TI-99/2 questions

[+mizapf]

Klaus and Fabrice (and Ciro), do you see socket names on the board where the EPROMs are plugged in? Typically, in MAME, the dump file names contain an indication of the socket where the dump was taken from, e.g. "994a_rom_hb.u610".

[fabrice montupet]

As a very early prototype, my 99/2 has no Silk/Top Assy mark. ROM4 and ROM2 are piggy-back, ROM0 is alone on its socket. If you want, I can trace the EPROM signals to the two 74LS138 to study the adresses decoding.

Edited April 22, 2018 by fabrice montupet

[kl99]

Circuit Board is labeled with 1055197-1
Eproms are stored on the sockets labeled
U2 2x TMS 2564JL-45 MEP8327 (0000, 4000B)
U3 1x TMS 2564JL-45 MEP8327 (4000)
U12 1x TMS 2564JL-45 MEP8327 (2000)
assignment according to page 1 of the TI992_Tech_Diagrams.pdf

[fabrice montupet]

The unique information written on the PCB on mine (production #44) is on the bottom side, we can read the product reference: 1055184-1

[+mizapf]

Fabrice, I think in your case it suffices to use some names like "rom0000.single", "rom2000.bottom", and "rom4000.top". The objective is to give a clear indication from which circuit the dump was taken without knowing details about the selection.

 

Without socket names, this should be the best we can achieve.

 

I suppose we are sure about this ordering.

[+mizapf]

Klaus, the TI992_Tech_Diagrams.pdf do not fully tell the truth, as it seems to me. There are two 4016 circuits (U3 and U4) which are the RAMs, but all four ROM circuits are represented by a single chip 32Kx8 on U2 (look for the 15 address lines).

 

Maybe you can find the names on your PCB.

 

[Edit: I just saw that the socket names can be found in the Byte article. Is that your PCB? Where are the stacked ROMs - can't tell from the picture.]

Edited April 22, 2018 by mizapf

[kl99]

U2, U3, U12 naming is coming from my PCB. U2 is stacked as stated in my post.

i only took the Tech pdf page 1 to map 0, 2, 4, 4B to those locations.

[+mizapf]

OK, thanks. The TI-99/2 documentation feels very preliminary at many places.

[+mizapf]

Klaus, have you noticed that your website is down? You seem to have a database issue, maybe your db server crashed.

[kl99]

yes, when i discovered it on friday, the provider had reported having many issues. now those seem resolved. so either they screwed up or it is really unrelated.

will check

[kl99]

website got fixed.

[+mizapf]

The document "TI992_Video_Chip.pdf" mentions a special video mode "bitmap", which I found fairly surprising. It can be selected by using the byte at EF00.

 

Klaus, and all others with a real 99/2, could you please test the effect of the byte at that address for values from 0 to 7? If there is a bitmap mode on bit 2 (counting from the right), it should be turned on by values 4,5,6, and 7, and it should be visible by a screen with random patterns.

 

The manual states that for the bitmap mode, you need at least 6K RAM for the screen (192 * 256 / 8 = 6144). They also say that the area starts at E000, which means it would intersect with the on-chip RAM, which I believe is not true. Maybe this was the joyful dream of some TI engineers.

[fabrice montupet]

Value 4 : Background and the screen deadzone turn to dark color. Text is white.
Value 6 Screen deadzone turns to white color.
Value 5 : Background turns to white color and the deadzone .Text is white.
Value 7 : Background and the screen deadzone turn to white color. Text is white.

 

In all cases, there's no random patterns.

Edited April 29, 2018 by fabrice montupet

[+mizapf]

Thanks so far, Fabrice. Are these effects the same with the values 0,1,2,3? In that case, bit 2 has indeed no effect. But you have an early 99/2 anyway, so we shall see what Klaus says.

 

Deadzone means border, right?

[fabrice montupet]

Yes, dead zone is border :-)

 

Value 0 : Background turns to white color and the deadzone to dark.Text is black.

Value 1: Background, screen deadzone and text turn to dark color.

Value 2 : Background and the screen deadzone turn to white color. Text is black.

Value 3: Background turns to dark color, screen deadzone to white and text to black.

 

And to resume (with a correction):

 

Value 4 : Background and the screen deadzone turn to dark color. Text is white.

Value 5 : Background turns to white color , screen deadzone to dark. Text is white.

Value 6 Background turns to dark color, screen deadzone to white. Text is white.
Value 7 : All is white.

 

 

 

When I will have some free time, I will upgrade my 99/2 with a newer ROM set. Of course, when the tests will be realized, I will re-set it to its original state.

[+mizapf]

I mixed up the bits, as I noticed. However, this is an unexpected result. It means that all three bits are significant, since you get 8 different results.

 

bit 2 = text color (1=white)

bit 1 = border color (1=white)

bit 0 = background color (1=text color, 0=inverted)

 

So all odd numbers are not useful, since you won't be able to read the text on the screen. For the even numbers, the background color is the inverted text color, so you can equivalently call this the background color, with 1=black. This is what we find in the specs.

 

I have some doubts that you can simply use the newer ROM set. I am sure that the custom chips CF40051 and CF40052 have also changed. Remember that the BEOL character for the older version is any character from >70 and higher, while the newer version defines more characters up to >7F, so the video controller must display them.

[fabrice montupet]

I thought about the possibility of some modification made on the two TAL004 but I also thought that I was not risking anything to try :-)

[+mizapf]

I thought I should let you know ...

 

Edited May 2, 2018 by mizapf

[fabrice montupet]

Yes!!! :-)

[+mizapf]

I still need to do some work, making cassettes work, and of course the Hexbus. If I did it correctly, it should just mean to add an emulation of the interface chip. We'll see how much logic is inside the custom chip. The Hexbus controller in the HX5102 was quite simple, while the Oso chip in the 99/8 was very elaborate.

 

Also, we will have to do some speed measurements to verify that the emulation runs at the real speed.

 

 

 

When this is done, I'll continue with the rarest of the rarest, the 99/4B and 99/5.)

 

 

[fabrice montupet]

You really do a great job! Congratulation!
I'm happy that people who don't have these TI prototypes will be able to play with them too :-)

I would like to know how many guys have a 99/4B or the 99/5. Do you have information about that? I can't imagine that it only remains one 99/5. During so many years I search another owner of this rare bird but I haven't found... Please owner(s), let us know!

 

[Casey]

This is fantastic!!!

[+acadiel]

I still need to do some work, making cassettes work, and of course the Hexbus. If I did it correctly, it should just mean to add an emulation of the interface chip. We'll see how much logic is inside the custom chip. The Hexbus controller in the HX5102 was quite simple, while the Oso chip in the 99/8 was very elaborate.

 

Also, we will have to do some speed measurements to verify that the emulation runs at the real speed.

 

 

 

When this is done, I'll continue with the rarest of the rarest, the 99/4B and 99/5.)

 

 

 

 

We need to encourage the CC-40 emulation guy to add your Hexbus to it

[kl99]

You really do a great job! Congratulation!

I'm happy that people who don't have these TI prototypes will be able to play with them too :-)

 

I would like to know how many guys have a 99/4B or the 99/5. Do you have information about that? I can't imagine that it only remains one 99/5. During so many years I search another owner of this rare bird but I haven't found... Please owner(s), let us know!

 

 

 

Amazing Progress mizapf! Sorry I had no time to test the 99/2 with your questions. I was busy with a broken TV, and cleaning up the boards of the DS990-1, another candidate for emulation

 

Short time ago I talked with Steve Eggers, also the 99/4B was a topic, he published a photo of a waxwing board (99/4B) on his website

https://aug.99er.net/images/Unreleased/994btop.jpg

https://aug.99er.net/unreleased.htm

The story is that once a member came to the group meeting who had got the board from a TI employee. So Steve made a photo. The rest is history.

The photographed board looks really naked. What I find interesting about the 99/4B is the two unique buses.

 

Also I have heard from TI employees, back then the primary engineering stuff was busy with 99/4A, 99/8 and peripherals. So the 99/4B and 99/5 were done in a seperate team (or teams).

 

Steve had the following documents:

- TI-99/4B Block Diagram

- TI-99/5 Block Diagram

- 99/5 Memory Map

- TI-994B "Waxwing 4B" Schematics

 

All documents that Steve had are with Ksarul now.

 

We could find more information if somebody goes to the TI Archive library, located here

https://legacy.lib.utexas.edu/taro/smu/00063/smu-00063.html

Sometime in the past TI donated the whole archive to a library, occupying over 1500 cubic feet of material.

I got this information when I asked TI support about buying/getting some TI-99 internal documents this year.

 

Regarding prototypes or board survivors: afaik Michael Becker is the former owner of your 99/5. I only know that Lee (Labtech1) had a waxwing board once but it got sold with his 99/8.

Besides that nothing.

[fabrice montupet]

Yes, they are the very few informations about 99/4B & 99/5 that I have found during my searches during these last years. There is also an interesting article on CAUG9404 and a little paragraph in MP8912.... but It is very weak .
I hope that we'll find new ones.
I have sent the ROM dump of my 99/5 to Michael and I stay available for any other operation to help its emulation.