Jan Panteltje]

I did mention in my first post I was using RGBs...analog 15.7 KHz rgb with
composite sync as output. This needs to be converted to YPbPr for connection
to a TV with component video. I pretty much just embarrassed myself with my
last post about not uderstanding how the sync was getting into the Y...the
circuit I was looking at must have been for a sync on green rgb input. That
circuit was here...

http://www.siliconchip.com.au/cms/access/purchaseRequired.html?buy=A&id=102648

But anyway...that circuit was helpfull in my understanding of how to use op
amps at matrix the rgb to get Y, R-Y and R-B....but I still need to
introduce sync...would another op amp do this to add sync to the green
before R, G, B get combined to create a Y = 0.3R + 0.59G + 0.11B signal? Or
even after?

After.
So once you have Y, and say it is 700mV pp, add 300mV composite neg sync.
This then gives 1Vpp, and you need to drive into 75 Ohm likely, so you may
want to go for 2V pp low impedance out with a 75 Ohm resistor.
Usually works fine with some transistors, emittor followers, no ICs needed.
I hope your TV is also 15.7 kHz, and not a VGA monitor.
If it is a VGA monitor then you need to convert to double the 15.7kHz for
example, and that is a whole other story, I have done that with DA / AD and
FPGA, so I could play the old DVD player on the 19inch monitor.

 

[petrus bitbyter]

After.
So once you have Y, and say it is 700mV pp, add 300mV composite neg sync.
This then gives 1Vpp, and you need to drive into 75 Ohm likely, so you may
want to go for 2V pp low impedance out with a 75 Ohm resistor.
Usually works fine with some transistors, emittor followers, no ICs
needed.
I hope your TV is also 15.7 kHz, and not a VGA monitor.
If it is a VGA monitor then you need to convert to double the 15.7kHz for
example, and that is a whole other story, I have done that with DA / AD
and
FPGA, so I could play the old DVD player on the 19inch monitor.

Jan,

Can you tell more about that conversion? I've got questions about an old
monochrome monitor (15.67kHz) output to be connected to a VGA monitor.

petrus bitbyter

 

[Jan Panteltje]

Jan,

Can you tell more about that conversion? I've got questions about an old
monochrome monitor (15.67kHz) output to be connected to a VGA monitor.

petrus bitbyter

Hi, Petrus, yes, I used a Spartan2-200 FPGA, and 2 Philips ADC converters,
lemme look (long time ago) TDA8708 8 bit flash.
I did the BW only, but worked out the color, you can sample Y at any speed
you like, and use a 2 to 1 CMOS 4053 and sample U and V sequentially at half
that speed each.
If you sample low freq you just get lower resolution ;-)
Then you need 3 DA 8 bit converters, and I used r2r ladders.
I used the FPGA dual port block RAMS (these allow read and write at the same
time), to store a line in 64 uS, and then clock it out twice at double speed,
would have to look up the Verilog code.
The most challenging was to do a digital PLL 2x to multiply the H.
(Jitter was a problem especially if fed from a VCR source).
It also worked as a timebase corrector, but not 100% good, did not finish it.
Had a sync separator IC first, but then just clamped the Y bottom sync with
a simple diode and did the sync slicing in Verilog.
Was just an experiment on the Digilab board.
Doing it in HDL makes things more simple to change.
When using a TV signal as source, as the OP does, then 1024 samples per line
should be enough.
That fits nicely in a Spartan2 block-RAM.
I have tried several sample frequencies and configurations.
You need some input filters if you think you have high frequency components in
the input, and a low pass after the DA (R2R with high R is low pass enough ;-).
1024 samples with 15625 Hz gives you 16MHz sample freq, and limits Fmax video to
below 8 MHz, filter at say 6.
If you use 2 4 bit block RAMs you get 8 bits 1024 sample RAMS to play with.
I just looked up some HDL code I wrote, seems I did it that way.

Actually I also used color in the timebase corrector, with composite in and out
with this setup... color is tricky as your time resolution becomes critical
(subcarrier phase), it is always once clock, even as I used the internal 4x FPGA
clock multiplier to go to 200MHz...
PAL is forgiving...
Finally played the DVDs in the PC .

It is a fun project, maybe I should play with it again some day, but all went digital....
satellite...

If you need code or diagrams maybe I can scan something in.

 

[Jerry Lynds]

"Jerry Lynds" wrote ...

OK, so you are saying that your source equipment has four
coax outputs: R,G,B, and composite (H+V) sync. Right?

Actually it is a multi-pin din....and yes, it has R,G,B and composite (H+V)
sync.

I apologize if I was being vague...I thought that was a pretty good
description Anyways, the source is older video game systems such as a
Sega Genesis or Sega Saturn. Pin outs here...
http://www.gamesx.com/avpinouts/genesisav.htm Oddly enough I created these
diagrams about 12 years ago when hooked up to my Comodore 1084D RGB monitor.
That monitor, as input, required RGB Analog 0.7 P-P, 75 Ohm. The cables I
built to drive the 1084 were straight through except a resistor of 470 Ohmn
on the composite sync. I am only mentioning the input of the 1084 to
possibly better describe the output of these sega systems.

Sorry, still not clear exactly what your destination equipment wants. Are
you saying that it wants sync
on Y? Separate composite sync? What, *exactly*?

Component video found on every DVD player as output, and found as input on
pretty much any brand name TV over 20" in the past year....

How many connectors does it have? How are they
labeled? You are unlikely to get any useful specific advice with only
generic descriptions of your situation.

It just has 3 RCA connectors labeled Y, Pb and Pr. I am assuming sync is on
the Y...as this circuit to convert from Component (YPbPr) to RGB uses a
LM1881 to remove the sync from Y

OTOH, you could identify the source/destination
equipment by make and model and make everything
much easier to decode here.

I think this is starting to be more trouble than it is worth...

 

[petrus bitbyter]

On a sunny day (Fri, 27 Oct 2006 19:56:08 +0200) it happened "petrus
bitbyter"


Hi, Petrus, yes, I used a Spartan2-200 FPGA, and 2 Philips ADC converters,
lemme look (long time ago) TDA8708 8 bit flash.
I did the BW only, but worked out the color, you can sample Y at any speed
you like, and use a 2 to 1 CMOS 4053 and sample U and V sequentially at
half
that speed each.
If you sample low freq you just get lower resolution ;-)
Then you need 3 DA 8 bit converters, and I used r2r ladders.
I used the FPGA dual port block RAMS (these allow read and write at the
same
time), to store a line in 64 uS, and then clock it out twice at double
speed,
would have to look up the Verilog code.
The most challenging was to do a digital PLL 2x to multiply the H.
(Jitter was a problem especially if fed from a VCR source).
It also worked as a timebase corrector, but not 100% good, did not finish
it.
Had a sync separator IC first, but then just clamped the Y bottom sync
with
a simple diode and did the sync slicing in Verilog.
Was just an experiment on the Digilab board.
Doing it in HDL makes things more simple to change.
When using a TV signal as source, as the OP does, then 1024 samples per
line
should be enough.
That fits nicely in a Spartan2 block-RAM.
I have tried several sample frequencies and configurations.
You need some input filters if you think you have high frequency
components in
the input, and a low pass after the DA (R2R with high R is low pass enough
;-).
1024 samples with 15625 Hz gives you 16MHz sample freq, and limits Fmax
video to
below 8 MHz, filter at say 6.
If you use 2 4 bit block RAMs you get 8 bits 1024 sample RAMS to play
with.
I just looked up some HDL code I wrote, seems I did it that way.

Actually I also used color in the timebase corrector, with composite in
and out
with this setup... color is tricky as your time resolution becomes
critical
(subcarrier phase), it is always once clock, even as I used the internal
4x FPGA
clock multiplier to go to 200MHz...
PAL is forgiving...
Finally played the DVDs in the PC .

It is a fun project, maybe I should play with it again some day, but all
went digital....
satellite...

If you need code or diagrams maybe I can scan something in.

Thanks Jan. It's about a monochrome display mainly used for 80 columns
characters and only two steps of brightness except black. It can do graphics
as every dot is programmable but it is slow. So AD will be overkill. I think
some comparators will do. Problem may be how to synchronise the sampler with
the dotclock. If you have material about the storage and recovery of the
samples and the way you treated the sync I should be glad to receive it.

BTW My e-mail address is valid once you've removed the obvious additions.
Petrusdotbitbyterathotmaildotcom works as well.

petrus bitbyter