QI model schematics

[JasonACT]

Were they ever released, or has anyone reverse engineered the QI differences to the original schematics?

 

I'm just wondering if there isn't some quick mod that could be done to undo the side-port block out they implemented (which I think was only to stop 3rd party module-devices, given the GROM changes in v2.2 were designed to do just that).

[+OLD CS1]

The schematics are readily available.  Most of us here have original copies of the schematics of both (from technical manuals.)

 

Do you mean to say there is a modification to the expansion port on the QI?  That is a change I have never seen referenced, though I am no expert.  The only physical change of which I am aware in terms of expansion is the missing CRU lines from the GROM ports, and the removal of support for CS2.

[JasonACT]

Just now, OLD CS1 said:

Do you mean to say there is a modification to the expansion port on the QI?

Yes sir.  Cartridge GROM addresses are not routed to the side port, and I think CPU address >6000->7FFF is also not routed there.

 

It wasn't something commonly known BITD - at least no one ever talked about it that I knew of, only that v2.2 didn't run ROM only carts which was well known.

[+OLD CS1]

1 hour ago, JasonACT said:

Yes sir.  Cartridge GROM addresses are not routed to the side port, and I think CPU address >6000->7FFF is also not routed there.

Ah, neat.  Not routing >6000->7FFF makes sense to me on two levels: first, why have the same address ranges routed to two different locations with two different intended functions, and second, because of all the combined logic into the new VLSI.

 

I suspect one of the gurus has a digital copy of the schematics they can provide.

[+Ksarul]

The various side port cartridges all used the memory expansion and DSR space to operate because of those memory addressing choices (and by using the DSR space, they also circumvented the V2.2 operating system). It is also why the TI GSIM and EGROM Box devices all used a cartridge on a cable to route the cartridge memory space to the devices. This memory quirk dates all the way back to the Dimension 4, it isn't unique to the V2.2 consoles.

[Gary from OPA]

1 hour ago, Ksarul said:

The various side port cartridges all used the memory expansion and DSR space to operate because of those memory addressing choices (and by using the DSR space, they also circumvented the V2.2 operating system). It is also why the TI GSIM and EGROM Box devices all used a cartridge on a cable to route the cartridge memory space to the devices. This memory quirk dates all the way back to the Dimension 4, it isn't unique to the V2.2 consoles.

Grom access on all consoles before QI could still be accessed via the side port, reason why PGram works in the PEB just fine, but not on the QI.

 

V2.2 only thing it did was remove searching the 6000 space for cartridge headers, forcing you to have a grom in the cartridge.

 

On the QI the 6000 space should still be accessible on the side port, as RAMBO works on the QI just only the QI with a v2.2 it will not be searched for cartridge headers, but some QI consoles still had the normal V2 groms in them.

 

The other major change was removing CRU lines on the cartridge port which stopped things like my Dataobitics Superspace II from working even tho it includes a E/A grom.

 

The biggest change on the QI is all the glue logic moved to the custom chip, and the memory mapped devices stay internal, previously only the VDP still internal on the /4a but now the GROM does as well.

 

There no easy solution to regain 9800/9c00 outside access, the best if just side port like the Pico w sidecar is to run the lines needed from the cartridge port on a cable just 3 or 4 wires at max.

Edited March 25 by Gary from OPA

[JasonACT]

On 3/1/2022 at 8:17 AM, matthew180 said:

I sent six TI-related documents to @acadiel for scanning.  They are done now, and @arcadeshopper kindly uploaded them to WHTech.  Thanks to both of your for making this digitizing possible and available.

 

Since it can be hard to see what is new on WHTech, here is the list of additions:

 

Interface Standard and Design Guide - Western Horizon Tech

Panasonic CT-110MA CT-110MCA Color Monitor Service Manual

SAMS TI-99/4A ComputerFacts

TI_XB_Assembly_Programmers_Guide

TI994A_Home_Computer_Servicing_Manual_Oct_1983

TI-994A Console and Perhipheral Expansion System Technical Data 1049717-2 (1983)

 

Some notable mentions:

 

The Servicing Manual includes 99/4 schematics and detailed troubleshooting flow-charts.

The Console and PEB Technical Data manual includes 99/4A and QI schematics.

 

Note: Although the "TI XB Assembly Programmer's Guide" looks like a poor quality scan, I assure you it improves the readability over the physical paper original.  The original has yellowing paper, and the quality looks like the ribbon in the dot-matrix printer used to produce the print was long overdue for replacement.

 

Enjoy!

 

Ok, the 1983 tech data pdf has the QI schematics.  I was using a slightly older version, so this is great because it's so much clearer/cleaner/complete.

 

And I know for sure I never owned a QI model.  I got my v2.2 beige model for xmas '83 and it had a dual cassette cable which wouldn't have come with a QI, but it went "pop" after about 6 months and TI Australia exchanged it for a silver model.  That's the one I used for years.  But Chue does mention here, his non 2.2 QI model doesn't run the (no GROM) ROM only memory tester from my device on the side port...

On 2/25/2024 at 5:22 PM, chue said:

Here are more results from my testing as well.  I only tested a single PI, but it seems they both run at 266 mhz.

 

I tested these firmwares: 252 mhz (1MB SAMS), 260 mhz (2 MB SAMS), and 266 mhz (1 MB SAMS)

 

These all passed the SAMS burn-in (3+ passes) on both the silver TI and the beige TI.  I tested using the onboard cart functionality (without FG).

 

I hadn't mentioned it before, but I have another beige TI.  It came with a Mitsumi keyboard, which isn't working.  I did a keyboard swap so I could do some testing using the PI Pico.  The TI is a QI model (non v2.2 1981 ROM), so it has potentially some differences in the motherboard timings.

 

I had to test with the FinalGROM on the QI TI because the Pico onboard cart would not work.  I was able to choose the cartridge but after the reset, the cartridge would not show as an option.

 

In any case, I tested using the same firmwares as the other two TIs.  All passed the SAMS test.

 

I will try to test the 264 mhz 2 MB SAMS firmware next, as the others have been accounted for.

Anyway, I see there's an extra 74LS245 chip that connects the 8 data lines for the GROM and ROM to the board circuits.  You would need to remove that (U32) and join the 8 data lines from one side of the empty chip 'socket' to the other side...  Because that seems to connect to the other 74LS245 (U22) which connects to both U32 and the side port.  I.E. I think what it's doing is forcibly overriding what's coming in from the side port (not such a good thing, if you had a side port device responding at those addresses).  Seems unnecessary, extra cost on what should have been a cheaper low component count build, and probably there to just be a bit nasty.

 

Anyone else want to take a look and confirm this?

[JasonACT]

There is also this in an original console:

 

 

Which I suspect (Q500) gives a >00 reading when nothing is plugged into the cartridge port.  I use that to detect if my device should disable Pico carts by testing >00.

 

The QI model seems to just pull up the data lines:

 

 

Does this mean you will see >FF when nothing is plugged in at that space?  I may need to change my firmware, at least, at the moment, things are "safe" with my device.

 

EDIT:

 

Oh, looks like you would need to remove the 2nd set of stronger pull ups too, if you remove U32 and wire up the 8 data lines:

 

Edited March 26 by JasonACT

[JasonACT]

Another nail in the proverbial coffin, if this is really how the QI is designed, then they may have had a simple fix...

 

 

They have a "W7" wire on the enable pin - so you don't even need to remove U32 - just cut the wire and connect the pin 19 end to +5v to disable the chip.  You can then solder 8 wires from one side of the chip to the other...  Also, maybe, just cut the +5v input to the S1P2 2.7K resistor pack so even it doesn't need to be removed?

[HOME AUTOMATION]

The QI's schematics can be wanting at times...

...In another thread @helocast, suggested that he'd checked the board circuitry, and that it's the schematic that is somewhat flawed.

[JasonACT]

1 hour ago, HOME AUTOMATION said:

The QI's schematics can be wanting at times...

...In another thread @helocast, suggested that he'd checked the board circuitry, and that it's the schematic that is somewhat flawed.

Interesting 

 

Do you know what the solution to the QI+FinalGROM problem was?

[HOME AUTOMATION]

Just what I've read about it in the Sunday Papers...

Some mention of the "pull-ups" there too.

[JasonACT]

34 minutes ago, HOME AUTOMATION said:

Just what I've read about it in the Sunday Papers...

Some mention of the "pull-ups" there too.

I was about half way though some other threads leading up to the one you mentioned, thanks - I would have missed it otherwise.

 

Here's a quote from @JJB

Quote

As you can see in the left LA pic, on the QI there is a 50ns delay between GROM Select (GS) becoming low and valid data on the databus. The classic TI (right pic) has no such delay.

And here's the timing of a TI 74LS245 chip:

 

So bypassing this abomination included on the QI model may fix the original FG99/QI issue too.

 

[Gary from OPA]

1 hour ago, JasonACT said:

I was about half way though some other threads leading up to the one you mentioned, thanks - I would have missed it otherwise.

 

Here's a quote from @JJB

And here's the timing of a TI 74LS245 chip:

 

So bypassing this abomination included on the QI model may fix the original FG99/QI issue too.

 

i would if replacing it with faster 245 chip would also work, there is newer models with better switching timing then the LS version.

 

Quality wise it is smart to have 245 chip there it acts as protection buffer to prevent damage for those idiots out there inserting and removing cartridges while the power is supplied. - reason for the QI label (quality improved)

[+OLD CS1]

15 minutes ago, Gary from OPA said:

Quality wise it is smart to have 245 chip there it acts as protection buffer to prevent damage for those idiots out there inserting and removing cartridges while the power is supplied. - reason for the QI label (quality improved)

Easy there, Tex... TI said we could.

[Gary from OPA]

4 minutes ago, OLD CS1 said:

Easy there, Tex... TI said we could.

original grom cartridges are very stable, almost impossible to damage them, but other types of cartridges using LS logic and PALs, not so smart to remove and insert while the power is running.

[HOME AUTOMATION]

4 hours ago, JasonACT said:

...I would have missed it otherwise.

The FG99, in the title, doesn't help, particularly since the Search eng., doesn't do two letter words(qi).

[Gary from OPA]

13 minutes ago, HOME AUTOMATION said:

The FG99, in the title, doesn't help, particularly since the Search eng., doesn't do two letter words(qi).

plus the FG99 thread so long, unless you been reading from day one, its not easy. - i been searching thru it now trying to figure out how I update the AVR firmware, as clearly my FG99 build is using a older one, as it doesn't work as it should on my v2.2 QI

[+OLD CS1]

2 minutes ago, Gary from OPA said:

plus the FG99 thread so long, unless you been reading from day one, its not easy. - i been searching thru it now trying to figure out how I update the AVR firmware, as clearly my FG99 build is using a older one, as it doesn't work as it should on my v2.2 QI

I had mixed results (50/50) with the QI test release firmware.  One FG99 took it fine and works like a champ, the other has some issues so I just left it on the original firmware.

[Gary from OPA]

43 minutes ago, OLD CS1 said:

I had mixed results (50/50) with the QI test release firmware.  One FG99 took it fine and works like a champ, the other has some issues so I just left it on the original firmware.

how did you update the firmware, is there an easy way - i have my little TG minipro programmer, or can I do it via the SDCARD?

[+OLD CS1]

9 minutes ago, Gary from OPA said:

how did you update the firmware, is there an easy way - i have my little TG minipro programmer, or can I do it via the SDCARD?

Yup.  Just put the firmware files on the SD card.

 

https://endlos99.github.io/finalgrom99/

[Gary from OPA]

9 minutes ago, OLD CS1 said:

Yup.  Just put the firmware files on the SD card.

 

https://endlos99.github.io/finalgrom99/

Thanks. I will try that out tomorrow with some fresh coffee, and nice new SDCARD just for updating... -- Can I downgrade the firmware to older version if the test one for QI doesn't work well?

[+OLD CS1]

Just now, Gary from OPA said:

Thanks. I will try that out tomorrow with some fresh coffee, and nice new SDCARD just for updating... -- Can I downgrade the firmware to older version if the test one for QI doesn't work well?

Yes.  That is what I did.

[JJB]

8 hours ago, JasonACT said:

I was about half way though some other threads leading up to the one you mentioned, thanks - I would have missed it otherwise.

 

Here's a quote from @JJB

And here's the timing of a TI 74LS245 chip:

 

So bypassing this abomination included on the QI model may fix the original FG99/QI issue too.

 

FYI the 50ns diff turned out to be a red herring. @OLD CS1 I remember you mentioned that particular console was already a bit icky and you would test further; did you in the end? 

[JasonACT]

2 hours ago, JJB said:

FYI the 50ns diff turned out to be a red herring.

So just the pull-ups then..

 

Still, I can't see any reason for the cartridge 8 bit data bus to pass through 2 x 74LS245's for quality reasons.  To be sure, to be sure.