Help with AD724 RGB > Composite / S-Video

[mbarszcz]

I'm working on an RGB to analog video encoder for a Sega Genesis. 

I now know that many people before me have had compatibility problems with the AD724 and the RGB output of the Genesis, but forum posts on Sega16 and Assembler Games haven't really yielded any technical discussion or solutions other than "it doesn't work" or "the colors are wrong".  I haven't tried my exact circuit on other inputs, but the consensus from many other people trying something similar before me is the AD724 works fine for most RGBS sources...except the Sega Genesis for some unknown reason.

I've gone through the datasheet and read what other people have tried, and incorporated that into my design, but unfortunately with limited results.

As a little background, the Sony CXA1145 is the original video encoder in the Sega Genesis Model 1.  My project is to try to use a different encoder to help improve the composite and s-video output.  I already have an "ArcadeTV RGB Bypass" installed in the board that taps the R,G,B and CSYNC signals directly from the video processor, buffers RGB them using a THS7316, and has a few resistors to bring down the level of the TTL sync signal. 

I'm trying to do something similar, but in the composite video world.  My schematic pretty much follows the AD724 datasheet, with the exception that the Fsc clock (3.58Mhz Chroma Subcarrier) comes from the system board as a TTL level signal, and there are voltage dividers on the RGB inputs to bring the TTL RGB levels down to the 714mV the inputs expect.  I made sure to AC couple the inputs and outputs and the TTL Sync feeds hsync directly (while pulling up vsync to 5V to indicate HSYNC is a CSYNC signal).  There is a pullup resistor elsewhere on the board, so a pullup of the CSYNC shouldn't be necessary.



The problem with the output is the is the hues are totally shifted around.  Red comes out green, green comes out blue, blue comes out red.  And yes, I am sure the inputs are going to the correct pins. 

When i look at the output with my scope and compare with the original output,

,

I can see that the AD724 is not inserting the color burst at the correct time, which I believe is the cause of the colors being way off.  It is way late and very weak.  The fact that the monitor locks onto it at all is surprising.  The yellow trace is the composite output from the original CXA1145 video encoder, and the pink is the composite output from the AD724.  The different output voltages are because the AD724 is connected to a monitor with 75 ohm termination and the CXA1145 is just floating, but it makes no difference in the colorburst issue.

Here is the csync and AD724 Composite output:


and here is the csync and CXA1145 Composite output:


Any help or thoughts as to why the AD724 isn't generating a proper color burst with the Sega Genesis RGB output would be greatly appreciated.

Matt

[mbarszcz]

When looking at the lines in the beginning of the video field, there appears to be a normal colorburst after the sync pulse (~3.2μs):



at line 10 though, the length of the sync pulse gets longer (~6us), which seems to cause the colorburst signal to to be delayed and missing the bottom half of the waveform.



Then once the actual image starts, the late and malformed colorburst gets the end chopped chopped off by the start of the video signal.



In the datasheet, the AD724 specifies this as the timing specs:



Perhaps the "Sega Genesis' 6μs" horizontal sync pulse is what causes the AD724 to get confused and not produce a properly timed and formed colorburst?