Interesting not-ti99-related patent

[jbdigriz]

Found this the other day. Very detailed, should  be very useful for anyone working with the referenced apparatus. There is a user's manual floating around out there  (on an ftp site which is offline temporarily, due to awaiting replacement for a failing disk in an array, guy says), but little else. 

 

jbdigriz

EP0134386A2.pdf

Edited September 5, 2022 by jbdigriz

[Jimhearne]

This UK Ebay seller still has some TEN X Cobal accelerator boards for the TM990 as well as TM990/10A processor boards and a few other cards.

 

https://www.ebay.co.uk/usr/rkn704

 

No connection to the seller except i bought a TM990/10A processor board from him to play with.

I've powered the board up and it it runs the console on the serial port, i need to make a copy of the TM990 programmers panel to do any more with it.

 

Jim

[jbdigriz]

Cool. Actually, those are TILINE boards for the 990 mini, though. Don't think they'd even fit in a TM990, not that you'd want to try. 🙂 Richard may still have a KLE 13-slot TILINE chassis, though.

 

Great that you got it running. Hmm, a panel, yeah, that would be a good cloning project. or the Maintenance Diagnostic Unit (MDU)  TI FE's used. Got a guy working on a /12 needing to diagnose a faulty cache controller, so you're not the only one needing a panel. It's a CRU device IIRC, so it shouldn't be too hard. Heh, you got me thinking how to do one, now. Maybe with a terminal or GUI interface on a PC. Just interface the signal lines and hack the panel in Dave Pitt's simulator or something.

 

Edited September 6, 2022 by jbdigriz

[jbdigriz]

And there's alway following the trend and making a 990/PI, complete with a repro panel. But let's keep it in-house and make it a 990/BB, instead 🙂 There's already interest in a TILINE adaptation of the Unibone for Unibus machines... Nah, strike that, I'd be more interested in a TILINE board with some AM5729s, lots of RAM, and a prototyping/interfacing  area. Hmm... Anyway, if you want to keep things completely old-school, you could go with a TILINE 320C80/MVP board  for period-possible, if not correct.

Edited September 6, 2022 by jbdigriz

[Jimhearne]

I thought the TM990/10A board i have is one option as a processor board for a TM990 system  ? 

Does the TM990 not use TLINE as a interface.

I know the place i bought it from used it in a custom chassis for controlling oil refinery or something similar.

They did have some available but it was a bit big , i intended to make a smaller backplane with less slots, there are backplane schematics in the service manual.

At the moment i have 2 pcb's that make a single slot breakout for the processor card.

There seems to be enough information to make the programmers panel , just got to decide if i make it as original, or just use a PIC micro (don't know anything about these funny pie things) and emulate the panel.

I couldn't find the correct size connectors for the board so had to get 2 shorter ones and carefully cut and join them.

 

Actually, i said above i had the console running, now i've dug out the board i don't think i've got quite that far, it was about 18 months ago i last played with it, before moving house.

 

I'm always more interested in the real hardware than emulating it.

 

Jim

 

[+Ksarul]

Here's an 80-pin (2x40) pin connector with .100 spacing. . .and another that isn't wire-wrap style.

[jbdigriz]

7 hours ago, Jimhearne said:

I thought the TM990/10A board i have is one option as a processor board for a TM990 system  ? 

Does the TM990 not use TLINE as a interface.

I know the place i bought it from used it in a custom chassis for controlling oil refinery or something similar.

They did have some available but it was a bit big , i intended to make a smaller backplane with less slots, there are backplane schematics in the service manual.

At the moment i have 2 pcb's that make a single slot breakout for the processor card.

There seems to be enough information to make the programmers panel , just got to decide if i make it as original, or just use a PIC micro (don't know anything about these funny pie things) and emulate the panel.

I couldn't find the correct size connectors for the board so had to get 2 shorter ones and carefully cut and join them.

 

Actually, i said above i had the console running, now i've dug out the board i don't think i've got quite that far, it was about 18 months ago i last played with it, before moving house.

 

I'm always more interested in the real hardware than emulating it.

 

Jim

 

 

7 hours ago, Jimhearne said:

I thought the TM990/10A board i have is one option as a processor board for a TM990 system  ? 

Does the TM990 not use TLINE as a interface.

I know the place i bought it from used it in a custom chassis for controlling oil refinery or something similar.

They did have some available but it was a bit big , i intended to make a smaller backplane with less slots, there are backplane schematics in the service manual.

At the moment i have 2 pcb's that make a single slot breakout for the processor card.

There seems to be enough information to make the programmers panel , just got to decide if i make it as original, or just use a PIC micro (don't know anything about these funny pie things) and emulate the panel.

I couldn't find the correct size connectors for the board so had to get 2 shorter ones and carefully cut and join them.

 

Actually, i said above i had the console running, now i've dug out the board i don't think i've got quite that far, it was about 18 months ago i last played with it, before moving house.

 

I'm always more interested in the real hardware than emulating it.

 

Jim

 

 

Neat! I was wondering how you powered the board up and so on. The /10A CPU is perfect for experimenting because you have RAM on board. Main thing left would be some form of storage, but those breakout boards give you a way to interface some kind of controller. Nice work!

 

It is a 990 minicomputer TILINE board, though. The TM990 uses T-bus, I think it's called. Stuart Conner has plenty of info on that. I am getting a wild thought that you might be able interface to a T-bus floppy controller and boot TXDS or something, but that would take some serious R&D, and getting a working programmer panel would be a first step to that anyway. 

 

I did recently get a copy of the /10A General Description manual and will be scanning it in. There''s an excerpt on bitsavers that may be useful in the meantime, though.

[Jimhearne]

2 hours ago, Ksarul said:

Here's an 80-pin (2x40) pin connector with .100 spacing. . .and another that isn't wire-wrap style.

Thank you for looking for those, but these board connectors are 0.125" pitch, a lot less common.

 

Jim

 

[+FarmerPotato]

2 hours ago, Jimhearne said:

Thank you for looking for those, but these board connectors are 0.125" pitch, a lot less common.

 

Jim

 

Is this  one correct? 2x40 wire wrap.  0.125” pitch, 0.250 between rows. 

https://www.electronicsurplus.com/gc-electronics-41-858-connector-pcb-edge-40-position-80-pin-dual-a7d40ww27a3

I find odd connectors there.

[+Ksarul]

2 hours ago, FarmerPotato said:

Is this  one correct? 2x40 wire wrap.  0.125” pitch, 0.250 between rows. 

https://www.electronicsurplus.com/gc-electronics-41-858-connector-pcb-edge-40-position-80-pin-dual-a7d40ww27a3

I find odd connectors there.

Electrosurp is one of my go-to places for odd, hard to find bits and parts. That is where I found the Hex-Bus controller chips. . .

[Jimhearne]

Thank you, i will do some dimension checks and maybe get some for my bigger motherboard/backplane.

Looks like they will ship to the UK

 

Jim

 

[jbdigriz]

Got the /10A manual scanned in last night, have to redo part of it today do some post-processing and get some pics of these 990/1 boards, maybe some rom dumps. Soon, folks, soon.

Edited September 9, 2022 by jbdigriz

[Jimhearne]

Thank you, anything you have will be interesting.

 

Jim

 

[jbdigriz]

I'm thinking now that the reason this is a patent application, rather than an actual patent, apparently, is because microcoded language processor engines were not exactly a new thing by 1984. Specifically there was the WD MCP-1600 5-chip microprocessor that was microcoded with the P-system as the WD-9000 chipset in the WD Pascal Microengine, ca. 1979. Thank you, Mr. Appelt, for applying, though!

 

Interesting to note by the way that the UCSD Pascal list on groups.io woke up the other day. Laurie Boshell reports that there is a porting effort to put I.5 on the RPi Pico, The sources for the H3 distribution shipped with Microengine are sought in this regard, if anyone knows where to find a copy. Can't help but think a TI IV.0 version on the Ten X board would be slick, myself. Apologies to the fans of COBOL, which, yes, is interesting too. Very.

Edited September 11, 2022 by jbdigriz

[+Ksarul]

TI also had a full IV.11 port ready for the 99/8.

[+FarmerPotato]

I had heard that HP BASIC had microcode support on some HP portable. eevblog did an episode to open one up.

 

This HP 3000/45 manual gives source listings for its COBOL micro-programs. 

The source reads like the logic of the TI 99000 floating point (etc) macro instructions. But it is at a level beneath—operations of the ALUs.  
 

It made me laugh that the Table of Contents is giving both Physical Page and Hexadecimal Adresss. 
 

 

 

 

 

[jbdigriz]

22 hours ago, Ksarul said:

TI also had a full IV.11 port ready for the 99/8.

I haven't kept up on all the 99/8 stuff as I'd have liked. Is there a list of the TI UCSD versions somewhere? I see IV.12B in the Mame romset. IV.12 was also available for the TIPC. Looks like DX10 had UCSD version 4.1.0. as well as TI Pascal. Not sure how 4.1.0  corresponds to the IV. notation, exactly.

 

Huh, MPP has *both* an intrepreted mode using P-codes, and an native object compiler with a separate runtime. Got in a hurry and forgot to RTFM.

Edited September 13, 2022 by jbdigriz

[+FarmerPotato]

@jbdigriz Are you sure you’re reading about a shipped product? One manual , it may have been  RX, indicated that Pascal compiled code was planned. It was documenting  the binary API for procedures etc. (See AMPL directory on Bitsavers)

 

 

 

[jbdigriz]

I'm looking at http://bitsavers.org/pdf/ti/990/pascal/1603730-9701_TI_9900_The_Microprocessor_Pascal_System_1981.pdf

 

From section 1-2:

 

"Execution of Microprocessor Pascal· is in one of two modes, '·.each
supported by its own set of executive run-time libraries:

Interpretive execution - Pseudo code (or PCODE) refers to
the code resulting from execution of the Microprocessor
Pascal compiler. Pseudo code is executed in the target
via an interpreter (and for this reason, pseudo code
may be referred to as interpretive code), a program
that "looks at" each PCODE instruction in turn and
executes a set piece of code to perform the task
indicated.

 

Native code - Native machine code (or obiect code) refers
to the code resulting from execution of the
Microprocessor Pascal computer and code generator. It
is generated from the interpretive code produced by the
Microprocessor Pascal compiler. Native code is executed
directly by the microprocessor in the target
environment."

 

Same year as the RX manual in the ampl directory, so not sure how to account for the apparent discrepancy. Will look at it some more later. Busy with other things the next couple of days.

[jbdigriz]

On 9/9/2022 at 10:24 AM, Jimhearne said:

Thank you, anything you have will be interesting.

 

Jim

 

Finally got it done. I could take some more time, doctor up the covers and do a little more cleanup, but it's OCR'ed and bookmarked, and a bunch of other stuff has come up here. Maybe later.

2302633-9701C_Model_990_10A_Computer_General_Description_May85.pdf

[Jimhearne]

Wow,  thank you so much for taking the time to scan and process that, must have taken ages just adding the index.

 

Jim

 

[+FarmerPotato]

@jbdigriz Wow. That is wonderful.  I assume that where the manual says it has a 9900 microprocessor, they are describing the 99000.  It’s fascinating to read how TI used the 99000 microprocessor in a large system. 

 

For instance: 

 

1. The 990/10A auxiliary mode.   If a jumper is set, the LOAD function executes LIMI 3 and IDLE. I haven’t read the rest of the story but I guess it is interrupted when it recognizes another CPU’s macro instruction, and asserts the APP line. (I just read in my dad’s notes how a TI 980 was attached to the  990 as a speech synthesizer. Emulating 0280 for instance. Perhaps a second 990/10A could have done such a job.)

 

And! It’s spelled out in detail in Appendix A.

 

2. The rest of the status register at >1FC0. Like the CPU ID bits. These are 0 on my 99105 so I assume that was a mask option! Also TILINE timeout, memory error, mapping error. I’m not able to guess how these would get set from outside. 
 

3. CKON and CKOFF really do enable/disable the periodic interrupt from the power line frequency. (I guess not really new on the 99000.) 

 

4.  slot 1 handles most  of the  interrupts from the backplane. (Except Int14.)

 

5. Byte or Word CRU really is used to load the memory mapper. I guessed!

 

 

 

 

 

[jbdigriz]

5 hours ago, Jimhearne said:

Wow,  thank you so much for taking the time to scan and process that, must have taken ages just adding the index.

 

Jim

 

It went quickly when I realized that none of the available/affordable/acceptable gui PDF viewers/editors were up to the task of building an outline. Foxit Reader came closest, but was difficult to use without corrupting, or allowing one to corrupt, the bookmark nodes. I gave up after 3-4 tries, deleted back to before the corruption, pulled out pdftk, dumped the info file, manually edited it, and then update_info'ed the PDF. Downhill from there.  Learned lots in the process, anyway, and yeah, there's a lot of useful info in there.

[jbdigriz]

45 minutes ago, FarmerPotato said:

@jbdigriz Wow. That is wonderful.  I assume that where the manual says it has a 9900 microprocessor, they are describing the 99000.  It’s fascinating to read how TI used the 99000 microprocessor in a large system. 

 

For instance: 

 

1. The 990/10A auxiliary mode.   If a jumper is set, the LOAD function executes LIMI 3 and IDLE. I haven’t read the rest of the story but I guess it is interrupted when it recognizes another CPU’s macro instruction, and asserts the APP line. (I just read in my dad’s notes how a TI 980 was attached to the  990 as a speech synthesizer. Emulating 0280 for instance. Perhaps a second 990/10A could have done such a job.)

 

And! It’s spelled out in detail in Appendix A.

 

2. The rest of the status register at >1FC0. Like the CPU ID bits. These are 0 on my 99105 so I assume that was a mask option! Also TILINE timeout, memory error, mapping error. I’m not able to guess how these would get set from outside. 
 

3. CKON and CKOFF really do enable/disable the periodic interrupt from the power line frequency. (I guess not really new on the 99000.) 

 

4. Only slot 1 can receive interrupts from the backplane. (Except Int14.)

 

5. Byte or Word CRU really is used to load the memory mapper. I guessed!

 

 

 

 

 

Yeah, I figured it would be useful. Now we need to find the maintenance manuals, 2302634-9701 "Field Theory and Maintenance", and especially 2302635-9701 "Depot Theory and Maintenance". I'm hoping these will contain logic diagrams for the mapper and other VLSI chips, as well as schematics.

 

Looks like the /10A really wants to be the master processor if anything other than another  /10A is coprocessing. . That certainly makes sense for something like the Ten X board.

Edited September 21, 2022 by jbdigriz

[jbdigriz]

Naturally, of course, after submitting this scan to bitsavers, I noticed that rev A and rev B are already up there. Oh well, now there's rev C, and I did learn a lot in the process.