Beige Bug

[grayin99]

Looking to get a beige TI-99/4a.  I was looking at this one https://www.ebay.com/itm/Texas-Instruments-TI-99-4A-Computer-Working-with-cables/303749528684

What can anyone tell me by looking at it?  From the expansion port silver "fingers" I would think this is a QI module, but the screen seems to show 1981 so maybe not.  What about that keyboard?  I know there are tons of FAQs on the QI's, but I thought I would ask the experts on this one.

 

Thanks.

 

Edit, looks like I put this in the wrong forum.  Maybe someone can move it to the right section?  Sorry about that.

Edited November 5, 2020 by grayin99

[+arcadeshopper]

Qi can have 81 ROMs qi only means the hardware revision you can see the difference in the FAQ

Sent from my LM-V600 using Tapatalk

[Omega-TI]

Considering how late in the game that unit looks to be, and the square keytops, I highly suspect it has a Mitsumi keyboard.  I'd contact the seller and see if he'll take a photo of the bottom.  If you see a flat brown bottom through the vents, I'd decline.  As to the cartridge aspects, take @arcadeshopper's advice above.

[+Ksarul]

One other note: as the console has a QI motherboard, that means that two signal pins on the cartridge port are not connected, and this affects some cartridge software (only a few cartridges used either of those signals, but the FinalGROM board is one that will definitely have trouble because of it).

[grayin99]

Thanks for all the replies.  I think I will keep looking and skip this one.

[+FALCOR4]

On 11/5/2020 at 9:30 AM, grayin99 said:

Looking to get a beige TI-99/4a.  I was looking at this one https://www.ebay.com/itm/Texas-Instruments-TI-99-4A-Computer-Working-with-cables/303749528684

What can anyone tell me by looking at it?  From the expansion port silver "fingers" I would think this is a QI module, but the screen seems to show 1981 so maybe not.  What about that keyboard?  I know there are tons of FAQs on the QI's, but I thought I would ask the experts on this one.

 

Thanks.

 

Edit, looks like I put this in the wrong forum.  Maybe someone can move it to the right section?  Sorry about that.

I usually send these two pics to the seller and ask them to id which side port is on the console.  But, in either case, the Matsumi keyboard is a killer.  Also, like arcadeshopper mentioned, I have a QI with the old GROM 0 in it but it has the cartridge port limitation that Ksarul talked about.

[+acadiel]

18 hours ago, Ksarul said:

One other note: as the console has a QI motherboard, that means that two signal pins on the cartridge port are not connected, and this affects some cartridge software (only a few cartridges used either of those signals, but the FinalGROM board is one that will definitely have trouble because of it).

Are the pins just not exposed on the pop-off cartridge connector, or do they not go to the bottom female motherboard connector?  Or both?

 

[+OLD CS1]

2 hours ago, acadiel said:

Are the pins just not exposed on the pop-off cartridge connector, or do they not go to the bottom female motherboard connector?  Or both?

 

According to the QI motherboard schematics, the CRU lines are not passed to the GROM port.  I would assume the cartridge port extender has the lines for them them as I expect it to be easier just to have one of those rather than two.  But at the same time, there are two completely different constructions of the beige cartridge port support module of the upper console body, one each for the QI and standard motherboards as I show in my gallery.

[+OLD CS1]

On 11/5/2020 at 9:56 PM, Ksarul said:

One other note: as the console has a QI motherboard, that means that two signal pins on the cartridge port are not connected, and this affects some cartridge software (only a few cartridges used either of those signals, but the FinalGROM board is one that will definitely have trouble because of it).

Are the missing CRU lines the definitive cause of the FG problems on the QI?  If so, that is a bummer and I may just have to mod a couple of my QIs to restore the CRU.

[+jedimatt42]

Is there a well documented de-QI mod? Seems like a void in the knowledge base that adds to FUD.

[+OLD CS1]

3 minutes ago, jedimatt42 said:

Is there a well documented de-QI mod? Seems like a void in the knowledge base that adds to FUD.

Not of which I am aware.  I was looking at the schematic and planning to just run a couple of jumper wires up to the cartridge port.

[+FALCOR4]

From Michael Bunyard's "Hardware Manual for the TI-99/4A.....":

 

".....The CRU options were apparently never used, and were removed from the late model TI-99/4As with a Gate Array for Timing and Control."  The *CRUCLK, pin 4, and the CRUIN, pin 6, are non-connects on the cart port of the QI models.  Another, "why'd you do that TI?!" to add to the list. ?

[+OLD CS1]

1 hour ago, FALCOR4 said:

Another, "why'd you do that TI?!" to add to the list. ?

Simple: TI did not use them, so obviously they were not being used.  Never mind that third-parties were using them.  Considering the ROM-only lock-out in the v2.2 OS, that only third-parties were using the CRU is what I believe was a large part of its consideration on the design change.

[+OLD CS1]

Okay, an amendment and clarification after a more close review and comparison of the 4A and 4QI schematics.

 

First, the QI motherboard is apparently missing the CS2 remote circuit.  Pin 23 on the 9901 (P7) should go to an opto-isolator driver but this pin goes nowhere and the associated circuitry is missing, as are the connections to pins 6 and 7 on the cassette port.  Now, the QI motherboard likely still outputs to CS2 and may still accept a SAVE CS2 statement as the audio output connection goes to both CS1 and CS2.  It will just not be able to control the CS2 motor.

 

Second, the CRUOUT is routed to pin 8 on the GROM port.  However, _CRUCLK, which is CRUCLK inverted^†, does not route to pin 4 of the QI GROM port.  It appears the QI's GROM port pin 4 is NC, not even grounded.  As well, CRUIN is not routed to pin 6 on the QI GROM port, which is NC like pin 4.  (Per @FALCOR4's post above.)

 

So, CRUOUT is present on the QI GROM port, which makes sense since CRUOUT also doubles as A15, but CRUIN and _CRUCLK are not present.

 

† Note that CRUCLK is not inverted for use inside the console, but is inverted to active-low before being sent out of the console at the GROM and I/O (side expansion) ports.

[HOME AUTOMATION]

So, I can't check this... as I have no QI board...

 

The standard TI schematic, below... shows the GROM's READY/HOLD signal, ganged to all chips, including those on the cart. port.

 

 

While(below) the QI's cart. port and GROMs 0,1, seem to be isolated. Leaving GROM 2, to speak for all?

 

I drew in a little red jumper to show the discrepancy.

[+OLD CS1]

Too dark for me to pull out my schematics (on the night stand here, but "we" are sleeping.)  That would seem to prevent the cartridge port from "over-riding" the console GROMs, which I believe the GRAM Kracker can do.  Thierry's TI Tech Pages indicates that GROM 2 is a continuation of GROM 1.  If my guess is correct, it would seem silly to ignore GROM 2 in this regard and not just tie it in with the other console GROMs.

 

(hrrrmmmm or maybe not... since the cartridge port device could still see the GROM addresses, it could reasonably respond more quickly than the console GROM...)

[HOME AUTOMATION]

Interesting analysis! I was thinking that this would stop the carts from delaying powerup while booting up a cart. based GROM emu.

[+OLD CS1]

1 minute ago, HOME AUTOMATION said:

Interesting analysis! I was thinking that this would stop the carts from delaying powerup while booting up a cart. based GROM emu.

I am likely off on this one.  I edited my post just as you responded.  It might still be possible for the cartridge port to take over for the console ROMs.  I need to peruse the full sheet to figure out what is going on.

 

I think your analysis has merit, and I am thinking again toward the problems with the FG99 vs. QI.  But, I do not think the FG99 pulls any tricks during power up -- it is ready when it is ready.

[+OLD CS1]

Nope.  Looking at the schematics in full context, the RY lines of GROMs 0 and 1 go to cartridge port pin 31.  GROM 2, however, is the only GROM fed to the logic array CD40050.  This indeed does make it look like only GROM 2 ever actually talks to the system.  But in this configuration it would seem that GROMs on the cartridge port never talk to the system, either.  That does not compute.

 

In the 4A console, the GREADY signal is inverted by U508 then fed into a NAND with the GS (GROM Select) signal, the result being fed into the READY/_HOLD line (SYSRDY) which takes a convoluted path back to the 9900's READY.

 

I have to assume this contrivance is replicated in the logic array and is spit out as CPURY as that is fed to the 9900's READY.

 

The mystery: where does the GREADY line from the cartridge port and GROMs 0 and 1 go?  Is the QI schematic in error?

 

 

[+OLD CS1]

I went through the technical manual to see if this situation is addressed.  The only thing I can find is a mention that the cartridge port memory space should be considered logically connected and is not accessible from the I/O port.  The schematics play this out.  On the 4A the data lines from both ports are run together.  In the 4QI the cartridge port data lines have an additional gate before them.

[Tursi]

11 hours ago, OLD CS1 said:

I went through the technical manual to see if this situation is addressed.  The only thing I can find is a mention that the cartridge port memory space should be considered logically connected and is not accessible from the I/O port.  The schematics play this out.  On the 4A the data lines from both ports are run together.  In the 4QI the cartridge port data lines have an additional gate before them.

That's interesting... it'd be interesting to know if the side port literally prevents cartridge port addresses from being read. Seems like that'd be hard to block.

 

[+dhe]

Maybe way off base. But from what I know, reasons for QI:

   1) Cheaper to make (a real selling point, when your paying everyone a $100 rebate on a $100 dollar computer, cause you'll make it up selling software).

   2) Lock out 3rd party modules.

           The was done in the firmware, and also apparently from what this thread says, locking out of the side port.

   3) Some of the side port modules sold had really hard teeth (fingers?), and TI issued a particular warning, that they would NOT be responsible for damage to the side port, if 3rd party modules were used.

 

[+helocast]

17 hours ago, OLD CS1 said:

Nope.  Looking at the schematics in full context, the RY lines of GROMs 0 and 1 go to cartridge port pin 31.  GROM 2, however, is the only GROM fed to the logic array CD40050.  This indeed does make it look like only GROM 2 ever actually talks to the system.  But in this configuration it would seem that GROMs on the cartridge port never talk to the system, either.  That does not compute.

 

In the 4A console, the GREADY signal is inverted by U508 then fed into a NAND with the GS (GROM Select) signal, the result being fed into the READY/_HOLD line (SYSRDY) which takes a convoluted path back to the 9900's READY.

 

I have to assume this contrivance is replicated in the logic array and is spit out as CPURY as that is fed to the 9900's READY.

 

The mystery: where does the GREADY line from the cartridge port and GROMs 0 and 1 go?  Is the QI schematic in error?

 

 

You're pretty close from what I've been able to deduce all these years focusing on Figure J of the QI schematic. Despite how it is drawn, the open collector pullup on GRDY does go to all three groms (physically checked best my eyes can do on a populated motherboard). I keep meaning to publish all the oops! I've found so far, but don't really have the software to make a professional presentation here to the collective.

Way back when Tursi pointed me to the Hombre TAL stuff on WHT it became one of my main projects.

CPURY is a fairly complex equation taking GRDY, GCS, and SYSRY into account driven/sampled on @CLK2 phase and then finally NANDed with three separate conditions off of the shift-hold 8/16-bit bus conversion logic circuit.

 

Hombre Logic TAL004 Gate Array.pdf

Edited November 9, 2020 by helocast
Added an additional file from previous 2018 work

[HOME AUTOMATION]

It's starting to sound like maybe there's a software enable trick ...but that sounds like it would interfere with backward compatibility.

[+helocast]

On 11/7/2020 at 1:54 AM, OLD CS1 said:

Are the missing CRU lines the definitive cause of the FG problems on the QI?  If so, that is a bummer and I may just have to mod a couple of my QIs to restore the CRU.

For this one specifically and to a couple previous posts: Again I have visually identified (desoldered the female GROM riser on a a dead QI board) and those two traces fly by the port without connection, unlike the earlier 4A.

Not to raise what I believe a (possible) separate issue to explain the FG's behavior on a QI, but imho I really feel it is more related to how the 4A generates Vss pin 33 on the grom connector vs "real ground" as the QI is wired.

What I can't explain yet is why the ~ -.7v bias difference still works on QI built-in groms interchangeable with 4A groms. I really hope I'm not following the wrong rabbit down the proverbial hole on this research.?

Doug