Jump to content
IGNORED

99/5 TI documents Quest


Recommended Posts

PHROM = Speech Rom ?

 

Here is another internal document that covers the 99/4B in detail:

http://ftp.whtech.com/datasheets%20and%20manuals/Specifications/Peripheral%20Expansion%20System%20Theory%20Of%20Operation%20and%20Technical%20Traning%20Guide%201982-09-03.pdf

 

There is a great interview from Dan Eicher with Don Bynum taken in 2002. Here are some extracts:

 

Q.Please tell us anything and everything about the 99/8.
A.This was an awesome machine for its time. It supported real development without the horriblerestrictions of a 64K address space. Further, it leveraged every technology that we had found usefulby third-party developers on the 99/4A. Politically, consumer group could not produce a machine with80-character text, but we planned for the eventual elimination of that concern (unfortunately thatconcern was eliminated when the business was shot down.)

Q.How many units were produced?
A.I don't know how many were actually produced. I left before the first units were built. Herb Shanzersent me one which I kept around for a few years then scrapped.
Q.Who were the engineers involved?
A.There were a bunch, from all over TI.
Q.How compatible was it with the 99/4A?
A.The best I can recall, every single cartridge application worked. Machine language programs, of the"well behaved program" category made famous by Bill Gates, from third parties all worked, as far asI know.
Q.What problems did it have?
A.The joystick interface on the unit I had had X&Y crossed over on one of the sticks. A trivial changewas needed to fix that. Otherwise it was simply amazing. I think the one I had may have had 512K ofRAM … at a time that the IBM PC was not coming near that for $5,000.
Q.What was your involvement with the 99/7, 99/5, 99/2?
A.I was involved with the 99/2. I scared Timex out of the computer business. It was a very well executedsilly idea put in place because some sales guys were losing a few sales to Timex at some schools. It wasapparent that technology would overtake Timex as they were running margins inadequate to sustaintechnology renewal. We just made that extremely apparent for Timex management.

Q.What was your relationship with Michael Bunyardas he tended to do a lot of work on 4A hardware?
A.What a great guy! Mike worked for Alan Lawson, who reported to me. He was the architect of thehardware virtual memory design for the 99/8. His chip worked the first time through. Lots of peopleat TI, including a few in semiconductor, rather publicly declared that the design would never work.It not only worked … the first chips off the fab worked flawlessly!
Q.Do you know who wrote the OS (ROM and GROM 0) for the 4A?
A.Yes, Johnny Ackers was the head of that group.

Q.How did the idea of moving towards the Hexbus come along?
A.The Calculator Division was building handheld, today we would call them mobile, computing devicesand were moving up in complexity. It was apparent that LCDs were going to become viable for muchlarger screen sizes and that a true portable, handheld, not just luggable, computer was going tohappen. They needed low cost, low power peripherals. We anticipated the need for similar capabilityat the low end of the home computer line and possibly in future home networking functions (this waslooking out a decade or more with the very murky crystal ball available to product marketing folksthen and now). It was an easy decision to sign on to include Hexbus in future designs. Today's USBis a much more functional descendant of Hexbus at TI and GPIB4 and HIL at HP.
Q.It was rumored the 99/4 was to use the 9980 — an 8-bit version of the 9900 — is that why the bus was"crippled" out to the PEB?
A.I think you mean 9995. This pre-dated me, but it was actually a reasonable cost compromise. The truthis that the 8-bit bus was three orders of magnitude less damaging to speed than was the wretchedGROM-based implementation of the O/S.

Q.The 99/4B design was not implemented. Was the reason because it came along too late?
A.Please clarify the question.
Q.I found the info I have on the 99/4B. Can you tell me who was the lead engineer and why it was neverproduced?
A.Beats me! If there really was such a thing, it would have had Alan Lawson as the lead engineer. WhenI left, our real plan was to create a stripped down 99/8 to replace the 99/4A, about 2 years out, but wewere not thinking of retaining the /4(A) part of the product identity as it implied a particular level ofcapability that would have understated what was going to be available. At the same time the 99/ partof the identity was well associated with easy to use technology that enriched the life of a family. Giventhe consumer research that we had done, I would be very surprised to learn that Herb would havetried to roll out a "B" extension of the family.

Q.The TI-99/4B was a revamped 4A based on the TMS9995 processor and using the Hexbus forexpansion (floppies, hard disks, serial and parallel ports, etc.) RAM and speech were to be added bythe user to the console. It ran between five and ten times faster than the 99/4A and could be slowedto 4A speed at power-up ("Press 1 for slow, 2 for normal, 3 for fast"). We had a contest to see who couldsurvive to level two in Parsec. (I've also seen these called Waxwings.) The speed feature, developed forthe 99/8 was necessary for many of our educational games. The 99/8 architecture delivered some muchfaster execution that the games became pretty much unplayable. Arcade players would keep tryinguntil they broke the joystick, but kids trying to run Scott, Foresman software were quickly frustratedand gave up (did you have at least one teacher like that in college?).

A.I think we only made about nine of the 99/4Bs and 99/5s. All of which went to the engineering andsoftware folks.
The 99/4B described above is the stripped down 99/8. It was not slated to be marketed as a /4extension. The speed feature described above was from the 99/8 bios (and involved some trick codewhen serial interfaces and disc controllers were involved.).
The 99/5 was a 4B with speech and ROM installed at manufacture. This is a 99/8 with no p-System (wesaw a need for that as many would not need the p-System. Adding p-System was going to be a depotupgrade (about 5 minute job.)
The 99/2 was a low end unit designed to compete with the Timex Sinclair. It too was based on the 9995processor and used the Hexbus for peripherals. It was laid to rest once the 4A vs VIC-20 price warheated up.
The 99/8, also 9995 and Hexbus based (see a pattern here?) came with speech, 64K RAM, Basic andp-Code all in the console. The 4A's PEB could also be used for floppy/hard disk expansion, RS232, andup to 12 MB of RAM expansion (never tell an engineer you wanted more RAM without putting anupper limit on it). There were over 300 99/8s produced as a pre-production run just before the plugwas pulled on the whole program.
The 99/8 would support up to 128 MB of onboard RAM. It was designed to use memory chips for thelarge memory versions, that were on the S/C product roadmap, but which were not yet in production.External RAM, in the PEB was going to be an interim, and much slower, option.

Q.How was the decision made to go from the classic black-and-silver 99/4A to bland beige?
A.Again, that was easy. The polished aluminum overlays were easily damaged by kids, who were ourprimary intended users, were easily damaged in handling in the factory (therefore adding cost), andserved no utilitarian purpose, so we eliminated them. My own preference would have been to stay witha black case, but ergonomic standards were emerging in Europe which dictated a lower contrast color.

Q.On the 99/7 do you know if any where ever produced?
A.There were two 99/7 projects. The Ranger was also called 99/7 for a while. None of either were everbuild in the "production" sense.

 

Some Magazine extracts:

 

Compute! Magazine, 1983 August:

The fall computer collection at the summer Consumer Electronics Show

...

Pre-show rumors were that TI would introduce one or two new computers, possibly the TI-99/4B and theTI-99/8. So much for rumors.

The 99/8, however; is said to be very near. Insiders say it will come with 80K RAM, built-in speech, andsell for roughly $500. Basic, Forth, Logo, and UCSD Pascal will be the available languages. The 99/4B, theysay, will fall somewhere between the 99/4A and 99/8 in features and price.

It could be that TI is somewhat gun-shy after its recent experience with the 99/2. Introduced at the winterCES, the 99/2 was an economy version of the 99/4A without color or sound. It was designed to sell for $99.Unfortunately for TI, the ongoing price war with Commodore and Atari heated up a little faster thananticipated. To compete, TI slashed the price of the 99/4A again and started another rebate program. Thisbrought the 99/4A to under $100. Unable to cut the 99/2's price accordingly, TI was forced to drop the newmodel it had spent months (and millions) developing. Now that each one of the Big Three has been burnedin a similar way — Atari with its 1200XL and Commodore with its P128 and Max Machine — they maybe more circumspect about making splashy introductions of new computers.

Although TI unveiled no new machines at CES, the company did introduce a 99/4A with a redesignedwhite housing. Word is the new plastic case is cheaper to manufacture, and that it will match the designof the coming 99/8.

...

 

Micropendium 1989 December:

TI's unreleased legends - Products that never reached the market [by Richard Fleetwood]

...

The TI-99/4B
Also about the time good things were happening with software development, TI was looking toward thefuture and the possibilities of system expansion. TI engineers played with ways to make the computersimpler to use, yet more complex. They threw together a half-dozen "all new" 99/XX computers based onmarket research and other criteria. These new computers were little more than proposals and prototypesfor the new wave of home computers that would renew the infamous home computer wars of the early1980s.
These new consoles were known as the 99/4Bs. They had a brand new memory mapper and used the bigbrother of the 99/4As microprocessor. They were equipped with the new 9995 microprocessor that enabledmuch more memory to be accessed, as well as much faster throughput of machine code. These new chipswere much more efficient than the 9900, and soon proved very capable and powerful in a small home-basedsystem. The 99/4B was equipped with 32K of memory, a faster system clock, and a semi-new keyboard.All six of these prototypes were basically hand-built and weren't meant to be marketed. They were thetesting ground for the 99/4As big brother — the 99/8.

The TI-99/8
The 99/8 was TI's final attempt at making it to the top of the home computer market. At the time it wasgoing to be released, it would have been more powerful than any other home computer in its price range.(Rumors about the imminent release of the 99/8 were hot and heavy the fall of 1983, with many expectingit to be on dealer shelves in time for the Christmas rush. — Ed.)

The 99/8 was equipped with 64K of memory and could be upgraded to 2 Megabytes. Unlike all othercomputers at the time, it had built-in speech capabilities. Instead of Basic as a menu selection on power-up,the 99/8 came with Extended Basic II. XB II was an upgrade of Extended Basic, with improvements ingraphics commands, string handling and new routines that made use of hexadecimal/decimal numberhandling. Also available on the power-up screen was the Pascal p-Code system.
With all these standard features, the 99/8 was in a class by itself. Also included were ports for cassette,video, AC power, and the all-new Hexbus port. The expansion port on the side of the console had 50 pins,compared to the 44 on the 99/4A. The extra pins and some juggling of signals gave the 99/8 true 16-bitperformance on its I/O bus.

The cartridge port was mounted on top of the unit, and installing a cartridge consisted of inserting itstraight down instead of pushing the cartridge into the front of the console as with the 99/4A. Thekeyboard was redesigned and included several new keys to reduce dependency on the Function keys forsuch characters as ?, ", _, ', ~, |, [, ], and so forth. The FCTN key was also moved to the left side of thekeyboard, so that users could maintain full cursor control with one hand instead of two.

The 99/8 keyboard was almost four inches wider than the 4A keyboard, and touch-typing was easierbecause it felt like a full-size IBM Selectric typewriter keyboard.

Incidentally, the power-up menu of the 99/8 offered another option: system speed. You could choosebetween "slow" mode, 99/4A mode or "fast" mode. This control over operating speed made it possible tochange the speed at which a program ran. It was interesting to try to play Munch Man at full speed on the99/8 because it ran much faster than on the 4A. Similar effects were noticeable with other cartridges aswell.

250 were manufactured

In discussions with others who know about the 99/8 project, I have put together the following facts:

* The 99/8 project almost died in the prototype stage because of the complexity of the memory mapper.A big breakthrough by one engineer kept the project going.

* There were about 1,000 etched PC boards made. Only 250 of these were assembled into working units.Of these, only about 150 were considered to be final, pre-production versions. These early units, if theyhad the Pascal system installed, held the code on ROMs instead of GROMs. This was to facilitatedebugging until the final version was ready.

Speaking of Pascal, I talked with the fellow who had the responsibility of taking the actual silicon wafersfrom the SC building after etching to Singapore, where the final GROM chips were to be manufactured.He made it as far as Los Angeles before he got a call on Black Friday to come back home. That was whenhe learned that TI was getting out of the home computer business.
I'll bet that fewer than two dozen 99/8s have the Pascal system intact. My 99/8 doesn't have it. I have seenabout two dozen of the 99/8s, and none seems to be exactly the same. Each had a different "feel" and someof the operating characteristics.
TI's code name for the 99/8 project was "Armadillo," which for non-Texans is a feisty, little armoredmammal that roams the Texas plains and Hill Country. On more than one occasion, while displaying my99/8, a former or current TI employee would remark when they saw my computer, "Wow, an Armadillo!"Most of these TI'ers had heard about the project but had never seen one. The ones who had seen it whileit was being developed provided me with much of the information for this article.

Next month: More on the 99/8 and software compatibility with the 99/4A, the 99/2 and the GROM box.

 

Micropendium 1993 August:

Reader to Reader:

...

Vandsteene Carlo, Elsrakenstraat 52, B-3500 Hasselt, Belgium, writes:

I'm about to write an article about the things (computers and peripherals) Texas Instruments wasgoing to release, but never did because of the withdrawal from the home computer market, now 10years ago. I've already found some (general) information about the 99/8, the GROM Box … But overhere in Belgium, it's really hard to find. I'm still looking for more information, photographs,photocopies, printouts, articles, etc, in order to get a complete (as possible) survey of what TI intendedto produce or had produced but didn't release.

That's why I address myself to you; can anyone send me some information concerning the 99/4,(99/4A), 99/4B, 99/8, 99/2, CC-40, hard drive controller, four-part RS232 card, 128K Super RAM and374K Ultra RAM card, GROM Box, smart modem, video controller? Anything would help; aphotograph you took from your 99/8 if you own one (yes, they have been sold in Texas!) or a photocopyof it out of a magazine or a brochure. I don't expect you to send me your owner's manual, but especiallydescriptions, test reports, documentation, advertising, pictures and so on (may be photocopies too).So if you have anything, could you please send it to the address above? However little it may be, it willbe well appreciated. But, there is a but, be sure that I receive it as soon as possible, because it will takea while to get the article ready in time. Thanks beforehand.

...

 

Micropendium 1994 April,

Ah, the elusive TI-99/5

You'll probably never see one, but what the heck — wouldn't it have been nice if TI had come out with theTI-99/5, if not the TI-99/8. In the February Bugs & Bytes column we published an item that brieflyoutlined a controversy regarding the 99/5 that had appeared on the Internet and elsewhere. The focus ofthat controversy was on a so-called marketing plan for the 99/5 that allowed users to trade in their 99/4Asfor more money than they'd have to pay for a 99/5A.
Glen Bernasek, of the Cleveland Area TI-99/4A User Group, published an article in the group's newsletterdetailing some of the scuttlebutt about and photos of the 99/5. His information was provided by someonewho actually owns one. Here are some excerpts:

The cassette port was replaced with a Hexbus port. The Hexbus was an eight-wire, four-bit widecommunication cable. Devices designed for the Hexbus included a floppy disk controller, a serial port,streaming tape drive, 80-column video controller and portable printers, among other things.
The TI-99/5 used a 9995 processor.
The TI-99/5 had a 32K memory expansion and speech synthesizer built into the console.
The motherboard was imprinted with "99/5 11/2/83." TI was working on it right up to the end.

According to other sources, who posted information on the Internet, the 99/5 was also known as the 99/4B.It had no PEB port because it was meant only for use with Hexbus peripherals. He also reported that TIwill disavow any knowledge of the 99/5 or the 99/8 or virtually any other hardware or software meant forthe home computer market that was never actually shipped. However, he said, when he asked whetherhe could distribute such software to his user group members TI prohibited it on the basis of holding acopyright to the software it wouldn't acknowledge existed.
Other items that were developed in small quantities for the home computer market include a hard drivecontroller, a GROM box similar to a GRAM Kracker, an IEEE interface card, a 128K Super RAM card, a374K Ultra RAM card, a 4-channel music card, a PEB interface card for the 99/8 and an RS232 card withfour ports.

 

  • Like 1
Link to comment
Share on other sites

SmartProgrammer 1984 July,

Questions & Answers:

What happened to the rumors about Toys R Us and/or GE buying the rights to produce the99/4A?
Last we heard about this item was that TI was asking too much money for the rights, so both companiesdecided not to buy! Also, everything has been quiet on the 99/8. At one time we heard that Control DataCorp. was thinking about buying this or a version of the 99/5, but that was a long time ago.

 

Micropendium 1994 December,

Feedback: "Hoax" a misnomer

As an ex-user of the 99/4A, 99/8 and the Geneve, I still stay current with the TI community. The reasonI'm writing is this: I was told by a subscriber of your publication that there was an article about a hoaxregarding Texas Instruments developments, namely the 99/5 "Waxwing 5," the 99/8 "Armadillo" and so on. Well, I used to own a 99/8 with all the Hexbusperipherals (Wafertape, printer/plotter, 80-column printer, 5.25-inch DSDD disk drive/controller, 300-baudmodem, RS232 interface and the Video Interface for the CC-40) and the schematics for each including the99/8 itself. These were sent to me by Mike Bunyard of the Bunyard Manual fame with his own personalnotes regarding the "Armadillo" project. I've held in my hands one of the very few "Waxwing 5"motherboards and have a copy or the schematics and I have photographs of the motherboard. I've ownedsuch peripheral cards as the IEEE-488, EPROM programmer card, the Super Modem card, the 128K cardand TI's DSDD card.
So, as far as a hoax goes, I don't know about that! It seems to me that someone is very misinformed or Iwas told the wrong thing by my friend.

Steve Eggers, Abilene, TX

  • Like 1
Link to comment
Share on other sites

Interesting ... this code is executed right after RESET. Looks like a mapper setup. FFFE should be the part of the NMI vector (aka LOAD interrupt). We have a "crippled" 9995 here, but AFAIK it just lacks the on-chip RAM and the decrementer, but the NMI is still available.

 

 

0B64:     LI   R15,>8C02
0B68:     LI   R1,>8000
0B6C:     LI   R2,>0B8A
0B70:     LI   R3,>0010
0B74:     INCT R1
0B76:     MOV  *R2+,*R1+
0B78:     DEC  R3
0B7A:     JNE  >0B74
0B7C:     LI   R0,>6000
0B80:     MOV  R0,@>FFFE
0B84:     B    @>0032
0B88:     DATA >001F
0B8A:     DATA >F000
0B8C:     DATA >F000
0B8E:     DATA >0000
0B90:     DATA >1000
0B92:     DATA >F000
0B94:     DATA >F000
0B96:     DATA >F000
0B98:     DATA >F000
0B9A:     DATA >F000
0B9C:     DATA >F000
0B9E:     DATA >2000
0BA0:     DATA >3000
0BA2:     DATA >4000
0BA4:     DATA >5000
0BA6:     DATA >6000
0BA8:     DATA >7000
Link to comment
Share on other sites

No, at this point I'm just supposing this, but since there are many parallels to the 99/8 mapper, I think it is safe to assume that the "TMS9500" is in fact the one used in the 99/8 (TMS9995-MP9537). But we'll be able to find out with some appropriate assembly programs.

 

I just wondered at first whether the write operation to FFFE is something meaningful for the mapper, until I remembered that this is the NMI vector.

 

What we can see is that there is a mapper at 8000 - 803F which takes two words per page frame for 16 frames with 4 KiB each. The "F000" could mean unmapped or pass-through (like the "00ff0000" in the 99/8). Other than the 99/8, only the first word seems to be used.

 

Still, it does not reveal how we can get to the hidden 8K. My latest guess is that this hidden 8K is the "master DSR" for the Hexbus. Unlike the typical peripherals, the hexbus peripheral DSRs are not mapped into the 4000 space (can't, because they are buried inside their cases).

Link to comment
Share on other sites

Some more interesting stuff. I did some analysis of the ROM code:

 

 

...
Interrupt handler (Level 1)
0924:     LIMI >0000  
0928:     LWPI >83E0  
092C:     CLR  R12    
092E:     SBZ  0      
0930:     COC  @>0034,R14
0934:     JNE  >093A    
0936:     B    @>1404       Cassette driver
093A:     TB   2            Signal from VDP?
093C:     JNE  >0950       
093E:     TB   1            *
0940:     JEQ  >094C        *  99/5-specific
0942:     MOV  @>400C,R2    *  Hexbus DSR
0946:     INCT R2           *
0948:     MOV  *R2,R2       *
094A:     BL   *R2          *
094C:     B    @>0ACC       Handler done
0950:     SBO  2            Signal came from VDP

 

This essentially means that the DSR search from the original interrupt handler has been replaced by a simple branch to the address that is held at >400C. Since there is no code to turn something on or off, this looks as if there is a master DSR at 4000-5FFF at all times. I just have to update my dump program to dump this area instead of 0000-1FFF. (Fabrice, you may of course do it by yourself if you like, you got the source code from me.)

 

Also, I will send you a dump program for the speech ROM (which I used for the 99/8 already).

 

Another indication of a master DSR:

 

 

XTAB
12A0:     DATA >11AE   CSN (XML >10)
12A2:     DATA >11A2   CSN (XML >11)
12A4:     DATA >12B8   CFI (XML >12)
12A6:     DATA >1648   Lookup (XML >13)
12A8:     DATA >164E   Stack entry (XML >14)
12AA:     DATA >1642   Assign (XML >15)
12AC:     DATA >15D6   Search (XML >16)
12AE:     DATA >163C   VPUSHG (XML >17)
12B0:     DATA >1F2E   VPOP (XML >18)
12B2:     DATA >5934   DSRLNK (XML >19)
12B4:     DATA >5968   GSRLNK (XML >1A)
12B6:     DATA >1868   XML >1B

 

The DSRLNK and GSRLNK point to an address in 4000-5FFF as well.

  • Like 1
Link to comment
Share on other sites

No, at this point I'm just supposing this, but since there are many parallels to the 99/8 mapper, I think it is safe to assume that the "TMS9500" is in fact the one used in the 99/8 (TMS9995-MP9537).

I see in the 99/5 code a lot of similarities to the 99/8 mapper, so I am tempted to conclude that the TMS-9500 and the TMS-9995 MP9537 are the same CPU. But you are right, It can still remain a doubt.

 

Fabrice, you may of course do it by yourself if you like, you got the source code from me.

This week-end, I will have time to put the computer guts on the outside and I'll extract the 27128 to dump it thanks to my DATA I/O.

 

 

OK for the PHROM, thank you :-)

Edited by fabrice montupet
Link to comment
Share on other sites

Thank you!

 

All dumps done! :-) Here is the new image disk, containing all ROM, GROM and PHROM:

http://www.ti99.com/99_5/TI-99_5_DUMP_ROM_GROM_PHROM.DSK

 

Anyway, the computer will go under the knife saturday. I will examine the motherboard again, to route the traces to each chips to see if there's the way to upgrade the RAM to 64KB.This can be very interesting.

  • Like 2
Link to comment
Share on other sites

Woohoo! This is the jackpot.

 

 

000000: aa 01 00 00 40 50 00 00 40 56 40 fa 40 b4 00 00     ....@P..@V@.@...
000010: ff ff ff ff ff ff ff ff ff ff ff ff 04 60 40 26     .............`@&
000020: 00 40 04 60 00 74 00 20 04 60 00 74 ff ff ff ff     .@.`.t. .`.t....
000030: 0a de 5a 74 5a dc 5b 88 5b 32 5b 5a 59 2a 59 28     ..ZtZ.[.[2[ZY*Y(
000040: 59 2c 59 2e 59 30 59 32 04 60 4d 12 04 60 41 b0     Y,Y.Y0Y2.`M..`A.
000050: 00 00 41 42 00 00 40 5e 40 ba 03 44 53 4b 40 68     ..AB..@^@..DSK@h
000060: 40 c0 04 44 53 4b 31 00 40 72 40 c6 04 44 53 4b     @..DSK1.@r@..DSK
000070: 32 00 40 7c 40 cc 04 44 53 4b 33 00 40 86 40 d2     2.@|@..DSK3.@.@.
000080: 04 44 53 4b 34 00 40 90 40 d8 05 52 53 32 33 32     .DSK4.@.@..RS232
000090: 40 9c 40 d8 07 52 53 32 33 32 2f 31 40 a8 40 de     @.@..RS232/1@.@.
0000a0: 07 52 53 32 33 32 2f 32 00 00 40 e4 06 48 45 58     .RS232/2..@..HEX
0000b0: 42 55 53 00 00 00 41 2e 00 00 02 0a 00 03 10 13     BUS...A.........
...
001fb0: 83 73 d2 60 83 73 09 89 02 29 83 00 c6 4a 04 5b     .s.`.s...)...J.[
001fc0: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff     ................
001fd0: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff     ................
001fe0: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff     ................
001ff0: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff     ................

 

Unfortunately, no CRC16 at the end, but I guess we can live with that. We'll compare with your read-out.

 

Also, the PHROM is identical to the 99/4A speech synthesizer's ROM; I just compared the dumps.

  • Like 1
Link to comment
Share on other sites

Guys, since we now have the ROMs, there is clear way to get this emulated in MAME, but first I'd like to continue with the 99/2 in MAME, if that is OK for you. :) (I still try to keep some order in the pile.)

 

Some scientists here could try to figure out the specifics of the 99/5 in the meantime and probably try to draw schematics.

Link to comment
Share on other sites

I wonder why TI engineers have deprived of static RAM their 99/5 prototype computer, while they provided their 99/8. It's suprising because this feature has been even carefully studied. Looking at the 99/5 motherboard, there is an empty IC location near the CPU ROM, with no IC support.

99_5sram.jpg

When I traced it, I noticed that it exactelly fits with a JEDEC pinout of 2 Kb x 8 SRAM. The connexions between it and the ROM or the CPU are all valid. And holes for pins #19, #22 and #23 (for A8, A9 and A10) are even grounded to force a TMS-4016 or 6116 SRAM to be used as a 256 bytes memory.
Maybe they wanted to simplify the memory management/organization or it was just to reduce costs. I tend to this last version. If the SRAM IC has been installed on the 99/5, this project could have begun to overmuch encroach on the 99/8 Computer lands. Having said this, the 99/5 performances are very good. The ones that the 99/4A should have had.

 

The motherboard in my hands has six TMS-4416 IC for a total of 48 Kb of RAM. But the memory banks are not fully populated, there is room for two other IC that should logically offer 64Kb of RAM if the memory mapper and chip selecter permits... Again, there's no IC support. 
This option has been considered by TI: Here is a portion of the ROM code:

0010f0: 00 40 01 00 00 00 00 00 00 47 52 4f 4d 20 54 45     .@.......GROM TE

001100: 53 54 20 34 38 4b 20 43 50 55 20 52 41 4d 20 54     ST 48K CPU RAM T

001110: 45 53 54 36 34 4b 20 43 50 55 20 52 41 4d 20 54     EST64K CPU RAM T

001120: 45 53 54 50 48 52 4f 4d 20 54 53 54 39 00 01 01     ESTPHROM TST9...

I am going to see how to execute this memory test with E/A because the start menu of the 99/5 does not offer this feature.

Edited by fabrice montupet
  • Like 2
Link to comment
Share on other sites

  • 2 weeks later...

I did some research on that memory test, as you suggested. Here are some lines from GROM 0, shortly before the master title screen shows up.

 

 

...
00BA: CALL  GROM@>112C                Set characters, colors, fill screen with 00
00BD: MOVE  >0050 BYTES FROM GROM@>090E TO VDP@>0808     TI symbol
00C4: XML   >6B                       Table at 4030
00C6: DCZ   @>834A
00C8: BR    GROM@>1055                ROM test
...

 

Later, there is

 

 

1055: DST   >4000,@>8300
1059: MOVE  >0006 BYTES FROM GROM@>10F3 TO @>8302
105F: XML   >66
1061: CALL  GROM@>112C
1064: CALL  GROM@>10CE
1067: MOVE  >0001 BYTES FROM GROM@>0409 TO VREG>01
106D: MOVE  >0009 BYTES FROM GROM@>10F9 TO VDP@>014B         GROM TEST
1074: DCLR  @>834A
1076: XML   >68
1078: DST   >2000,@>834A
107C: XML   >68
107E: DST   >4000,@>834A
1082: XML   >68
1084: MOVE  >0009 BYTES FROM GROM@>10FA TO VDP@>014B         ROM TEST
108B: DCLR  @>834A
108D: XML   >67
108F: DST   >4000,@>834A
1093: XML   >67
1095: CEQ   >4B,@>000C
109A: BS    GROM@>105F
109C: MOVE  >0010 BYTES FROM GROM@>1103 TO VDP@>0147         48K CPU RAM TEST
10A3: DST   >F000,@>8000
10A9: DST   >AA55,@>0000
10AF: CEQ   >55,@>0000
10B4: BR    GROM@>10BD
10B6: MOVE  >0010 BYTES FROM GROM@>1113 TO VDP@>0147         64K CPU RAM TEST
10BD: DCLR  @>8000
10C1: XML   >69
10C3: MOVE  >0009 BYTES FROM GROM@>1123 TO VDP@>014B         PHROM TEST
10CA: XML   >6A
10CC: BR    GROM@>105F
10CE: ALL   >20
10D0: MOVE  >0008 BYTES FROM @>8300 TO @>834A
10D5: MOVE  >0008 BYTES FROM GROM@>10F1 TO @>835C
10DB: XML   >06
10DD: MOVE  >0008 BYTES FROM @>834A TO @>8300
10E2: ST    >09,@>8355
10E5: CLR   @>8356
10E7: XML   >63
10E9: CLR   @>8355
10EB: MOVE  @>8355 BYTES FROM @>835A TO VDP@>028E
10F0: RTN  
10F1: BR    GROM@>0001
10F3: RTN  
10F4: RTN  
10F5: RTN  
10F6: RTN  
10F7: RTN  
10F8: RTN  
10F9: TEXT 'GROM TEST '
1103: TEXT '48K CPU RAM TEST'
1113: TEXT '64K CPU RAM TEST'
1123: TEXT 'PHROM TST'

 

So we have some calls to XML code. The first digit "6" says that its table is at 4030. In the first snippet, there is a call to XML >6B and a test on 834A, so it seems as if the result is relevant. I did not find any indication that 834A is modified elsewhere.

 

Looking into ROM4, we see

 

 

XML table
4030:     DATA >0ADE    60
4032:     DATA >5A74    61
4034:     DATA >5ADC    62
4036:     DATA >5B88    63
4038:     DATA >5B32    64
403A:     DATA >5B5A    65
403C:     DATA >592A    66 
403E:     DATA >5928    67 
4040:     DATA >592C    68 
4042:     DATA >592E    69 
4044:     DATA >5930    6A 
4046:     DATA >5932    6B 
 
...
 
5926:     RT 
5928:     RT 
592A:     RT 
592C:     RT 
592E:     RT 
5930:     RT 
5932:     RT

 

Hence I conclude that the memory test is disabled. It would supposedly be possible with a different ROM4 (4000-5FFF), maybe a diagnostical ROM.

  • Like 1
Link to comment
Share on other sites

  • 2 years later...
On 6/26/2015 at 5:08 PM, fabrice montupet said:

Hi Bryan,
Thank you for your reply !

I contacted Erik a few years ago about his TI-99 prototypes et peripherals, he told me that he has sold most of his TI stuff, only very few documentation remained, he still has got some papers like a documentation about test data for the TI-99/4 (not A) cassette output (maybe interresting for you) , but nothing about the 99/5 ou 4B for exemple.

Bringing this thread back to life...

That's me..

I don't remember that conversation, but, I still have the same amount of documentation as in 1998. 

 

The fact is that in 1992 I was forced to downsize. I sold most of my rarities to Gerd Weismann in Germany, that's a name that Berry Harmsen recognized later. It was documentation, schematics, and some bare boards.  I have not sold anything else since then :) 

 

What I still have is a binder of progress reports about the 99/4B. There is a spec for the Hexbus disk. Yes, also a report on TI testing different consumer cassette recorders. The remainder is common things like printouts of the Disk Drive Interface and so on.

 

 

  • Like 1
  • Thanks 1
Link to comment
Share on other sites

6 hours ago, FarmerPotato said:

What I still have is a binder of progress reports about the 99/4B. There is a spec for the Hexbus disk. Yes, also a report on TI testing different consumer cassette recorders. The remainder is common things like printouts of the Disk Drive Interface and so on.

 

 

 

Are you for real??? Getting those Documents for the 99/4B would be adding a real value to the community.

 

We might have the spec already for the Hexbus Disk if it is one of the following Specifications:

- Hex-Bus™ Disk Drive/Controller 5102 Preliminary Manual
- Hex-Bus™ Floppy Disk System Product Specification
- Software Specification for the Hexbus Floppy Disk System

 

Edited by kl99
  • Like 1
Link to comment
Share on other sites

2 hours ago, kl99 said:

 

Are you for real??? Getting those Documents for the 99/4B would be adding a real value to the community.

 

We might have the spec already for the Hexbus Disk if it is one of the following Specifications:

- Hex-Bus™ Disk Drive/Controller 5102 Preliminary Manual
- Hex-Bus™ Floppy Disk System Product Specification
- Software Specification for the Hexbus Floppy Disk System

 

Here is a good faith offering. I have just setup a scanner for this purpose.

This is the beginning of a TI internal memo telling part of the 4B story.

 

4B_Design_Review_01.PDF

  • Like 4
  • Thanks 2
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...