Getting S-Video from the TI console

[+OLD CS1]

Okay, kids. Tonight I decided I would implement both the resistor and capacitor video fixes on my console. While I was in there I decided to grab the schematics to try to determine some good places to put the capacitor inside the console. This was a mistake which lead to over an hour of soldering and wiring, and missed sleep and cleaning.

 

While looking at the video circuit, I noticed an option for B-Y and R-Y video signals to exit the console, but the default strap selection is to ground those ports instead. (I also noticed that R212 of the resistor fix is supposedly an option, as well as an option to completely bypass the composite amplification circuit.)

 

According to the schematics in my "TI-99/4A Console and Peripheral Expansion System Technical Data" book, U100, the TMS-9918A, pins 35 and 38 output the B-Y and R-Y signals. The TI Tech Pages' TMS9918 reference page indicate that in the 9918A, these are CPUCLK and EXTVDP, respectively. I have great faith in Theirry's technical information, though I proceeded based upon the schematics none the less.

 

After replacing R212 with the 330ohm resistor per the Mainbyte resistor fix article, I also put matching resistors in the missing R210 and R211 locations, which both indicate 560ohm, the same as R212.

 

I also decided to connect the .0047uF capacitor to the unused shunt W205 leg which connects to video port pin 4, as well as to the grounded leg of R212, based upon the Mainbyte capacitor fix article, which shows the capacitor across the video output to ground.

 

Then I took an S-Video cable and grounded the red and black wires to a grounded strap (W201/202,) the white wire ( C ) to the combination of the two lines from U100 pin 35 and 38 (this page says C is B-Y combined with R-Y,) and the green wire ( Y ) to the leg of added capacitor closest to U100's pin 36.

 

The pictures attached are the results when connected to the S-VHS input of my monitor. Notice the odd coloring and pixelation. As well, these show off some of my better and poorer soldering skills (it is late and I should be cleaning up the house for tomorrow's events. Whoops.)

 

The system does not seem to suffer any adverse effects from combining what could be CPUCLK and EXTVDP. Anyone have any feedback on this? To me, it looks like the monitor is getting the Y (luma) signal but no working C (chroma.)

 

Oh, I also found an error in Mainbyte's motherboard graphic regarding the traces from pins 35 and 38 to the shunt vias.

Edited November 25, 2010 by OLD CS1

[+retroclouds]

Nice project. Thanks for sharing

[+OLD CS1]

Thanks. But, to my dismay, after taking a closer look at the schematics after having woken up today, I realize that it shows TWO pin 35s and 38s. One set is grouped in with the video circuitry, and the other is part of the host interface. The host interface set of the pins shows pin 35 as EXTVID with no connection and pin 38 connected to SGCCK and an apparent shunt option to ground.

 

I cannot answer as to why the TI manual would show configurations for multiple versions of the video chip and not make these clear -- the chip is labeled "U100, TMS 9918A," with no references to the 9928A which would be the chip required for this project to be successful.

 

So, some lost time in the project itself. But it has served to quell the project demon in me for at least a while and, if I say so myself, my soldering job looked damn good. Now I must go clear off the dining room table before I find myself in more trouble.

 

EDIT: Okay, I may have stumbled across some hope here. The QI schematics show only one set of pin 35 and 38, and these are the expected video pins. However, the schematic shows the 9918A. Does anyone have a picture of the QI motherboard which shows the VDP?

Edited November 25, 2010 by OLD CS1

[matthew180]

The 9918A does not have the component output. You need a 9928A. See pages 3-9 and 3-11 in the TMS9918A datasheet:

 

9918A:

Pin 35: EXTVDP - External VDP input

Pin 36: COMVID - Composite video out

Pin 38: CPUCLK - Not used in the 9918A or the MSX, not sure about the Coleco

 

9928A/9929A:

Pin 35: B-Y - B-Y color difference

Pin 36: Y - black/white luminance and composite sync

Pin 38: R-Y - R-Y color difference

 

I have some 9928A chips if you want them. Tursi gave them to me last year at the Faire, but since I'm making the F18A I don't need them any more.

 

Matthew

[Tursi]

I have some 9928A chips if you want them. Tursi gave them to me last year at the Faire, but since I'm making the F18A I don't need them any more.

 

I tried to give more away this year.. nobody would take them.

 

I did successfully drop a 9928 in my US console, and ran the output of that into a component->VGA converter - this did give me a picture that was not horrible. But I moved away from that being the best answer due to the cost of the converters, eventually deciding that a new replacement VDP was the way to go. And Matthew's taking care of that wish.

 

No photo unfortunately. I remember thinking there was still some ghosting though, which annoyed me, though the picture was definately better. That was with no balancing circuitry in the middle, with some proper parts you can get quite a nice picture. Search AtariAge, because someone here actually documented their path down that same road about two years ago.

[+OLD CS1]

I have some 9928A chips if you want them. Tursi gave them to me last year at the Faire, but since I'm making the F18A I don't need them any more.

 

I tried to give more away this year.. nobody would take them.

 

I would not mind having a few on-hand, actually. I am more of a fan of the S-Video interface for use with entertainment systems. Matt's going to send me a couple, so if you have a couple more, I do not think it would be too wasteful to have them hanging around here.

 

I did successfully drop a 9928 in my US console, and ran the output of that into a component->VGA converter - this did give me a picture that was not horrible. But I moved away from that being the best answer due to the cost of the converters, eventually deciding that a new replacement VDP was the way to go. And Matthew's taking care of that wish.

 

I am really excited about the F18A project and am waiting patiently. Your problem with the component video converter; I read somewhere that R-Y and B-Y are not REALLY component signals...?

 

Oh oh oh... speaking of cool 9918 tricks, do I recall correctly someone lamenting that "Zaxxon" was never released for the TI because of some lame-ass excuse about the graphics? Well, I turned up a ColecoVision "Zaxxon" cartridge in my storage room. (Not to mention "Mr. Do!") How difficult would it be to take the video data from the cart and convert the Z80 code to 9900? (The CV is Z80, yeah? Too lazy to WikiGoogle it right now.)

[matthew180]

Before I started the F18A, I looked into every other form of getting a better signal from the 99/4A. The problem with the S-Video, Component, Pb,Pr,Y, or what have you, is the analogue electronics is beyond my current electrical engineering skills. The conversion from one video / color space to another is very complex, and any drop in 9918A solutions required some form of modification to the motherboard, which is beyond most of the people using the computers today. Even people who can solder like a drop in solution.

 

I think the biggest problem you are going to have with using a 9928A is that, you can't just take the output from the Pb, Pr, and Y pins and run them to the video monitor. You need active filtering, amplification, and impedance matching.

 

Matthew

[+retroclouds]

Oh oh oh... speaking of cool 9918 tricks, do I recall correctly someone lamenting that "Zaxxon" was never released for the TI because of some lame-ass excuse about the graphics? Well, I turned up a ColecoVision "Zaxxon" cartridge in my storage room. (Not to mention "Mr. Do!") How difficult would it be to take the video data from the cart and convert the Z80 code to 9900? (The CV is Z80, yeah? Too lazy to WikiGoogle it right now.)

 

I don't see an immediate reason why a "Zaxxon" cartridge wouldn't be possible on the TI.

But it's a hell of job though. Basically you'd have to rewrite "Zaxxon" from scratch in TMS9900 language. Ofcourse you can "borrow" the colecovision graphics

and sounds. But that's about it.

[+OLD CS1]

Oh oh oh... speaking of cool 9918 tricks, do I recall correctly someone lamenting that "Zaxxon" was never released for the TI because of some lame-ass excuse about the graphics? Well, I turned up a ColecoVision "Zaxxon" cartridge in my storage room. (Not to mention "Mr. Do!") How difficult would it be to take the video data from the cart and convert the Z80 code to 9900? (The CV is Z80, yeah? Too lazy to WikiGoogle it right now.)

 

I don't see an immediate reason why a "Zaxxon" cartridge wouldn't be possible on the TI.

But it's a hell of job though. Basically you'd have to rewrite "Zaxxon" from scratch in TMS9900 language. Ofcourse you can "borrow" the colecovision graphics

and sounds. But that's about it.

 

Well...? Give me your address and I will send you the cart Of course, for that matter, I am sure a dump exists already for ColecoVision emulators. Maybe this can be a start of a new project for me. I would love to have a version of "Zaxxon" for the TI. Oh, and "Mr. Do!" as well.

[Tursi]

The biggest problem for me with the 9928, S-Video, Component, and so forth enhancements, and the reason I abandoned that path, is that none of these fix the video output from the console to be compatible with modern display devices.

 

To explain: NTSC video is normally 525 lines of video, interlaced so that two frames of 262 lines are drawn each 1/60th of a second (I'm not going to get into the details of the extra line or how it decides which is which... look it up online if you want to know).

 

However, a lot of consoles of the day, and the TI as well, actually outputs a PROGRESSIVE scan signal -- that's right, we had non-interlaced video long before the HD Video guys jumped all over it. We get a 262 line, 60hz progressive signal from our little console.

 

The problem is, that more and more modern video devices do not recognize this signal. If they see a 15kHz NTSC-compatible signal, they want to see the interlace. I've owned three devices now that can not sync reliably to my TI, 2600, Genesis, or ColecoVision.

 

So no matter how you enhance the signal coming out of the chip, if you don't insert some kind of scan doubler to add the interlace, you will still have the same progressive, low res image. I did consider some external enhancements that would be able to do it, and that may someday be my FPGA project, but the new chip is a cleaner solution.

 

As for ColecoVision, yes, same video chip (it uses the 9928 version), compatible sound chip (possibly slightly different white noise pattern), and Z80 CPU.

[Tursi]

Your problem with the component video converter; I read somewhere that R-Y and B-Y are not REALLY component signals...?

 

I couldn't find anything definate one way or the other that wasn't conjecture, so I just tried it. The color was good and the sync was solid, which is good enough for me. I didn't have to amplify anything but I did use a resistor to tone the signal down a bit, which is why I believe some balancing is required.

 

I can't find the thread I was talking about now. This person got a very nice (15Khz) RGB signal out of his 9928 and some adapter circuitry (stolen from Baby PacMan, IIRC). But since the adapter I tried worked for VGA out it's not necessary to hack up a classic arcade machine. But there was still a tiny bit of ghosting IIRC, which I'm pretty sure just come from the inside of the chip itself. About a thousand times better than composite though.

 

And if you want the 9928s I have, just PM me your address

[jchase1970]

Oh oh oh... speaking of cool 9918 tricks, do I recall correctly someone lamenting that "Zaxxon" was never released for the TI because of some lame-ass excuse about the graphics? Well, I turned up a ColecoVision "Zaxxon" cartridge in my storage room. (Not to mention "Mr. Do!") How difficult would it be to take the video data from the cart and convert the Z80 code to 9900? (The CV is Z80, yeah? Too lazy to WikiGoogle it right now.)

 

I don't see an immediate reason why a "Zaxxon" cartridge wouldn't be possible on the TI.

But it's a hell of job though. Basically you'd have to rewrite "Zaxxon" from scratch in TMS9900 language. Ofcourse you can "borrow" the colecovision graphics

and sounds. But that's about it.

 

Well...? Give me your address and I will send you the cart Of course, for that matter, I am sure a dump exists already for ColecoVision emulators. Maybe this can be a start of a new project for me. I would love to have a version of "Zaxxon" for the TI. Oh, and "Mr. Do!" as well.

 

I always wanted to play zaxxon on the TI to. My best friend had it on the coleco and I remember playing it alot. This was one of those 'where the hell is it games' I saw it listed as coming for the TI in ads but never could find it and I was never in any TI groups so I never knew it was never made.

[jchase1970]

I just looked up the game play of ZAXXON on the COLECO and there is also a MSX but the COLECO I think is a better port, it's smoother anyway.

[+OLD CS1]

The short: I got some 9928s from Matt, installed one in my console, made appropriate wiring changes in the console as well as a 5-pin DIN to S-Video cable. I am not happy with the results. I am too tired and brain-dead to post pictures or anything else. Just that I am not happy at the moment and will go to bed angry.

 

But I did get to burn things in the fireplace tonight. No, nothing related to the project.

[+OLD CS1]

Okay, so here is a quick run-down. I installed the 9928 without a problem. Made the strap changes as necessary, W205 to bypass the composite amplification circuit, and W201-W204 changes. I decided to leave one of the pins as ground, since the new configuration would normally require two more pins, leaving no ground. More details will follow when I am not watching our game.

 

In the end of two hours of work, and I did pretty damn good soldering work if I say so myself, I was greeted by the screen below.

 

I did some fumbling around and figured out that there is a signal coming from the B-Y and R-Y ports, but it is apparently just not right.

 

I did some digging around and found that what I am doing may not work, anyway. The NTSC S-Video standard is Y/C, luma and chroma. The B-Y and R-Y outputs are the U and V parts of a YUV signal, however, NTSC S-Video is not *really* YUV, but is a slightly different color space, YIQ.

 

SGI has a good page which goes over the differences, here

 

Appendix A. Video Basics

http://techpubs.sgi.com/library/dynaweb_docs/0530/SGI_Developer/books/Ind2Vid_PG/sgi_html/apa.html

 

Okay, pictures in a little bit.

[marc.hull]

Okay, so here is a quick run-down. I installed the 9928 without a problem. Made the strap changes as necessary, W205 to bypass the composite amplification circuit, and W201-W204 changes. I decided to leave one of the pins as ground, since the new configuration would normally require two more pins, leaving no ground. More details will follow when I am not watching our game.

 

In the end of two hours of work, and I did pretty damn good soldering work if I say so myself, I was greeted by the screen below.

 

I did some fumbling around and figured out that there is a signal coming from the B-Y and R-Y ports, but it is apparently just not right.

 

I did some digging around and found that what I am doing may not work, anyway. The NTSC S-Video standard is Y/C, luma and chroma. The B-Y and R-Y outputs are the U and V parts of a YUV signal, however, NTSC S-Video is not *really* YUV, but is a slightly different color space, YIQ.

 

SGI has a good page which goes over the differences, here

 

Appendix A. Video Basics

http://techpubs.sgi.com/library/dynaweb_docs/0530/SGI_Developer/books/Ind2Vid_PG/sgi_html/apa.html

 

Okay, pictures in a little bit.

 

 

I sure hope you don't live in Nebraska.....

[matthew180]

I did some digging around and found that what I am doing may not work, anyway. The NTSC S-Video standard is Y/C, luma and chroma. The B-Y and R-Y outputs are the U and V parts of a YUV signal, however, NTSC S-Video is not *really* YUV, but is a slightly different color space, YIQ.

 

Yup. This is the reason I think most people abandon this. The 9928's output requires quite a bit of analogue circuitry to amplify, impedance match, etc. to get it to work with a modern S-video or component input. I probably spent three or four straight days trying to understand all the differences and how to make it work. I hope you are more persistent than me! I'm looking forward to seeing the pictures you have.

 

Matthew

[Asmusr]

I attached my European console to my Sony Bravia KDL-40EX653 television using a simple component cable from Retro Computer Shack on ebay, and look what I got: an almost perfect picture with square pixels and none of the color bleeding I get when I use the RF modulator. From what I have been reading I'm not sure if anyone else have found a TV that works as well as this. Perhaps Sony is the way to go?

[sometimes99er]

Whauh ! That's a pretty good picture there. I wasn't planning on getting the real hardware. Been tempted from time to time. For testing purposes only. A better quality picture, without modifications to the original (inside jobs), is certainly a plus. Would go CF7 and no Peb, if it were. Thanks for posting.

 

[Asmusr]

I wasn't planning on getting the real hardware.

 

Perhaps something for you?

http://www.dba.dk/texas-ti-99-4a-komplet-med/id-91548477/

 

http://www.stargames.be/shop/105-accessoires

[sometimes99er]

Hehe, you're much too kind - handing it over on a silver platter like that.

 

[Asmusr]

Come on, you know you need it!

[TheMole]

http://www.stargames...105-accessoires

 

What? There's a webshop where you can buy these?!

I've been searching ebay for weeks in hopes of one popping up, even emailed the guy, to no avail. And these guys have them on stock? Maybe something to add to the resources thread?

[+OLD CS1]

ARISE CHICKEN! CHICKEN ARISE! I command this thread to resurrect!

 

If the amount of time passed does not give any indication, while I hate to have abandoned my project, I ran out of time to continue working a solution. I may dig into it later but not any time soon as classes have started up.

 

I am beating this dead horse because I have run into something which I can neither fit into another thread nor justify a new thread.

 

I recall reading in a thread somewhere that the 9918A did not exactly match NTSC standards, or something similar. I suspect this is the reason for the results attached with my new composite/S-Video-to-VGA converter.

 

The converter works fine with other composite devices, so I point my finger squarely at my TI. I am going to try with a QI console tonight to see if there is any difference.

 

(EDIT): I missed this above, but this is most likely the problem. The converter DOES work with the arcade joysticks (Jaxx Pacific, etc.)

 

 

However, a lot of consoles of the day, and the TI as well, actually outputs a PROGRESSIVE scan signal -- that's right, we had non-interlaced video long before the HD Video guys jumped all over it. We get a 262 line, 60hz progressive signal from our little console.

The problem is, that more and more modern video devices do not recognize this signal. If they see a 15kHz NTSC-compatible signal, they want to see the interlace. I've owned three devices now that can not sync reliably to my TI, 2600, Genesis, or ColecoVision.

[+OLD CS1]

Well, the problem is not the TI. The QI console did the same thing, but so does an Atari 130XE and an Atari 800XL. I tried a PC with a composite output, of course that works fine. I will be trying a few other composite devices, but my suspicion is this device does not work well with our old computers.

 

 

Oh, and as one last gasp for this project, I have pretty much passed on the idea of getting S-Video directly from the console. For one, with the F18A around (though in limited availability,) S-Video from the console seems moot. Secondly, I have found passive composite-to-S-Video adapters which produce a very good signal from the TI.