TMS-RGB Mod

[+retroclouds]

The folks in the Colecovision community created an open source hardware RGB mod for TMS9928A and TMS9929A VDP.

 

https://tms-rgb.com

 

 

 

[+retroclouds]

From their documentation:

 

Quote

This small adapter board can unlock the potential of many 8-bit systems from the 1980's including the Colecovision, MSX, SG-1000, SC-3000, TI-99/4, Memotech MTX, and more. Instead of the original output—often wavy, blurry RF with color fringing—soldering this tiny board directly to the back of your graphics chip will enable clean RGB output, ready to be sent to your favorite line doubler (like the OSSC), RGB monitor, or upscaler (like the Framemeister).

 

[Falonn]

For completeness: we haven't tested compatibility with any other systems besides Colecovision yet, though there isn't anything Colecovision-specific in the circuit.  My confidence is high that it should work anywhere the '992xA series VDP has been correctly integrated.  Better news: I have seen the exact same mod board working on both an NTSC and PAL system without any changes.  So that seems to bode well for compatibility, too.

 

If anyone decides to try it with a TI-99 or otherwise, it would be great to hear about positive (or negative  ) results!

 

As for that list in the documentation, I did my best to track down which products in the 80's used the '9928.  It's tricky because there seems to be a lot of (slightly) incorrect information out there.  (E.g., I've seen Colecovision described on more than one website as having a '9918 which has never been true as far as I know.)  You guys would know a lot better than I would: was I correct in listing the TI-99/4 as using a '9928A?  Thanks!

[Omega-TI]

1 hour ago, Falonn said:

was I correct in listing the TI-99/4 as using a '9928A?  Thanks!

For the U.S. Version it was the TMS9918.  My memory is not what it used to be, so I cannot remember if the European PAL version used a different version or not.

 

Was this thing developed because the other product was no longer available?

[Falonn]

7 hours ago, Omega-TI said:

Was this thing developed because the other product was no longer available?

Do you mean the ColecoRGB?  There is a more complete version here, but the shorter story is that all three or four of the widely-known circuits had at least one problem.  Almost all of them suffered from the "too much blue" problem (due to an interesting choice on TI's part when designing the VDP) and pixels that are too narrow when coming from black (making text very hard to read) was a problem unique to ColecoRGB (among others).

 

TMS-RGB combines the strengths of each while working around the weaknesses of each.  You don't need an oscilloscope during install anymore to tune the colors, etc.

[+OLD CS1]

Not the "A" version for the 99/4, at least for the 9918.  I do not know if there are non-A versions of the 9928A and 9929A.

 

EDIT: While working to confirm or deny, I found some things in the Wikipedia article for the TMS-9918 which could be cleaned up:

 

Screen mode 2 details

[...]

When manipulating the screen in BASIC with the LINE command, one easily could exceed the maximum 2 colors per 8×1 area[...]

 

What LINE command in what BASIC??

 

Scrolling

The TMS9918 does not have any scroll registers, and so scrolling must be done by software. Furthermore, scrolling can only be done on characters boundaries.^{[citation needed]}

 

Citation needed, indeed.  Pretty certain this assertion is well debunked.

 

[wierd_w]

Too bad Wikipedia has a rule against Original Research;  You will need to publish documentation showing this is false, THEN get it well circulated for a credibility score, THEN get SOMEBODY ELSE to battle with the page maintainer to make the correction and add the citation.

 

 

[+mizapf]

The LINE command is from some other BASIC; there are some machines apart from the TI that use the 9918A. The "Furthermore" sentence is wrong and should be removed completely (update comment "scrolling possible by tile scrolling, pattern updates"). This is not a matter of original research; the wrong claim is just removed.

[wierd_w]

hmm.. I suppose just linking to any of Rasmus's fine titles that make use of this method on the 99 would be sufficient citation. Still, I have heard too many horror stories about trying to get a simple correction in edgewise on curated articles.

 

It's worth a shot; there are many examples in the wild of this being done that could be pointed to.

[Falonn]

Hmm, that sounds like there are about three votes for the TI99/4 not having a '9928A.  Should I remove it from the list on the mod's front page?  That was the machine I was the least sure about while trying to find things that used it.

[JB]

I believe it is true of the 99/4a in PAL land, but not the kingdom of NTSC.

In America, TI used a 9918a that directly outputs composite video.

 

The 99/4 with no a used a 9918, not a 9918a. Not sure on the PAL-land situation there.

 

[+Ksarul]

For a more complete note on VDPs and the TI computers: The TI-99/4 (1979-1981) used the TMS9918, and was only released as an NTSC machine. The TI-99/4A (1981-1984) used the TMS9918A in NTSC land and the 9928A/9929A pretty much everywhere else. There were also mods for the NTSC machines to replace the video chip with a 9928A to get RGB capability. The TI-99/8 used the TMS9118, as did the Tomy Pyuuta/Tutor line. The Powertran Cortex could use the 9918A, 9928A, or the 9929A, IIRC.

[JB]

16 hours ago, Falonn said:

As for that list in the documentation, I did my best to track down which products in the 80's used the '9928.  It's tricky because there seems to be a lot of (slightly) incorrect information out there.  (E.g., I've seen Colecovision described on more than one website as having a '9918 which has never been true as far as I know.)

Oooooh yeah. There's a tendency to write everything in the family down as 9918 or 9918a. From a specs and programming point of view, this is even accurate.

...

And then you flip a coin to determine if the writer writes "9918a" for both the a and non-a families, or writes "9918" for both. Of which neither approach is right in any way.

[Tursi]

I don't know if you can come up with enough 9918-NO-A chips to be worth even considering anymore...

 

[Falonn]

Thanks for the clarifications, everyone.  I think it'll be better if I remove it from the list.  I'd rather have the PAL version be covered by the "and more" at the end of the list than have some unsuspecting NTSC TI-99/4A user go to the trouble of building or buying one of these things just to find out it won't work for them.

 

Eventually as more of these RGB boards make their way out into the wild, the right answer will be to add an "actually tested on" compatibility list as those reports come in.  In the meantime, it's better to be safe than to inadvertently make someone else sorry.

 

Thanks again!

[ChildOfCv]

10 hours ago, OLD CS1 said:

Citation needed, indeed.  Pretty certain this assertion is well debunked.

In the most naive sense, that is exactly right.  There is no pixel scrolling.  If you want to do so, you need to write multiple pattern maps into the RAM and switch between them.  But due to limited memory, this also limits the total number of unique tiles you can use.

[ChildOfCv]

8 hours ago, JB said:

I believe it is true of the 99/4a in PAL land, but not the kingdom of NTSC.

In America, TI used a 9918a that directly outputs composite video.

 

The 99/4 with no a used a 9918, not a 9918a. Not sure on the PAL-land situation there.

 

As far as I know, there was never a 9919 or the like.  So PAL equipment had to do their own composite encoding.

[+Ksarul]

Actually, the TMS9919 was a sound chip. . .later called the SN94624.

[Tursi]

I want to say that I /also/ found evidence years ago that the 9919 - a PAL VDP with composite output - was at least specified. But I can't remember now why I think that.

 

[JJB]

On 7/12/2020 at 12:05 AM, Falonn said:

For completeness: we haven't tested compatibility with any other systems besides Colecovision yet, though there isn't anything Colecovision-specific in the circuit.  My confidence is high that it should work anywhere the '992xA series VDP has been correctly integrated.  Better news: I have seen the exact same mod board working on both an NTSC and PAL system without any changes.  So that seems to bode well for compatibility, too.

 

If anyone decides to try it with a TI-99 or otherwise, it would be great to hear about positive (or negative  ) results!

 

As for that list in the documentation, I did my best to track down which products in the 80's used the '9928.  It's tricky because there seems to be a lot of (slightly) incorrect information out there.  (E.g., I've seen Colecovision described on more than one website as having a '9918 which has never been true as far as I know.)  You guys would know a lot better than I would: was I correct in listing the TI-99/4 as using a '9928A?  Thanks!

Just confirming that TMS RGB works very well in my US TI after replacing the TMS9918A with a TMS9928A I purchased from Ebay. Pin 38 (CPUCLK) wasn't used for the sound processor so I just terminated it with a 560R resistor, same with pin 35 (EXTVDP). The composite video signal from pin 36 goes to a inductor (L202) first, I cut the incoming side and bent it out of the way. Right under that inductor is an unused wire link which can be conveniently used to terminate pin 36 with another 560R resistor. 

 

I re-used the existing TI video connector to provide RGB out  and made a custom DIN - SCART cable to connect to a cheap RGB - HDMI converter. SCART needs a positive potential on pin 16 (Fast Switching) to select RGB instead of CVIDEO but with a 5-pin DIN I didn't have a spare pin to carry 5V from the console. I just opened up the converter and connected pin 16 internally to a suitable 5V point via a current-limiting resistor.

 

Cheers,

 

Jochen

[Calab]

On 1/25/2021 at 10:40 PM, JJB said:

Just confirming that TMS RGB works very well in my US TI after replacing the TMS9918A with a TMS9928A I purchased from Ebay. Pin 38 (CPUCLK) wasn't used for the sound processor so I just terminated it with a 560R resistor, same with pin 35 (EXTVDP). The composite video signal from pin 36 goes to a inductor (L202) first, I cut the incoming side and bent it out of the way. Right under that inductor is an unused wire link which can be conveniently used to terminate pin 36 with another 560R resistor. 

 

I re-used the existing TI video connector to provide RGB out  and made a custom DIN - SCART cable to connect to a cheap RGB - HDMI converter. SCART needs a positive potential on pin 16 (Fast Switching) to select RGB instead of CVIDEO but with a 5-pin DIN I didn't have a spare pin to carry 5V from the console. I just opened up the converter and connected pin 16 internally to a suitable 5V point via a current-limiting resistor.

Hello!

 

I've got one of these TMS RGB boards that I'd like to use in my TMS9918A NTSC QI machine. Could someone spell out the process of replacing the TMS9918A with a TMS9928A, while retaining the composite video (and/or SVideo if possible) so I could fit the TMS RGB board?

Edited January 22, 2022 by Calab

[+Ksarul]

As soon as you replace the TMS9918A with the TMS9928A, you basically lose composite video out, as it is replaced with component video out. This is a function of the video chip, so there isn't really an easy way around the issue. You might try and build a daughter board that pulls all of the signals from the video socket up to it--and set up some kind of switch between a 9918A and a 9928A, but that would probably get complicated fast. . .

[+OLD CS1]

5 minutes ago, Ksarul said:

You might try and build a daughter board that pulls all of the signals from the video socket up to it--and set up some kind of switch between a 9918A and a 9928A, but that would probably get complicated fast. . .

If you have access to the schematics, there are points on the logic board where you can pull the Y-Cb-Cr signals.  I did this way back when I was trying to get S-Video off the board from the 9928A.  (This experiment failed as S-Video is a different color space than what the 9928A outputs.)  Otherwise, yeah, find some way to intercept the signals direct from the 9928A.

[Calab]

On 1/25/2021 at 10:40 PM, JJB said:

Just confirming that TMS RGB works very well in my US TI after replacing the TMS9918A with a TMS9928A I purchased from Ebay. Pin 38 (CPUCLK) wasn't used for the sound processor so I just terminated it with a 560R resistor, same with pin 35 (EXTVDP). The composite video signal from pin 36 goes to a inductor (L202) first, I cut the incoming side and bent it out of the way. Right under that inductor is an unused wire link which can be conveniently used to terminate pin 36 with another 560R resistor. 

So, just to clarify, before I install the TMS9928A into my TI-99/4a, I should bend pins 35, 36 and 38 out so they don't go into the socket? Connect ping 35, 36 and 38 to ground through ~560ohm resistors?

[ChildOfCv]

Don't bend any pins.  Solder it in place just like the previous one was.  There will be a few resistors and jumpers to change.  There are a series of posts about this somewhere back in this thread, with pictures.