Atari A/V Mods - Quest for Perfection
Status: 2600 V2.2G - Final
5200 V2.1 - Redesign
7800 V2.1A - Final
The schematics, guides, and other stuff related to the Atari Video mods can be found here.
The Story
It all started a few years ago after I finished my first portable, the monstrous NESpV1. I received a few emails asking me if I could perform a simple video mod to make a Atari 2600 output S-Video and Composite. I accepted and finished the mod. When I tested it I could not believe how clear the image was compared to the stock RF.
A few Ataris later I decided to make a bunch of PCBs with the standard CD4050 mod. I had great success with this mod and was pleased at its performance. This was Version 1 of my Video mods.
The mod supported all Atari consoles and even Atari computers. A few months later I started receiving reports from my costumers that they where experiencing vertical lines much like scan lines on there LCD TVs. Also the Composite image was much to be desired. So I set out to figure out why.
The Quest
Well I sure didn't own all the systems I need to test on so I opened up free Video modding service to one of each kind of Atari. I decided to first tackle the Composite image. I decided to use a S-Video filter chip, the FMS6400, to do this. The chip has inputs for S-Video and outputs filtered S-Video and Composite signals. Pleased at the chips performance. Well that was easy I thought and promptly moved on to tacking the vertical lines.
[Prototype of Atari Video mod V2.0.]
Vertical lines seemed to have plagued the S-Video mod since it was first used. Some Tvs and games exhibit these lines more then others (Pole Position and KaBoom! come to mind) and there didn't seem to be any real connection between games that had them. Further more some Ataris had it worse then others. The Heavy Sixer models had light vertical lines but had more bleed. 4 Switchers had heavier lines but the least amount of blur. Juniors fell in the middle range. 7800's and 5200's where all over the place. Well I tried to figure out why this was the case.
Well maybe it is TIA quality control? When the 4 Switchers where being made they where made cheaply compared to the Heavy Sixers. A quick swap of TIA chips threw out this theory.
Comparing the schematics and board layouts of the Heavy Sixers, 4 Switchers, and Juniors I realized that there are quite a number of different ways Pin 6 and 9, the blank and color signals, are handled by the video mixers.
This diagram shows the different color circuits as well as my color part of V2.1 of my mod. To come up with my idea I looked at Pin 6, the usually unused blank signal. This signal carries the horizontal blanking signal which tells the TV how to determine the correct DC levels for the AC-coupled video data (ripped from AJs post on Atariage. Looking back at the schematics I noticed that the Ataris that use there Blank signal had fainter vertical lines then ones that did not.
So I wired the blank into the mod just like the Heavy Sixer. Well the picture looked, unsurprisingly, just like the Heavy Sixer image. Still the image had those faint lines and bleed. Looked at the diagram again and noticed that the blank signal passed through the CD4050 that was on the board before hooking up with the color. I tried this and eureka! Little vertical lines and no blur!
Still not perfect however. I started tweaking the color circuit and by shear luck I bridged the 47pF cap with a 680ohm resistor and it eliminated all bleed and vertical lines. I am confused on why this is (no oscilloscope so all my experiments are my trial and error) as I have not progressed to video circuits in my degree yet.
1/07/2009 - 2600 Mods done, 7800 and 5200 in works
Well right now I am having preorders over on AtariAge and through this site. I decided from the beginning that I would offer my mod design and schematic free to anyone that would like to build one themselves. V2.2B supports stereo sound and has the latest mod design.
Still working on the 5200 and 7800 version since I figured out how to remove the vertical lines on the 2600.
1/10/2009 - 2600 V2.2C
Had to change some layout and fix a floating input on the FMS6400. Was pretty easy to do.
1/18/2009 - 2600 V2.2D
I moved from ExpressPCB to KiCAD. It is open source and a non proprietary and is easy for me to make custom parts.
I did some digging around on the Atari 2600 mod V2.2C and realized that the Luma signal is delayed by 140ns because of the buffer chip! This is probably the reason why the composite image created by the FMS6400 had allot of bleed. I am guessing this delay caused a problem with the combining of the two signals. Reading a couple articles a delay of over 20ns causes issues with video quality. I found a replacement part and will be ordering it to replace the other Cd4050s. It is a direct drop in so no change to the layout of the board is needed.
CD74HC4050E - Gate speed is 16ns @ 5V.
I will be testing this theory on my Atari by ramping up the voltage to 15V for the CD4050. According to the spec sheet. The higher voltage will decrease the gate delay to 40ns. If there is an improvement in the image I will be changing the partslist and schematics to reflect this change.
V2.2D of the Mod will include this change.
2/12/2009 - 7800 Chroma Dilemma
Been kicking the 7800 version of the mod around. Seems the biggest issue is separating the 7800 and 2600 chroma lines from each other. The Luma lines go through a NOR gate (U3). I am assuming the NOR gate separates the 7800 and 2600 lines.
When the 7800 mode is on what is being sent by the TIA? 0 or 1? When the 2600 mode is on what is being sent by the MARIA? 0 or 1?
If it is a NOR gate then the "inactive" mode has to be set to low to even get a difference in the output but it is inverted. I guess I can head to the EE lab on Friday (have those off!) and figure it out if no one knows.
Then I have to figure out how to detect when the board is in 7800 mode. There has to be a signal somewhere that is high when 7800 mode is enabled and low when 2600 (or vise versa?). If I can figure this out then it will be pretty easy to let the Mod know when to switch to 2600/7800 chroma line.
The first challenge was to find a signal that could be used to switch the chroma lines. Someone on the boards suggested I look at pins 2 and 5 of the U11 chip. After a trip to the EE lab I found that pin 5 was high it was in 7800 mode and when low it was in 2600 mode. Using this I can switch between chroma lines. I also found out that the U3 chip is a OR gate not a NOR.
Using the TI CD74HCT4053. It is a bit overkill but its in DIP format and can handle +5V.
2/19/2009 - 2600 V2.2E
Had the Aud L and Aud R silk screen reversed. Fixed in this edition.
5/28/2009 - 7800 Chroma
I figured out the 7800 issue with the chroma. Using pin 5 on the U11 chip and the TI TS5A3159ADCKR chip (a analog multiplexer) I can have the mod auto determine the 2600/7800 mode. The 7800 chroma circuit needs to be worked on cause im still getting vertical lines there. In version 1.0 for the 7800 I got fairly close so I am going to look at that design and see if it can be improved. I will update with some pictures in a couple days of the mod. Oh and the PAL 2600 is coming along.
5/28/2009 - MIDI 2600
Been working on this puppy for awhile now. This is a Heavily modded Atari 2600 Heavy Sixer. It is modded with the Longhorn Engineer V2.2E S-Video mod and the MIDI2600 mod. The Video mod enables it to output both S-Video and Composite Video as well as splitting the Audio Signal into two distinct Left and Right channels. The MIDI2600 mod allows the user to plug in a MIDI keyboard to control the Atari 2600. These coupled with the Synthcart create a fairly unique 8-Bit "musical" device.
Highly Liquid MIDI2600: http://highlyliquid.com/kits/midi2600/
Synthcart: http://www.qotile.net/synth.html
12/02/2009 - 7800 Prototype
Been working on the 7800 V2.0 video mod. So far I got rid of the vertical lines that plagued V1.0 and now I am just working on removing the red bleed but I have a fairly decent idea what that is already.
12/03/2009 - More background information
By the way for those that want to read up on luma/chroma delay and signals there is a nice article here.
Figure 3 is what is happening with the 7800 mod and what used to happen with the 2600 mod.
Also this
Ok if I use some SOIC chips I can get the Luma delay to around 11ns which is within the spec needed.
SN74LVC07AD - Buffer Chip SN74F02DG4 - NOR Chip
No DIP chips have delays that are short enough. I think I am going to change the 2600 mod to use that SOIC chip as well just to cut down on the amount of different parts I have to buy.
Also, I am planning on having the 7800 mod install just like the 8bit domain mod. Remove that one chip and solder in a socket.
01/13/2009 - 7800 PCB
PCBs are made. Just a small error on the silkscreen, nothing that affects performance.
01/14/2009 - 7800 V2.1A will be the final
Installed into a 7800. V2.1A only differs from V2.1 in that the silk screen issue is fixed.
01-29-2010 - Atari 2600 A/V Mod Speed Build Video
A speed up video of me building a Atari 2600 Video Mod. A little blurry but it gets the job done. Need a better camera for close ups I think.
Music: Johnny and the Knuckleheads - Cadillac
The band, Johnny and the Knuckleheads was my Great Uncles back in the early 1980's. Have most of there stuff on LP.
I just sourced an Atari 5200 so the mod is very close to being done on the design side. Just need to take final measurements to ensure the mod does not bump any thing and to finalize the schematic.
