WikiTI - User contributions [en]

83Plus:Hardware Revisions

2021-12-12T10:53:48Z

CVSoft: Fix naming in category

The '''TI-83 Plus''' has gone through about 50 different hardware revisions, and has evolved significantly from the 5-chip original 1999 design to the 3-chip design used presently. Careful analysis of hundreds of TI-83 Pluses has led to the following conclusions.

== Produced by Inventec ==
Inventec produced the prototypes of the TI-83 Plus and the earliest production models; serial numbers indicate production began in March 1999, which is nearly a year later than the TI-73. Hardware revisions are not comparable between factory codes I and S, but they are comparable between factory codes S and P.

Interestingly, after having switched to the 3.3-volt Inventec 6SI837 for quite some time, Inventec produced a short run of 5-chip TI REF 9815455GAII-based TI-83 Pluses as factory code I, hardware revision D.

Production expanded to Shanghai, as factory code S, in 2000. By 2002, Shanghai was Inventec's sole production of the TI-83 Plus. Only hardware revision A used the 5-chip architecture, while all later models use a 3-chip architecture around either the Inventec 6SI837 or the TI REF TI-738X ASIC. Hardware revision E changed to a LCD with poor contrast, and use of this poor-quality LCD continued at least through F. Hardware revision J synchronized hardware revision lettering with Nam Tai, as both manufacturers switched to 3.3-volt LCD drivers and eliminated the LCD daughterboard entirely. During revision K, and without bumping the hardware revision, the LCD driver was switched from the T6K04 to the NT7564H; this happened sometime during 2007.

Late in 2007, production moved from Shanghai to Pudong. The Pudong LCDs have the poorest contrast ratio of any TI-83 Plus hardware, but the hardware is otherwise identical to Shanghai-produced calculators of the same revision.

=== Inventec (I) ===
{| class="wikitable"
!colspan=8|Inventec (factory code I)
|-
! Revision
! First Seen
! Last Seen
! CPU
! ASIC
! RAM
! FlashROM
! LCD Driver
|-
| XX-X Proto
|
|
| Z84C00
| 9815455
| SRM2B256SLMX
| Am29F400BT
| T6A04
|-
| A
| Jan 1999
| May 2001
| Z84C00
| 9815455GAII
| SRM2B256SLMX
| Am29F400BT
| T6A04A
|-
| B
| Mar 2001
| Apr 2001
|colspan=3 style="text-align:center"|6SI837
| MBM29LV400TC
| T6A04A
|-
| D
|colspan=2 style="text-align:center"|May 2001
| Z84C00
| 9815455GAII
| A62W5308M
| MBM29F400TC
| T6A04A
|}

=== Inventec Shanghai (S) ===
{| class="wikitable"
!colspan=8|Inventec Shanghai (factory code S)
|-
! Revision
! First Seen
! Last Seen
! CPU
! ASIC
! RAM
! FlashROM
! LCD Driver
|-
| A
| May 2000
| July 2001
| Z84C00
| 9815455GAII
| GM76C256
| HY29F400TT
| T6A04A
|-
| B
| Aug 2001
| May 2002
|colspan=3 style="text-align:center"|6SI837
|
|
|-
| C
| Jan 2002
| May 2003
|colspan=3 style="text-align:center"|6SI837
| MX29LV400TTC
| T6A04A
|-
| D
| May 2002
| Apr 2004
|colspan=3 style="text-align:center"|6SI837
| MBM29LV400TC
| T6A04A
|-
| E
| Apr 2004
| Apr 2005
|colspan=3 style="text-align:center"|6SI837
| MBM29LV400TC
| T6K04
|-
| F
| Feb 2005
| Jun 2005
|colspan=3 style="text-align:center"|6SI837
| MBM29LV400TC
| T6K04
|-
| J
|colspan=2 style="text-align:center"|Feb 2006
|colspan=3 style="text-align:center"|
|
|
|-
| K (early)
| Mar 2006
|
|colspan=3 style="text-align:center"|TI-738X
|
| T6K04
|-
| K (late)
|
| Aug 2007
|colspan=3 style="text-align:center"|TI-738X
| S29AL004
| NT7564H
|}

{| class="wikitable"
!colspan=8|Inventec Pudong (factory code P)
|-
! Revision
! First Seen
! Last Seen
! CPU
! ASIC
! RAM
! FlashROM
! LCD Driver
|-
| L
| Dec 2007
| Apr 2013
|colspan=3 style="text-align:center"|TI-738X
| S29AL004
| NT7564H
|-
| M
| Jul 2013
| Dec 2013
|colspan=3 style="text-align:center"|TI-738X
| MX29LV400CTTC
| NT7564H
|}

== Produced by Nam Tai ==
Nam Tai produced the TI-83 Plus alongside Inventec from the first year of production all the way until the end of 2010, if not later. The baby blue Target cosmetic variant was only produced by Nam Tai. Earlier Nam Tai calculators were plagued with the LCDs missing rows/columns, but the contrast of the LCD is superior to Inventec's. Hardware variants are scattered throughout date periods and are not exactly chronological.

=== Nam Tai (N) ===
{| class="wikitable"
!colspan=8|Nam Tai (factory code N)
|-
! Revision
! First Seen
! Last Seen
! CPU
! ASIC
! RAM
! FlashROM
! LCD Driver
|-
| A
| Oct 1999
| Aug 2000
| Z84C00
| 9815455GAII
| SRM2B256SLMX
| MX29F400TTC
| T6A04A
|-
| C
|colspan=2 style="text-align:center"|Aug 2001
|
|
|
|
|
|-
| D
|colspan=2 style="text-align:center"|Dec 2001
|
|
|
|
|
|-
| E
| Feb 2002
| Feb 2003
| Z84C00
| 9815455GAII
| W24258S
| MX29F400TTC
| T6A04A
|-
| F
| Feb 2003
| Apr 2003
|
|
|
|
|
|-
| G
| May 2003
| Dec 2005
| Z84C00
| 9815455GAII
| HY62WT08081E
| MX29F400TTC
| T6A04A
|-
| H
| Mar 2005
| Dec 2005
|
|
|
|
|
|-
| I
| Mar 2006
|
|
|
|
|
|
|-
| J
| May 2006
| Sep 2006
|colspan=3 style="text-align:center"|TI-738X
| S29AL004
|
|-
| L
|colspan=2 style="text-align:center"|Mar 2007
|colspan=3 style="text-align:center"|
|
|
|-
| Q
|colspan=2 style="text-align:center"|Apr 2007
|colspan=3 style="text-align:center"|
|
|
|-
| S
| Apr 2007
| May 2007
|colspan=3 style="text-align:center"|
|
|
|-
| T
| May 2007
| Jun 2007
|colspan=3 style="text-align:center"|TI-738X COB
|
| NT7564H
|-
| U
| May 2007
| Jul 2007
|colspan=3 style="text-align:center"|TI-738X COB
| MX29LV400CTTC
| NT7564H
|-
| V
| Jan 2008
| Jan 2009
|colspan=3 style="text-align:center"|TI-738X COB
|
| NT7564H
|-
| X
| May 2009
| Sep 2010
|colspan=3 style="text-align:center"|TI-738X COB
| MX29LV400CTTC
| NT7564H
|}

== Produced by Kinpo ==
Kinpo co-produced the TI-83 Plus with Inventec Pudong for a brief period of time, but was soon the sole TI-83 Plus manufacturer. Kinpo's calculators can be quickly identified by an angled plastic bezel immediately below the LCD. While LCD contrast and durability is excellent on these calculators, the new Kinpo LCD driver used on later-production calculators (perhaps isolated solely to the Philippines factory) can cause issues with community assembly programs; earlier hardware uses the NT7564H LCD driver. Hardware revisions are comparable between Kinpo China and Kinpo Philippines.

Certain calculators produced by Kinpo Philippines, especially those sold at Walmart, began using a new serial number format, with the first letter indicating year of production (A=2019, B=2020, ...) and the next two digits indicating the week of production. Since hardware revisions do not appear to be encoded in this format, direct comparison of PCBs is necessary to determine a calculator's hardware revision.

=== Kinpo (K) ===
{| class="wikitable"
!colspan=8|Kinpo China (factory code K)
|-
! Revision
! First Seen
! Last Seen
! CPU
! ASIC
! RAM
! FlashROM
! LCD Driver
|-
| pre-A
| May 2011
| Jul 2011
|colspan=3 style="text-align:center"|
|
|
|-
| A
| Apr 2012
| Jul 2012
|colspan=3 style="text-align:center"|
|
|
|-
| B
| Jun 2012
| Apr 2014
|colspan=3 style="text-align:center"|
|
|
|-
| C
| Apr 2014
| Aug 2014
|colspan=3 style="text-align:center"|
|
|
|}

=== Kinpo Philippines (L) ===
{| class="wikitable"
!colspan=8|Kinpo Philippines (factory code L)
|-
! Revision
! First Seen
! Last Seen
! CPU
! ASIC
! RAM
! FlashROM
! LCD Driver
|-
| C
| Jan 2015
| Nov 2016
|colspan=3 style="text-align:center"|
|
|
|-
| D
| Dec 2017
| Jun 2018
|colspan=3 style="text-align:center"|TI-738X COB
|
| New Kinpo
|-
| E
| Jul 2018
| Jan 2021
|colspan=3 style="text-align:center"|TI-738X COB
| MX29LV400CTTC
| New Kinpo
|}

[[Category:83Plus:General_Hardware_Information|TI-83 Plus Hardware Revisions]]

83Plus:Hardware Revisions

2021-12-12T10:51:50Z

CVSoft: Add to a category so the page isn't out on its own

83Plus:Hardware Revisions

2021-12-12T10:48:26Z

CVSoft: Create page

83Plus:LCD Drivers

2021-09-15T04:56:56Z

CVSoft: Create page

[[Category:82:General Hardware Information|LCD Drivers]]
[[Category:83:General Hardware Information|LCD Drivers]]
[[Category:83Plus:General Hardware Information|LCD Drivers]]

The TI-83 Plus (and TI-84 Plus) can be found with one of four different LCD driver ICs, depending on the age of the calculator and its manufacturer. These are accessed by ports [[83Plus:Ports:10|10]] and [[83Plus:Ports:11|11]].

==T6A04==
The original Toshiba T6A04 is found on TI-83 Plus prototypes, in addition to the non-Parcus TI-82 and TI-83s. It may also appear on early production TI-83 Pluses from Inventec. It has a 120x64 pixel video RAM.

[https://www.datasheets360.com/pdf/5917359191271458077 Datasheet]

==T6A04A==
TI-83 Pluses from Inventec before 2004 (changed in hardware revision S-'E'), and TI-83 Pluses from Nam Tai before 2007, use the Toshiba T6A04A. It has a 120x64 pixel video RAM, and is externally identical to the T6A04. TI-82 and TI-83 Parcus calculators can also have this LCD driver.

[https://pdf1.alldatasheet.com/datasheet-pdf/view/94019/TOSHIBA/T6A04A.html Datasheet]

==T6K04==
The Toshiba T6K04 is found in Inventec calculators manufactured after 2004 (changed in hardware revision S-'E'), uses both a smaller package and a smaller external component count compared to its T6A04A predecessor, and can operate at 3.3V in addition to 5V. All TI-84 Plus series calculators prior to the Novatek switch use the T6K04. It has a 128x64 pixel video RAM (command $3F maps to a valid column).

[https://www.datasheets360.com/pdf/-6902716969693637240?query=T6K04&pqid=111710275 Datasheet]

==NT7564H==
In 2007, all manufacturers switched to the Novatek NT7564H. Unlike the Toshiba chips, it only has a 96x64 pixel video RAM, so there is no 'extra' video RAM. These LCD drivers process instructions much faster than the Toshiba drivers, and need a LCD busy delay of only a few clock cycles on a TI-84 Plus. No datasheet appears to be available.

==New Kinpo LCD Driver==
These behave similarly to the NT7564H, except that reading the status port can scramble the contents of the internal video RAM pointer. The switch from the NT7564H to this new LCD driver happened during 2018, while production of the grayscale calculators was exclusively under Kinpo Philippines.
* On the TI-83 Plus, this corresponds with factory 'L', hardware revisions 'D' and newer.
* On the TI-84 Plus, this corresponds with factory 'L', hardware revisions 'AC' and newer.
Exact hardware revision boundaries are still being determined at the time of this article being written, but they are narrowed down to +/- one hardware revision.

The LCD driver IC itself is actually situated on the LCD itself, bonded to the top of the LCD panel by gray silicone. The chip is about 1mm x 4mm. All support circuitry is on the calculator's mainboard.

===Testing for the new LCD driver===
Write a known pattern of unique bytes to a column of the LCD, checking if the LCD is busy by reading the status port (10) after every write. Read the video RAM of that column back, and see if the pattern matches. If it matches, it probably isn't the new driver. I (CVSoft) write the column three times, as there's a possibility that an entire column can get written successfully. Failure appears to depend on what's on the bus, as it's very unpredictable.

===Avoiding issues with the new LCD driver===
See the documentation for [[83Plus:Ports:10|TI-83 Plus Port 10]].

83Plus:Ports:10

2021-09-15T03:36:18Z

CVSoft: /* Comments */ Corrected information about new display drivers

[[Category:83Plus:Ports:By Address|10 - LCD Command/Status Port]][[Category:83Plus:Ports:By Name|LCD Command/Status Port]]
[[Category:83:Ports:By Address|10 - LCD Command/Status Port]][[Category:83:Ports:By Name|LCD Command/Status Port]]
[[Category:82:Ports:By Address|10 - LCD Command/Status Port]][[Category:82:Ports:By Name|LCD Command/Status Port]]
== Synopsis ==
'''Port Number:''' 10h

'''Function:''' LCD Command and Status Port

This port is used to check the status of and send commands to the calculator's LCD driver.

= TI 84 Plus C SE =
Information on the LCD driver for the TI-84 Plus C SE is on [[84PCSE:LCD_Controller|another page]].

= TI-83 Plus, TI-83 Plus SE, TI-84 Plus, TI-84 Plus SE =
'''Note: Official documentation for the LCD controller uses "X" to refer to rows and "Y" to refer to columns.'''

=== Read Values ===
The state of the LCD can be determined by reading this port.
* Bit 0: Set if auto-increment mode (commands 05 and 07) is selected, reset if auto-decrement mode (commands 04 and 06) is selected.
* Bit 1: Set if auto-increment or auto-decrement will affect the current column, or reset if auto-increment/decrement will affect the current row.
* Bit 2, 3: Not used.
* Bit 4: Set if in reset state, reset if in operating state. (Whatever that means.)
* Bit 5: Set if the display is enabled. Reset if disabled. (Note: LCD is completely turned off via [[83Plus:Ports:03|Port 03h]].)
* Bit 6: Set if the LCD will transfer 8 bits at a time through [[83Plus:Ports:11|Port 11]]. Reset if the LCD will only transfer 6 bits at a time.
* Bit 7: Set if the LCD is busy. Reset if a command can be accepted. (Not on newer models, see [[#Necessary_LCD_communication_delay|Comments]].)

=== Write Values ===
Writing this port sends a command to the LCD. Once you send a command, wait until bit 7 reads 0 before sending another. You need a minimum delay of 10 microseconds (60 cycles of whatever else on the 6MHz TI-83+, or ~150 on a TI-83+ Silver or any TI-84+ when running in [[83Plus:Ports:20|fast mode]]).

* 00 : Switch to '''6-bit mode'''. ([[83Plus:Ports:11|Port 11]] transfers 6-bits at a time.)
* 01 : Switch to '''8-bit mode'''. ([[83Plus:Ports:11|Port 11]] transfers 8-bits at a time.)
* 02 : Disable the screen. This blanks the screen and disconnects the LCD RAM from the physical screen, thus allowing you to use the LCD RAM as extra saferam. (But accessing it is so horribly slow that you should only do this if the usual saferam areas aren't enough and you don't need to display anything.)
* 03 : Enable the screen. This resumes displaying the LCD's RAM contents to the physical screen. (So any garbage you put into LCD RAM while it was off will show up, so B_CALL ClrLCDFull before this!)
* 04 : Set X auto-decrement mode. Every read or write operation from the data port will cause the LCD's internal pointer to move '''up''' one row.
* 05 : Set X auto-increment mode. Every read or write operation from the data port will cause the LCD's internal pointer to move '''down''' one row. The TI-83+ expects the LCD to be in this mode for most display routines.
* 06 : Set Y auto-decrement mode. Every read or write operation from the data port will cause the LCD's internal pointer to move '''left''' one column.
* 07 : Set Y auto-increment mode. Every read or write operation from the data port will cause the LCD's internal pointer to move '''right''' one column.
* [08~0B]: Set power supply enhancement. You'd be well advised to just leave this alone.
* [0C~0F]: Applies screen mirroring on newer calculators. Bit 0 set enables horizontal mirroring, bit 1 set enables vertical mirroring.
* [10~13]: Set power supply level. You'd be well advised to just leave this alone.
* [14~17]: Undefined.
* [18]: Cancel test mode (see Comments).
* [19~1B]: Undefined.
* [1C~1F]: Enter test mode (see Comments).
* [20~3F]: '''Set column'''. [20~2E] are valid columns in 8-bit mode, [20~33] are valid columns in 6-bit mode. [34~3F] values are also accepted, but do not correspond to a drawable area.
* [40~7F]: "Z addressing" (It just changed what phyiscal row the top row of RAM is displayed on. The LCD will wrap the bottom of it's RAM to the top of the screen as needed.)
* [80~BF]: '''Set row'''.
* [C0~FF]: '''Set contrast'''. (C0 is lowest. Note that there is a [[83Plus:RAM:8447|System RAM area (contrast)]] for this which will need to be updated if you want 2nd+Up/2nd+Down to change the contrast as expected (instead of causing a sudden jump...).)

== Comments ==
===Necessary LCD communication delay===
As mentioned, a 10 microsecond delay is required after sending the command. TI provides a routine at 000B (use normal CALL, not BCALL) that will delay the required amount of time. (Because this was not originally an LCD delay routine, on OS's prior to 1.13, calling it destroys the zero flag.)

On older calculators, this code sequence should also provide sufficient delay while also preserving the carry flag:
<nowiki> ex (sp),hl ;19
ex (sp),hl ;38
inc (hl) ;49
dec (hl) ;60</nowiki>

But I would advise you actually make use of these waste cycles somehow (process a buffer or something).

On some newer calculators, a fixed delay can still be used in combination with slowing the CPU through port [[83Plus:Ports:20|20]] and/or adjusting ports [[83Plus:Ports:29|29]] and [[83Plus:Ports:2A|2A]].

Since recently (late 2018), new calculators have started appearing which may alter the internal video RAM pointer when the status port is read. Therefore, you can't use it to check if the LCD is ready. Routines like ION FastCopy, which use the status port to check if the LCD driver is busy, will cause the displayed image to [https://i.imgur.com/8fWThqo.jpg appear garbled] on these new LCD drivers. These LCD drivers, like the Novateks, are quite fast and don't actually need a busy delay (they can be written to reliably with OTIR at 6 MHz); however, eliminating the busy delay will break backwards compatibility with Toshiba LCD drivers. The workaround is to use a fixed delay of 10 microseconds instead of reading the status port, or use a different port to measure LCD delay.

===Unused screen area===
On the T6A04 and T6A04A, the LCD driver's video RAM holds a display 120 pixels wide, even though the screen is only 96 pixels wide. On the T6K04, the video RAM is 128 pixels wide. The extra off-screen area can still be read from and written to, and is not affected by soft power-off. Newer calculators with Novatek display drivers do not have this extra memory.

===Test mode===
This is a very dangerous mode. The first test mode command (1C~1F) causes the screen to receive an unusually high amount of energy. The result is that the screen seems to go blank, save for a single blue(!) line on the TI-83 Plus, but on the TI-84-series calculators, the entire screen turns blue! Each subsequent command adds more blue lines. After you've had your fill of blueness, exit test mode by sending a exit test mode command (18), which also sets the contrast to it's highest. Now why is it dangerous? Sure it looks cool to make a supposedly black and white LCD to display BLUE, but you are pumping so much energy through the LCD to get this effect that leaving it in this mode for too long can cause long-term damage. Ever left your monitor on for a day or two and forget to set a screen saver? Remember how when you turned it off a burnt image of your screen remained? Well that's what can happen if you leave test mode on for too long. Other possible side effects include "dead pixels" (pixels that stay off no matter what you do).

So don't use test mode okay? Just pretend you didn't hear about it...

On newer calculators, the test mode commands seem to be ignored, so this is yet another reason to not rely on this function.

===Power settings===
On my (fghsgh's) calculator, the power supply command extends to 17. The boot code even initializes this setting to 17. Settings 10~13 all make the screen slowly fade to black, then fade to white, and then it stays blank. Writing any command in the range 14~17 makes the display come back to life. The power supply enhancement doesn't seem to do anything, but the (1.03) boot code initializes it to 0B.

== Cursor movement and out-of-bounds behavior ==
It is possible to set the column number out of range, either by directly writing a column number out of bounds, or by switching to 6-bit mode, moving the column to a position outside of 8-bit mode's range, then switching back to 8-bit mode.

If the column is out of range, no write is made to the display, but the column will continue to increment or decrement.

===8-bit mode===
In 8-bit mode, the column will wrap from 14 to 0 when moving right, and wrap from 0 to 14 when moving left. If the column>14, then there is no wrapping, the column will continue to increment or decrement within the full range of 0-31. Exceeding column 31 returns to column 0.

===6-bit-mode===
Same as 8-bit mode, but column wraps from 19 to 0 when moving right, and 0 to 19 when moving left.

== Example ==
Turn the LCD "off" and back "on".
<nowiki> ld a,02h
out (10h),a
ex (sp),hl ;Or do something else that takes ~60 cycles.
ex (sp),hl
inc (hl)
dec (hl)
inc a
out (10h),a</nowiki>

Safely wait for the LCD to become available
<nowiki>lcd_busy_loop:
in a,($10) ;bit 7 set if LCD is busy
rla
jr c,lcd_busy_loop</nowiki>

== See Also ==
* [[83Plus:Ports:11|Port 11]] - LCD Data Port
* [https://archive.org/details/t6a04a-datasheet LCD driver datasheet]

Talk:83Plus:Ports:10

2021-09-15T03:18:51Z

CVSoft:

I think it'd be nice if someone can find out what the missing command bytes do. :-) --[[User:Aquanight|Aquanight]] 20:04, 3 Apr 2005 (PDT)

What about the T6K04 since there are small yet important differences it may by a good Idea to include info on it. Also any one have a link to the T6K04.pdf --[[User:Jim e|Jim e]] 20:03, 11 May 2005 (PDT)
:Do a google search for t6k04 and look who comes up first :) --[[User:Kalimero|Kalimero]] 01:49, 12 May 2005 (PDT)

----

> However, a large number of calculators have been produced recently whose LCD drivers, for whatever reason, do not respond as quickly. For this reason, except where the speed loss is absolutely unacceptable (such as for grayscale display) fixed delays should '''not''' be used. Instead either use the routine at 000B, or poll this port or [[83Plus:Ports:02|Port 02h]] manually.

I removed this text as it's the opposite of the solution for making the newest display drivers work. --[[User:CVSoft|CVSoft]] ([[User talk:CVSoft|talk]]) 20:18, 14 September 2021 (PDT)

Talk:83Plus:Ports:15

2021-08-30T07:47:03Z

CVSoft: /* ld a,i interrupt detection bug */ new section

== ld a,i interrupt detection bug ==

> Also, what does this (ASIC integration of the CPU) mean for the LD A, I/R interrupt-enable-detection bug?

Per testing, nothing, the bug is still present. An early TI-83 Plus with a Z84C00 CPU didn't exhibit the interrupt-enable-detection bug, bewildering everyone present in the chat at the time. Everything else I've tested exhibited that bug.
--[[User:CVSoft|CVSoft]] ([[User talk:CVSoft|talk]]) 00:47, 30 August 2021 (PDT)

83:OS:ROMPage0C

2015-12-13T11:20:53Z

CVSoft: /* Indirect ROM Calls, 4F0C to 51F2 */ whoops

[[Category:83:OS Information|ROM Page 0C]]
This ROM page is loaded as soon as an ASM program is launched.

==Structure==

===Direct Calls, 4000 to 4F0B===
Here is the ROM Call jumptable, which looks like this :

<nowiki>
call 0033
ret
call 010F
ret
call 011B
ret
call 111F
ret
...</nowiki>

... seldom having some jumps instead of calls.

===Indirect ROM Calls, 4F0C to 51F2===

At 4F0C, the ROM Calls start to look like this :

<nowiki>
push af
ld a,XX
ld (80C8),a
pop af
jr 5188
</nowiki>

Calling a routine that uses the above format in the jump table loads a table index in ASM_IND_CALL (80C8), then goes to one of two handlers. The jr address calls either PARSER_EXEC (for addresses between 4F0C and 50A9) or GRAPH_EXEC (between 510A and 51E6). PARSER_EXEC goes to page $09 and GRAPH_EXEC goes to page $04. After calling a routine in one of those two ranges, the OS calls a handler on the corresponding page that adds 2*(80C8) (memory addresses are two bytes) to the start address of the address table, puts that value on the stack, and rets to that address.

There is a region between 50B2 and 5109 that contain normal calls, like those around 4000. GRAPH_EXEC, PARSER_EXEC, and IO_EXEC are also normal calls.

===51F3 to 52E8===
This has code that has something to do with something, maybe parsing/handling/executing nostub assembly?

===52E9 to 7FFF===
This part only has FF FF FF FF FF ... until the end (7FFF).

83:OS:ROMPage0C

2015-12-13T11:17:38Z

CVSoft: Ugly spacing is ugly

83:OS:ROMPage0C

2015-12-13T11:16:29Z

CVSoft: Lots more documentation, mostly about indirect ROM calls.

[[Category:83:OS Information|ROM Page 0C]]

This ROM page is loaded as soon as an ASM program is launched.

==Structure==

===Direct Calls, 4000 to 4F0B===
Here is the ROM Call jumptable, which looks like this :

<nowiki>
call 0033
ret
call 010F
ret
call 011B
ret
call 111F
ret
...</nowiki>

... seldom having some jumps instead of calls.

===Indirect ROM Calls, 4F0C to 51F2===

At 4F0C, the ROM Calls start to look like this :

<nowiki>
push af
ld a,XX
ld (80C8),a
pop af
jr 5188
</nowiki>

Calling a routine that uses the above format in the jump table loads a table index in ASM_IND_CALL (80C8), then goes to one of two handlers. The jr address calls either PARSER_EXEC (for addresses between 4F0C and 50A9) or GRAPH_EXEC (between 510A and 51E6). PARSER_EXEC goes to page $09 and GRAPH_EXEC goes to page $04. After calling a routine in one of those two ranges, the OS calls a handler on the corresponding page that adds 2*(80C8) (memory addresses are two bytes) to the start address of the address table, puts that value on the stack, and rets to that address.

There is a region between 50B2 and 5109 that contain normal calls, like those around 4000. GRAPH_EXEC, PARSER_EXEC, and IO_EXEC are also normal calls.

===51F3 to 52E8===
This has code that has something to do with something, maybe parsing/handling/executing nostub assembly?

===52E9 to 7FFF===
This part only has FF FF FF FF FF ... until the end (7FFF).

83Plus:History of TI-8x hardware

2015-06-04T09:24:07Z

CVSoft: so apparently the TI-82 exists. also some minor edits

[[Category:83Plus:General Hardware Information|History of TI-8x hardware]]

TI's first graphing calculator was the TI-81, released in 1990. The TI-81 was also TI's first Z80-based graphing calculator. It provided full scientific calculator functionality, equation graphing, matrix math, basic statistics, and user programs. (The latter functionality recalls the pocket computers that had proliferated in the 80s.) Interestingly, the backup battery feature was not present until February 1992, but mold design indicates the feature was planned.

In 1993, TI introduced the TI-82, bringing some features from the TI-85 into the TI-81 market. It introduced some new statistics functions, allows 10 function/polar graphs instead of the TI-81's four, and updates the TI-BASIC implementation to include most functions seen in the TI-83 series. It also added a link port, and was the first calculator compatible with the TI CBL (which would not be announced for another ten months). The first two months of production (May and June 1993) saw 9 different ROM versions, and at least one bug noticeable to the user was fixed during that period. Assembly programming was a challenge on the TI-82, and it was not until 1997 that a shell was released.

The immediate predecessor to the TI-83 Plus was the TI-83, released in 1996. The TI-83 featured complex numbers, graphing of both equations and data, custom number lists, improved matrix operations, an even more sophisticated statistics package, and a slightly-updated BASIC. TI also updated the LCD to a higher contrast display, with black pixels on a white background instead of the TI-82's blue pixels on a green background. Perhaps most importantly, TI officially supported assembly programming with the TI-83, releasing an include file for assemblers and documentation on the TI-83's internal functionality.

In 1999, the TI-83 Plus came. It featured all of the same math functionality, but added support for operating system upgrades, persistent installable applications, and nonvolatile long-term user data storage (the archive); the assembly programming implementation was updated to support 'squished' programs, greatly reducing the binary sizes.

The TI-83 Plus originally was based around five main integrated circuit (IC) chips: A CMOS Z80 CPU, a 32 KB SRAM chip, a 512 KB flash chip, an LCD driver with embedded VRAM, and a glue-logic ASIC (Application-Specific Integrated Circuit) that shuttles data between the other chips as well as interfacing with the keyboard and serial port (called the DBUS). TI calls the ASIC the "gate array". All logic was 5.0 V TTL logic.

The TI-83 Plus caught on in time. The TI-83 Ion shell was ported, and games targeting both models were common for many years. TI sold an SDK (which is now free) for applications for many years, hoping to make money off an app store with pay-apps like CellSheet (an Excel-like spreadsheet app). The payment system was enforced through the calculator's certificate system. When a user started buying apps, the user would be issued a cryptographically-signed certificate update (often embedded into the app itself) containing one or more keys unique to that user, allowing the user to install apps signed for that user and that user only. In turn, pay apps would be signed differently than free apps.

Not long after the TI-83 Plus came out, TI started working on the next calculator, the TI-83 Plus Silver Edition, released in 2001. It integrated the CPU into the gate array, and added some significant functionality to the gate array. TI also switched over to 3.3 V CMOS logic, yet failed to lower the voltage specification to 3.3 V. (Indeed, if you remove the OS's battery level checks, you can run the calculator without issue with 3 AAAs instead of 4.) The functionality added to the advanced gate array includes support for 32-256 K of RAM, support for 1-4 MB of flash, more precise and sophisticated write and execution permissions, variable CPU speed, a hardware DBUS assist so bits need not be bit-banged, support for more precise battery level measurement, support for a 32768 Hz quartz crystal oscillator used for both interrupts and timing, MD5 hardware-assist, and support for wait-states on the LCD, RAM, and flash.

The TI-83 Plus SE was a sort of beta version of the TI-84 Plus series, with limited production intended from the start. Throughout its run, the TI-83 Plus also continued to be produced. The actual production units all feature 128 K of RAM and 2 MB of flash. For unknown reasons, the boot code (even in the TI-84 Plus C SE) never sets the correct RAM size, making every other RAM page executable. TI initially thought the TI-83 Plus SE could be operated at up to 25 MHz, and included 6, 15, 20, and 25 MHz CPU speed options, but production units were capped at 15 MHz. Modern overclocking tests show instability above 22 MHz, so it seems likely that TI capped the CPU speed at 15 MHz just to be safe. The TI-83 Plus SE also came bundled with many if not all of TI's official pay-apps.

In time, some apps using the extra hardware, RAM, and faster CPU of the TI-83 Plus SE were produced, but most programmers were unwilling to limit themselves to the TI-83 Plus SE only, and so such software remained rare. Michael Vincent probably produced the most notable programs: Omnicalc's RAM restore feature, which stored a backup of RAM into the extra RAM after every power-off; and Emu8x, which could emulate other Z80 models, including the TI-86 (with less RAM), TI-83, and TI-82.

In the early 2000s, Michael Vincent discovered a flaw in the TI-83 Plus SE boot code that allowed bypassing the OS cryptographic signature check. He started making an operating system, and TI shut him down. Future models would fix that particular flaw. In time, however, additional flaws in the OS and boot code would be discovered, allowing user programs to write and erase flash at will. Many of these flaws remain unfixed, and Brandon Wilson wrote a program that allows arbitrary unsigned OSes to be loaded.

In 2004, TI discontinued the TI-83 Plus SE, and introduced the TI-84 Plus and TI-84 Plus SE, based on the TI-83 Plus SE hardware. The two new models introduced a USB port, a real-time clock, and eight GPIO (General Purpose Input/Output) pins. All hardware in the TI-84 Plus and TI-84 Plus SE is completely identical, except for the flash chip; the boot sector in the flash chip initializes the flash chip size port differently. After that, the boot code and OS use that port to differentiate the models. (It's possible to query the flash chip itself for its size using autoselect commands, but TI doesn't do this. Instead, the boot sector is hard-coded to emit the correct value for a particular chip size.)

Work on the TI-84 Plus USB produced some interesting USB demos, but overall USB never caught on. Similar to the TI-83 Plus SE, programmers were reluctant to produce TI-84 Plus-only software.

For a few years, TI bought below-spec LCD drivers for the TI-84 Plus series. These drivers operated far slower than the specs said they should, and calculators containing them could not run many games unless users ran a program to configure the correct wait-state count.

Sometime in 2006ish, TI silently issued a revision of the TI-83 Plus hardware, moving it to 3.3 V logic and integrating the RAM and CPU into the gate array.

TI also issued a revision to the TI-84 Plus series hardware, in which they reduced the RAM from 128 K to 48 K, breaking any TI-83 Plus SE software that used the extra RAM. TI's own software only used one of the extra six RAM pages, so their software was not broken.

In 2009, Benjamin Moody realized that PC hardware had advanced to the point that 512-bit RSA keys could be factored in just a few weeks, and proceeded to do so. This generated some controversy, as it was the first time cryptographic keys had been factored, instead of reverse-engineered through weaknesses in implementation. Eventually, TI responded by upgrading the TI-84 Plus OS key to 2048-bit RSA without comment. The new signature check now takes over seven minutes (compared to less than thirty seconds for the 512-bit check) in part due to an inefficient implementation. The new RSA check prompted a new round of reverse-engineering of the TI-83/84 Plus series hardware. It was long known that the flash chips TI uses can lock certain sectors so they cannot be modified or erased, and the community long assumed that the boot sector was so protected. It turns out, however, that the TI-83 Plus SE and TI-84 Plus series never used that ability, and the new 3.3 V logic TI-83 Pluses don't either. Instead, the gate array uses the flash chip size port to determine what sectors to lock. Therefore, by changing the flash chip size port, the boot sectors could be unlocked and modified. (The TI-83 Plus is exempted from this hack because it lacks a flash chip size port.)

The TI-84 Plus series went without major updates for many years, until OS 2.5x in 2010. This update added MathPrint. MathPrint heavily modified many screen-related BCALLs, making their behavior completely incompatible with existing 3rd party software unless the software disabled MathPrint.

In spring of 2013, TI slapped a new color LCD driver onto the old 128 K TI-84 Plus gate array, and called the new model the TI-84 Plus C SE. They upgraded the flash chip to 4 MB, because using 8 MB chips would require 9-bit page numbers. The model also got a lithium-ion rechargable battery, because the color LCD would eat AAAs far too quickly. Meanwhile, the CPU speed remained unchanged. The new color LCD has too many pixels, and the 15 MHz CPU unpipelined CPU is too slow to write to all of them quickly, so overall the new model feels very sluggish. Even worse, TI did not very well optimize their screen routines.

Thus far, the TI-84 Plus C SE community has failed to produce the breadth of software the TI-83 Plus has. Some interesting games exist, but neat utilities like Symbolic and Omnicalc remain absent.

User:CVSoft

2014-10-19T07:50:43Z

CVSoft: Update links

Hi, I'm CVSoft. Here's my website [http://cvsoft.byethost32.com/] with all of my data.

I've moved to a different website, at [http://internetexploder.weebly.com/].

User:CVSoft

2012-11-03T02:49:36Z

CVSoft: Created page with 'Hi, I'm CVSoft. Here's my website [http://cvsoft.byethost32.com/] with all of my data.'

Hi, I'm CVSoft. Here's my website [http://cvsoft.byethost32.com/] with all of my data.