DATAMATH  CALCULATOR  MUSEUM

Project X

• The Setting
• Project X
• LCD III Project
• Enter TI-85
• Project Milestones
• Market Research
• Enter TI-66, TI-76, and TI-86
• Troubling Waters
• Cancelation of the TI-88
• Lessons Learned
• Quanto Costa?
• Impact on the ALC Program
• Sad Ending
• Legacy

• Thank you

• The Setting

The landscape of Programmable Pocket Calculators was clearly laid out between 1974 and 1976: Hewlett Packard respected as leader in innovation and marketing their perfectly engineering products in the upper price segment while Texas Instruments following up within roughly one year and appealing a broader market through aggressive pricing. This storyline changed dramatically on May 24, 1977 when Texas Instruments introduced the now legendary TI Programmable 59 and its sibling TI-58, featuring a novelty, the Solid State Software ModulesTM with up to 5,000 program steps. A small lid on the backside of the TI-58/TI-59 allowed for easy access to the modules roughly the size of thumb tip.

The Master Library, known as "Module -1-", with 25 different programs was included with the TI-58 and TI-59. Twelve additional Solid State Software Modules, known as "Module -2- to -13-" were available from Texas Instruments as Standard Libraries. Due to this innovative module concept both the TI-58 and TI-59 gained a lot of attention and dozens of Modules were released for applications ranging from insurance fee calculations to tax calculations to screw joint calculations, HVAC design, pool water analysis and even flight computers for the USMC Harrier.

As of May 1977 the market of Programmable Pocket Calculators was dominated by Hewlett Packard and Texas Instruments, having low-end, mid-range, and high end products in their respective portfolios:

Introduction Product MSRP Form Display Operation,
Programming
Program Size
(Merged?)
Data Size
(Backup?)
Archive Ports Battery Operating
Time
Year Mth Company Model Size Techn. Logic PGM Mode Steps M Memories B
1974 1 HP HP-65 $795 (1974) P 15 (10+2) 7-Seg LED RPN Keystroke 100 N 9 N Mag. Cards   3*AA NiCd <10h
1975 1 HP HP-55 $395 (1975) P 15 (10+2) 7-Seg LED RPN Keystroke 49 N 20 N     3*AA NiCd <10h
8 HP HP-25 $195 (1975) P 12 (8+2) 7-Seg LED RPN Keystroke 49 Y 8 N     2*AA NiCd <10h
9 TI SR-52 $395 (1975) P 14 (10+2) 7-Seg LED AOS Keystroke 224 N 20 N Mag. Cards   3*AA NiCd <10h
1976 5 TI SR-56 $380 (1976) P 14 (10+2) 7-Seg LED AOS Keystroke 100 N 10 N     3*AA NiCd <10h
7 HP HP-25C $200 (1976) P 12 (8+2) 7-Seg LED RPN Keystroke 49 Y 8 Y     2*AA NiCd <10h
7 HP HP-67 $450 (1976) P 15 (10+2) 7-Seg LED RPN Keystroke 224 Y 26 N Mag. Cards   3*AA NiCd <10h
8 Casio fx-201P   P 13 (8+2) 7-Seg VFD CHAIN Keystroke 127 N 11 N     4*AA <10h
9 Casio pro fx-1   P 13 (8+2) 7-Seg VFD CHAIN Keystroke 127 N 11 N Mag. Cards   4*AA <10h
10 National
Semicond.
7100 never
released
P 13 (8+2) 7-Seg LED AOS BASIC like 240 Y 37 Y External 1 tbd <10h
12 Casio fx-202P   P 13 (8+2) 7-Seg VFD CHAIN Keystroke 127 Y 11 N     4*AA +
2*LR44
<10h
1977 5 TI TI-57 $80 (1977) P 12 (8+2) 7-Seg LED AOS Keystroke 50 Y 8 N     2*AA NiCd <10h
5 TI TI-58 $125 (1977) P 12 (8+2) 7-Seg LED AOS Keystroke 480-0 N 0-60 N     3*AA NiCd <10h
5 TI TI-59 $300 (1977) P 12 (8+2) 7-Seg LED AOS Keystroke 960-160 N 0-100 N Mag. Cards   3*AA NiCd <10h

Note: Table excludes the lowest end of “programmable” calculators featuring only simple key-reply operations and sporting only one memory.

• Project X

Understanding the game of cat and mouse with Hewlett Packard, Texas Instruments initiated already in Summer 1977 the "Project X" with clear objectives in mind:

• To provide upgrade leadership product to replace TI-59
• Expand concepts of problem solving via calculator device
• Incorporate:
   - State-of-the-art Technologies
   - System Design
   - System Expansion
• To simplify ease of use and provide market expansion among professionals and end users:
   - Technical Professionals
   - Non-Technical Professionals

The ambitious requirements for "Product X" later called "Product 225" were finalized in January 1978:

• Equation Operating System (EOS)
   - Equation can be entered without execution
   - Display will scroll when full
   - Equation can be edited (Backstep, Insert, Delete)
   - Equation is saved to be executed or repeated with new data
• 16 Character 5*7 dot matrix display
   - Full alphanumeric display
   - Equation trace and prompting in the display
• More flexible user memory
   - Basic memory is 1,000 program steps or 125 data registers with partitioning by the user
   - Solid State Software approach (CROM) has been expanded to include Drop-in RAM (CRAM).
   - CROMs and CRAMs are interchangeable and two slots are available on the calculator
   - Basic memory can be expanded to 3,000 steps of 375 data registers
   - CROMs will be 15,000 program steps, up to 99 programs per CROM, up to 10,000 steps per program

Goal: Define an operating system or "language" for Product X that allows the user to enter problems as they are normally written

• Equation Operating System (EOS)
• More powerful than the TI-59
• Much easier and more obvious to use than the TI-59
• Unary and binary operations will be entered as they are written
   Example: SIN 30 =
• Automatic variable assignment
   Examples: 30 => B  10 => C   SIN B + C => A  (A = 10.5)
• Implied multiplication
   Example: 2A + B = Calculates (2 x A) + B
• Integer powers of variables
   Example: A5 means the value stored in "A" is raised to the 5th power

Or in other words: The capabilities of an SR-60A desktop calculator fitting into the package of a TI-30, operating 150 hours on the charge of a small AA-sized rechargeable battery. It is obvious that this challenging goal could not be realized with the TMC501E building blocks used with the TI-59 and tracing back to the SR-50 introduced already in January 1974 and manufactured with a power hungry PMOS process.

• LCD III Project

Understanding that sooner or later the complete portfolio of electronic calculators, educational toys and even products like travel clocks needed to be ported to a more recent manufacturing process, Texas Instruments decided to skip the common polysilicon gate NMOS process and jump right into a state-of-the-art CMOS process for future products, including "Product X".

Caltech2_large.jpg (59964 Byte)Texas Instruments did not only invent the Integrated Circuit, but defined with the portable electronic calculator its first killer application consequently leading into "Calculator Chips" with two different architectures:

• Building Blocks
   - Register-based Arithmetic Chip
   - External ROM, RAM, and peripherals
• Single-chip Calculator Chips
   - On board ROM
   - On board RAM
   - Integrated I/O decoding and buffering
   - Simple ALU

As of Summer 1978 the calculator chips in use could be divided into two categories:

Register Processors

• Operations are done on entire registers
   - Usually 64-bit registers (16 Digits * 4 Bits)
• Instruction specify one of several "masks" to operate on portions of the register: Mantissa, Exponent, or All
   - Operations include Decimal Additions with automatic carry between the digits
• No pointer system for data addressing
   - The instruction specifies which registers
   - Addressing individual digits is not possible
• Instruction cycles are typically 80-100 us
• Excellent for fast arithmetic calculations
• Not flexible enough for specialty products since digit manipulation is not possible

Examples: TMC0500 (TI-59), TMC1500 (TI-57), TMC0920 (TI-1050), TP0310 (TI-1030)

Digit Processors

• Operations are done on 4-bit digits
   - A 4-bit Binary Adder and 4-bit Accumulator (temporary holding register) are the basic tools
• A pointer system locates the 4-bit piece of data to be operated on
   - One pointer selects one of the registers
   - One pointer selects one digit in that register
• Instruction cycles are typically 15-30 us
• Execution of arithmetic operations is typically slower than register processors since these calculations
   require a subroutine instead of a single instruction
• Extremely flexible
   - Effective register size, number of Flags, and I/O are all controlled by software
   - Useful as controllers and for processing odd data structures (printers, specialty products)

Examples: TMC1000 (SR-16), TMC0980 (TI-30), TMC1980 (TI-1680), TMC0260 (TI-5040), TMC0270 (Speak & Spell), TP0320 (TI-50), TP0455 (TI-55-II)

Looking into the advantages and disadvantages of the two architectures and already experienced with porting the TMS1000 microcontroller-based TI-30 calculator chip TMC0980 to CMOS technology for the TI-50 resulting in the TP0320 chip, Texas Instruments defined the "LCD III" project with very ambitious goals:

• Convert calculator products from basic consumer through TI-57 to long battery life, LC-Displays
• Provide the basis for a high-performance TI-59 replacement, aka "Product X"
• Afford the flexibility to produce new, non-calculator consumer products, including complex timekeeping functions
• Non-volatile program and data storage for most calculator models
• Low cost per function
• Long design service life

Engineering teams at Texas Instruments decided early in 1978 for a very innovative approach with a modular design:

• Layout: Each processor is a subset of the next larger circuit. Little or no layout work is required to generate the
   cores of the smaller version of the system.
• Cell design: Wherever possible, cells with similar functions will be designed as identical units
• Architecture: I/O and special functions are treated as memory locations to eliminate special instructions. ROM and RAM
   addressing is paged to permit easy expansion with minimal burden on smaller versions.

The Approach Rationale of the LCD III Project Team stated in January 1978 for their stakeholders:

• Development of several chips from one design reduced money, time, and manpower requirements
• Modular design greatly reduces development time and cost for future custom processor requirements
• Common instruction set reduced algorithm development time by utilizing programmer learning and previously designed routines

Within the LCD III Project a total of three chip sizes A, B, and C where defined, differing mainly in ROM and RAM capacity and each chip could include additional modules for timekeeping functions and drivers for LC-Displays.
The original naming convention for the three A, B, and C chips defined:

• A: 1k ROM
• B: 2k ROM
• C: 3k ROM
• [NN]: REGISTER SIZE
• F: FAST ROM
• T: TIME KEEPING

Example: C22FT – 3k ROM, 22*16*4-bit RAM, Fast ROM, and Time Keeping.

These three chips were meant to cover the whole portfolio of calculator products to be converted to low-power, LC-Display based successors and some future products, too:

Products Chip
TI-1000, TI-1025 A
Little Professor A
Dataman B
Calculator/Clock B + TIME
Executive Reminder B or B + C
Travel Alarm B + TIME
Ext. RAM
TI-30, Business Analyst B
TI-55, TI-57 C
"Product X" Controller C

In general:

• Chip A replaces TMC1000, TMC0970, TMC1990, TMC0920, TP0310
• Chip B replaces TMC0980, TMC1980, TMC1100, TP0320
• Chip C replaces TMC1500

Each version was meant to be more powerful than the circuits it will replace and the Project Team defined the performance goals of the LCD III processor accordingly:

• Efficiency of Digit Processor
   - Well structured logic arrays with tight design rules, minimize bar area
   - Bit/Byte data access
   - Software definable I/O Bus structures
• Speed of a Register Processor
   - One cycle register arithmetic
   - Direct memory addressing

The Project Team chose a very interesting design approach to achieve fast execution times with a Digit Processor with three innovations:

• Fast ROM
   - 128*13 bits – Small size allows faster ROM access
   - Basic add and Shift Loops are placed in Fast ROM
   - Reduces Multiply time by 50%
• Dual-Register Pointer System
   - Each operand in an addition has a separate pointer
   - No wasted time moving single pointer between two registers
• Hold Program Counter
   - One Instruction Loops

Early benchmarking of the LCD III design versus TMC0500 showed clearly the advantages and disadvantages of the two different design approaches:

Benchmark LCD III TMC0500
ROM Fast ROM
Instruction Cycle 15 us 5 us 83 us
Multiply Subroutine   30 ms 27 ms
Flag Test 30 us   155 us
Memory Access    
- Program Step 0.65 ms 0.18 ms 0.5 ms
- User Memory 6.1 ms 4.1 ms 1.5 ms
- Absolute GOTO 3.6 ms 0.55 ms 36 ms

The advantages of the LCD III architecture based on the instruction set capabilities more than outweighed the performance deficiencies with respect to the TMC0500 building blocks:

Instruction Set LCD III TMC0500 TMC0980
Direct Memory-Addressing Flags
Bytes
Flags
Registers
No
Memory-Mapped I/O Yes No No
Logic Instructions Yes No Limited
Indirect ROM-Addressing PC Indexing Yes No
Indirect RAM-Addressing Yes No Limited
ROM Paging 1k 1k 128
Conditional Branching Set
Reset
Set
Reset
Set
Subroutine Levels 6 1 1

• Enter TI-85

The naming convention of the three chip sizes A, B, C was changed in Fall 1979 to the more common TP04XX style:

• TP0480: C22FT with LCD Driver (TI-55 replacement)
• TP0475: C22FT without LCD Driver (Product X)
• TP0470: C22F without LCD Driver (Product X)


Note: While the original "Product X" incorporated one TP0475 plus two TP0470 chips and schematics of a "TI-85" feature one TP0475 plus one TP0475/TP0485/TP0495 chip, the actual TI-88 sports two TP0485 chips labeled CD2901 and CD2902.

Based on the available information we triangulated the specifications of the TP0470/TP0485 chips accordingly:

• TP0470: 3k Bytes ROM, 128 Bytes Fast ROM, 22*16*4 Bits RAM, no Timekeeping, no LCD Driver
• TP0475: 3k Bytes ROM, 128 Bytes Fast ROM, 22*16*4 Bits RAM, Timekeeping, no LCD Driver
• TP0480: 3k Bytes ROM, 128 Bytes Fast ROM, 22*16*4 Bits RAM, no Timekeeping, LCD Driver
• TP0485: 3k Bytes ROM, 128 Bytes Fast ROM, 22*16*4 Bits RAM, Timekeeping, LCD Driver

The "Project X" demanded not only the development of these 4-bit Microcontrollers but new memory chips (both RAM and ROM) and cascadable display drivers for the alphanumeric display, too. The final design of the TI-88 included the following main components:

• CD2901 (TP0485) Timekeeping, Key Scan and I/O Controller
• CD2902 (TP0485) Master Controller
• 2*TP0530 Cascadable Display Drivers
• 2*TP0531 On-board Read/Write Memories
• CD5402 (TP0532) On-board Read Only Memory
• Plug-in Memories which may be either Read/Write Memory with 2*TP0531 or Read Only Memory with TP0532
• SN77203 Display Interface Voltage Controller Chip

The only existing component reused from the TI-59 is the keyboard which in succession created a lot of headaches with the TI-55-II due to the lower voltages of the new designs (3V versus 16V).

The scope of "Project X" included not only the Programmable Calculator with its CRAM and CROM modules, today known as TI-88, but two additional peripherals connecting with a 2-pin Peripheral I/O connector to the calculator:

• PC-800: Thermal Printer for printing 16 characters per line at 3 lines per second
• CA-800: Cassette Interface for archiving of both programs and data with a tape recorder

• Project Milestones

With the "Project X" starting in Summer 1977 and having in 1980 around 55 FTE (Full-time equivalent) assigned to the project (around 30 FTE in the Consumer Products Group and around 25 FTE in the Semiconductor Group), one would expect that the market introduction of the TI-59 successor was planned for May 1981 (Summer CES).

Four years would have been a rather long product cycle in the early days of pocket programmable calculators:

• HP-65 (30 months) HP-67 (36 months) HP-41C
• SR-52 (20 months) TI-59 (48 months) Product X

Note: The earliest TI-88 we discovered so far was manufactured in September 1981 or about 40 months after the introduction of the TI-59, most of the units were manufactured either in May 1982 or August 1982.

While Texas Instruments had in 1977 with Hewlett Packard only one serious contender in the market of consumer programmable calculators, by 1981 the market looked completely different. Not only did Hewlett Packard introduce its flagship calculator HP-41C, competing directly with the "Product X", but Sharp and Casio launched with their PC-1211 and FX-702P very successful "Pocket Computers" introducing BASIC programming instead of keystroke programming and Matsushita, also known as Panasonic, created the market for Hand-Held Computers with their Panasonic HHC.

As of May 1981 the market of Programmable Pocket Calculators and Pocket/Hand-held Computers was much more competitive than in May 1977:

Introduction Product MSRP Form Display Operation,
Programming
Program Size
(Merged?)
Data Size
(Backup?)
Archive Ports Battery Operating
Time
Year Mth Company Model Size Techn. Logic PGM Mode Steps M Memories B
1977 7 HP HP-29C $195 (1977) P 12 (8+2) 7-Seg LED RPN Keystroke 98 Y 30 Y     2*AA NiCd <10h
8 Sharp PC-1200 $70 (1977) P 14 (10+2) 7-Seg VFD AOS Keystroke 64 N 6 Y     2*AA +
2*LR44
<10h
8 Sharp PC-1201 $90 (1977) P 14 (10+2) 7-Seg VFD AOS Keystroke 128 N 12 Y     2*AA +
2*LR44
<10h
9 Commodore PR-100 $60 (1977) P 12 (8+2) 7-Seg LED AOS Keystroke 72 N 10 N     3*AA NiCd <10h
10 Sinclair Enterprise
Program.
$49 (1978) P 8 (5+2) 7-Seg LED AOS Keystroke 79 Y 7 N     9V PP3 <10h
1978 5 HP HP-33E $100 (1978) P 12 (8+2) 7-Seg LED RPN Keystroke 49 Y 8 N     2*AA NiCd <10h
5 HP HP-38E $120 (1978) P 12 (8+2) 7-Seg LED RPN Keystroke 99-8 Y 7-20 N     2*AA NiCd <10h
1979 1 TI TI-58C $110 (1979) P 12 (8+2) 7-Seg LED AOS Keystroke 480-0 N 0-60 Y     3*AA NiCd <10h
6 Sharp EL-5100 $100 (1979) L 24 Dot M. LCD AOS Formula 80 N 11 Y     3*LR44 100-1,000h
6 Sharp EL-5101 $80 (1979) L 16 Dot M. LCD AOS Formula 48 N 6 Y     3*LR44 100-1,000h
6 Sharp EL-5102 $90 (1979) L 16 Dot M. LCD AOS Formula 80 N 11 Y     3*LR44 100-1,000h
6 Sharp EL-5102 $80 (1979) P 12 Dot M. LCD AOS Formula 48 N 6 Y     3*LR44 100-1,000h
7 HP HP-33C $120 (1978) P 12 (8+2) 7-Seg LED RPN Keystroke 49 Y 8 Y     2*AA NiCd <10h
7 HP HP-34C $150 (1978) P 12 (8+2) 7-Seg LED RPN Keystroke 210-70 Y 1-21 Y     2*AA NiCd <10h
7 HP HP-38C $150 (1978) P 12 (8+2) 7-Seg LED RPN Keystroke 99-8 Y 7-20 Y     2*AA NiCd <10h
7 HP HP-41C $295 (1978) P 12 (8+2) 14-Seg LCD RPN Keystroke 441-0 N 0-63 Y External 4 2*AA NiCd 100-1,000h
10 Casio fx-501P   P 13 (10+2) 7-Seg LCD AOS Keystroke 128 Y 11 Y   External 2*LR44 100-1,000h
10 Casio fx-502P   P 13 (10+2) 7-Seg LCD AOS Keystroke 256 Y 22 Y External   2*LR44 100-1,000h
1980 4 Sharp PC-1210 $159 (1980) L 24 Dot M. LCD AOS BASIC 0.4k Bytes 26 Y External   3*MR44 100-1,000h
6 Sharp PC-1211 $229 (1980) L 24 Dot M. LCD AOS BASIC 1.4k Bytes 26 Y External   3*MR44 100-1,000h
9 Panasonic RL-H1000 $250 (1981) L 26 Dot M. LCD AOS BASIC
FORTH
1.0k Bytes RAM Y External 3 5*AA NiCd 100-1,000h
9 Panasonic RL-H1400 $500 (1981) L 26 Dot M. LCD AOS BASIC
FORTH
3.0k Bytes RAM Y External 3 5*AA NiCd 100-1,000h
9 Panasonic RL-H1800 $750 (1981) L 26 Dot M. LCD AOS BASIC
FORTH
7.0k Bytes RAM Y External 3 5*AA NiCd 100-1,000h
12 HP HP-41CV $325 (1980) P 12 (8+2) 14-Seg LCD RPN Keystroke 2233-0 N 0-319 Y External 4 2*AA NiCd 100-1,000h
1981 3 Casio fx-601P $99 (1981) P 14 (11+3) Dot M. LCD
7-Seg LCD
AOS Keystroke 128 Y 11 Y External   2*CR2032 100-1,000h
3 Casio fx-602P $149 (1981) P 14 (11+3) Dot M. LCD
7-Seg LCD
AOS Keystroke 512-32 Y 22-88 Y External   2*CR2032 100-1,000h

Note: Table excludes the lowest end of “programmable” calculators featuring only simple key-reply operations and sporting only one memory.

• Market Research

Surprised by the success of Sharp's, Panasonics, and Casio’s early Pocket/Hand-held Computers using BASIC as programming language (and later confirmed by Hewlett Packard introducing the HP-75C Pocket Computer in September 1982), Texas Instruments’ Marketing Team conducted in June a large market study pitting three fictional products against each other. The market study was funded by Texas Instruments Corporate to serve as an example and model of in-depth analysis of the market potential for new TI product concepts, and to provide a sound basis for market planning.

The major objectives of the study presented in July 1981 to TI’s management included:

• Evaluate three proposed new Texas Instruments products:
   = RM 1000
   = RM 2000
   = RM 3000
• As concepts with no price stated, and at three different price levels for each product, to determine:
   - Buying interest – size of market potential
   = Function attributes and feature configuration desired by prospective markets
• Among four target markets:
   - Technical/Scientific
   - Students/Professors of Business and Engineering
   - Business/Financial/Professional
   = Calculator Owners

The team interviewed 857 carefully selected people in ten markets, most of them by appointment. Each participant was shown two of the three concepts mixed, matched and controlled to insure both monadic and comparative exposures.

Texas Instruments developed three different concept boards to illustrate and describe the product concepts. All three boards were designed to accomplish the same objectives for each product concept:

• Introduce the product
• Show the real-life size of the product. Give a good impression of its shape, thickness and styling
• Show the appearance, color, and capacity of the display. Emphasize clarify and ease of reading
• Show the functions and configuration of the keyboard
• Describe what the product is and what it can do. Answer the question "How do I / would I use it"
• Tell what the benefits are of owning and using the product. Tell people how easy it is to use
• Describe the basic characteristics of each product - Size, Display, Keyboard, Memory, Language, Power Source, and Portability.
   Refer to options available and their potential.

The three proposed products are looking to us today very familiar:

• RM 1000: Keystroke Programmable Pocket Calculator with 16-digit display, two Plug-in memory modules - TI Programmable 88
• RM 2000: BASIC Programmable Pocket/Hand-held Computer with 32-digit display, Plug-in memory module - TI-74 BASICALC
• RM 3000: Portable Computer with two ports for definition modules like BASIC programming, Word Processor, Terminal Software
   and tiltable 6-line by 40-column display and the size of a briefcase - Compact Computer70

The market study suggested that - as of Summer 1981 – the two product ideas RM 1000 (TI Programmable 88) and RM 3000 (Compact Computer 70) should be prioritized over the RM 2000 (TI-74 BASICALC).

It is understood that the study favored the continuation of the two (with respect to engineering resources) competing programs:

• Product X: Sophisticated keystroke Programmable Calculator: TI-88 plus peripherals PC-800, CA-800
• ALC (Advanced Language Computer): Programmable Language Portable Computer: CC-40 and CC-70 plus peripherals

• Enter TI-66, TI-76, and TI-86

One of the outputs of the market study conducted in June 1981 was:

The Technical/Scientific community is sensitive to price and we have seen that buying interest in the RM 1000 can be highly leveraged by price. We believe the RM 1000 should be marketed at $125, or even lower, and positioned as the latest breakthrough in Programmable Calculators at a very affordable price.

Well, the TI-86, known as Product X, was designed for a MSRP of $250 and unlikely to be profitable at $125 or even lower. With the Casio FX-702P already on the horizon, Texas Instruments conducted on June 29, 1981 a simulation of the expected market share of the TI-86 and other market contenders including a "TI-66" and "TI-76". The simulations covered three different scenarios for the TI-86, Casio FX-702P and Sharp PC-1211, best case was obviously with the TI-86 at lowest price, Casio FX-702P without printer availability, and PC-1211 at highest price:

Product Price Market
Share
Sharp PC-1211 $200  9.0% 
Texas Instruments TI-58C $80  8.2% 
Texas Instruments TI-59 $200  8.0% 
Texas Instruments TI-86 $200  14.3% 
Texas Instruments TI-66 $90  6.4% 
Texas Instruments TI-76 $150  8.1% 
Hewlett Packard HP-67 $310  4.7% 
Hewlett Packard HP-97 $595  2.8% 
Hewlett Packard HP-41C $200  12.1% 
Hewlett Packard HP-41CV $325  11.3% 
Casio FX-602P $145  8.1% 
Casio FX-702P $200  7.2% 

The "TI-86 vs Casio FX-702P Morton Market Simulation Model" discussion concluded with an important statement:

We must introduce the TI-86 on time to prevent a reversal of this situation.

The phrasing "on time" would suggest - with the Summer Consumer Electronic Show CES in June 1981 already passed - an announcement of the TI-86 in January 1982, ahead of the Winter CES.

In retrospect a very optimistic market share for the TI-86 at an unexpected low price point. And yes, the TI-66 Programmable was introduced in 1983 with a MSRP of $69.95 and well received while the TI-76 Programmable (or TI Programmable 76?) was just some vaporware making a product portfolio look professional on the performance-over-price discussion.

• Troubling Waters

Texas Instruments announced the TI Programmable 88 with CRAM/CROM modules and its peripherals on May 26, 1982 ahead of the Summer Consumer Electronic Show CES, knowing that the polysilicon gate CMOS process used with the TP0485 Microcontroller was finally in good shape but the design of the chip still needed some fixes requiring one or two additional manufacturing runs. Consequently the small printed read: "Availability expected late 4th Quarter".

As of May 1982 the market of Programmable Pocket Calculators and Pocket/Hand-held Computers included even more contenders than one year earlier:

Introduction Product MSRP Form Display Operation,
Programming
Program Size
(Merged?)
Data Size
(Backup?)
Archive Ports Battery Operating
Time
Year Mth Company Model Size Techn. Logic PGM Mode Steps M Memories B
1981 9 HP HP-11C $135 (1981) L 10 (7+2) 7-Seg LCD RPN Keystroke 203-63 Y 0-20 Y     3*LR44 >1,000h
9 HP HP-12C $150 (1981) L 10 (7+2) 7-Seg LCD RPN Keystroke 99-8 Y 7-20 Y     3*LR44 >1,000h
11 Casio FX-702P $249 (1981) L 20 (12+4) Dot M. LCD
7-Seg LCD
AOS BASIC 1680-80 Y 26-226 Y External   2*CR2032 100-1,000h
1982 1 Sharp PC-1500 $279 (1982) L 26 Dot M. LCD AOS BASIC 1.8k Bytes 52 Y External 1 4*AA 10-100h
5 TI TI-57 LCD $50 (1982) P 11 (8+2) 7-Seg LCD AOS Keystroke 48-0 Y 1-8 Y     2*LR44 >1,000h
5 TI TI-88 $350 (1982) P 16 (8+2) Dot M. LCD EOS Keystroke 960-0 Y 0-120 Y External 2 AA NiCd 100-1,000h

Note: Table excludes the lowest end of “programmable” calculators featuring only simple key-reply operations and sporting only one memory.

The stronghold of the TI-59, its ability to be customized with its Solid State Software Modules for target applications like insurance fee, tax, mortgage calculations vanished completely with the introduction of BASIC programmable Pocket/Hand-held Calculators like the Sharp PC-1211, Sharp PC-1500, Casio FX-702P, and Panasonic RL-H1400 offering easier program development, avoiding costly Mask-programmed ROM modules and offering much more advanced user interfaces and peripherals. Secondly started Hewlett Packard’s innovative HP-41C already in 1979/1980 to capture most of TI’s target markets and to make things even worse, plateaued the market for keystroke Programmable Calculators in 1981 at 600k units and started to decline in 1982 with a projected volume of about 500k units.

A market survey conducted by Texas Instruments in Summer 1982 with ALC (Advanced Language Calculator) target customers identified that three Pocket/Hand-held Computers with BASIC programming language (TI ALC, Sharp PC-1500, and Panasonic RL-H1400) were overwhelmingly preferred over keystroke programmable HP-41CV and TI-88 devices. Seventy-one percent of the respondents interviewed preferred the BASIC Language computing device. The major points of the preference were:

• The language is perceived to be easier to use than keystrokes
• The keyboard layouts provide easier entry of alphanumeric information

• Cancelation of the TI-88

Hewlett Packard introduced on September 15, 1982 with the HP-75C a powerful Pocket Computer with BASIC Programmability, a 32-digit alphanumeric display, 16k Bytes of non-volatile RAM for both programs and data, 4 ports for expansion cartridges, and even a build-in reader for magnetic cards. HP was forced to introduce this product at a significantly higher price point ($995) to protect the HP-41CV’s entrenched sales position. The HP-41CV and TI-88 are particularly vulnerable to a competitively priced Advanced Language Pocket/Hand-held Computer such as the ALC (think Compact Computer 40). This vulnerability would be amplified by the availability of Pascal as a language of that computer.

Consider it the final nail in the coffin of the TI-88!

Management at Texas Instruments discussed early in September 1982 the impact of the TI-88 cancellation in three different areas:

• TI’s reputation in the marketplace
• Specific obligations to customers
• Financial write-offs required by the cancellation

The following alternatives were considered:

• Continue the TI-88
• Cancel the TI-88 and replace it with the ALC keystroke program
• Cancel the TI-88 and not produce a keystroke program device and attempt to capture the market with BASIC language ALC
• Expand the ALC product family to include Pascal program language to capture the scientific keystroke user

The decision was to abandon the keystroke Programmable Pocket Calculator market and expand the ALC family to provide a Pascal program product to re-capture the keystroke user.

The Pascal version of the ALC included three significant new product features:

• The alteration of the case work of the ALC to accept Plug-in ROM cartridges with different language processors, enabling
   multiple language options on the same programmable portable computer
• The coding of the Pascal compiler
• The cost reduction of the current ALC design with the objective of the product cost of the new machine with the Plug-in ROM
   feature being no greater than the current ALC product costs

On September 10, 1982 Herb Shanzer, then Manager of TI's Calculator & Compact Computers Division, sent a short note to Doug Dobbs:

We met with Bill Sick today. The strategy that was recommended was to cancel the TI-88. Not to the ALC (Advanced Language Computer – Compact Computer 40) but to launch a Marketing Program which is based on the fact that we believe keystroke calculators are obsolete and to begin work on Fortran and Pascal versions of the current ALC.

We have to have a White Paper to support this recommendation for Bill Sick by Wednesday morning in time for the current forecast cycle. What has to be done is:

1. Development estimates for the two extra language versions of the ALC.
2. Specify risk assessment around the schedule of the current ALC as well as those new programs.
3. Refinement of the write-off estimates around the TI-88.
4. An in-depth analysis of the market impact of this strategy.
5. A specific statement in regards to the applications strategy. All of these machines, given
   they have one-line displays, are viewed as application machines, not development machines.
   And there is significant concerns how marketable they will be without applications.
6. Please make sure that whatever numbers you put together, you work with Milt Kuser.
   And whatever negotiations you are in the midst of in terms of materials, tooling or any
   aspects, Milt helps you with.

This is a 5-minutes dump right after the meeting. I will be spending six hours with Bill Turner on the way to Santa Fe and you will probably receive further instructions.

You ought to get in touch with Milt and Glenn who were both in this discussion to understand where it ended up.

Herb Shanzer


• Lessons Learned

A memorandum from Herb Shanzer distributed on September 20, 1982 provided interesting insight into Project Management in Texas Instruments’ Consumer Product Group:

The TI-88 development was an exceedingly ambitious program which attempted to push forward the state-of-the-art simultaneously in three dimensions: semiconductor process, architecture, and complex chip design. In retrospect, it was unrealistic to assume that the significant risks inherent in each of these technology areas could be isolated and managed separately without confusing interactions. As a result, it became impossible to isolate the problems in each of the areas and problem solution became an iterative process. There is ample precedent for similar problems occurring on other equivalently ambitious digital systems programs.

If we were to do this again, we would divide the program into three specific phases. First, a top-down system design with a clear definition of the interface requirements and performance requirements for both the process and each of the architectural entities. Secondly, each of those high-risk technology efforts would be broken apart and managed separately. That is, the process would be independently developed, while the microprocessor and support chips are designed, on paper, and checked using software simulation. Thirdly, only after these technology achievements had reached a satisfactory level of completion, would they be re-combined, to complete the TI-88. The proven architectural design could then be implemented as an ambitious custom design on the proven process.

The product was announced in late May at CES as a calculated risk driven by the desire to maintain TI’s image and position in the keystroke programmable market. An attempt was made to minimize the risk by holding the ship date until Fourth Quarter 1922. The time-table required for successful announcement at CES required that extensive preparation be done prior to the receipt of the latest version functioning calculators. A working prototype calculator was available only shortly before the CES and burn-in data was not available and understood at the time of the CES. It was only during burn-in that a number of subtle problems were discovered, these caused the re-design cycle prior to the current set of problems.

• Quanto Costa?

Continuing the TI-88 program would have included fixing the logic problems in the TP0485 chip set, producing new samples, and then further analyzing additional problem symptoms to ascertain whether or not the logic fix made the product sufficiently reliable to ship. Electing to ship the next mask version of the product, first shipment would occur January of 1983, already missing the promised availability of the TI Programmable 88 by Fourth Quarter 1982. If, as a calculated assumption, it would take two additional mask changes to manufacture a shippable product, shipment date would be in June of 1983 and additional spending was calculated to be $1.75M or about $5M in today’s (2020).

Texas Instruments calculated the total investment in the TI-88, CA-800 and PC-800 peripherals, and the TI-88 portion of the LCD III process, exclusive of the write-offs (about $3M), as follows:

Manpower (MY) 1977 1978 1979 1980 1981 1982 Total
Consumer Product Group (CPG) 0.5  6.0  14.0  30.0  30.0  20.0  100.5 
Semiconductor Group (SC) 1.0  2.0  25.0  20.0  9.0  57.0 
Total (MY) 0.5  7.0  16.0  55.0  50.0  29.0  157.5 
OST & Tooling $K 50  533  1,157  3,413  3,661  2,326  11,140 

Assuming the cost of Manpower in 1979 to 1982 averaging around $50K per MY with both the Consumer Product Group (CPG) and Semiconductor Group (SC), we could add up to a total damage of around $19M, or in today’s money between 50M and $60M. Always remember this number when you proudly hold one of the few surviving TI-88 calculators and CA-800 / PC-800 peripherals in your hands.

• Impact on the ALC Program

As of September 1982, Texas Instruments was finalizing plans for a family of Programmable Language Portable Computers based upon extensions of the ALC-A (Advanced Language Computer, Product A) and a library of application programs and language compilers to be used with this family. The planned product portfolio included three products with various options:

• ALC-LC – SRP $119 (1k RAM, 18k ROM, BASIC)

The ALC-LC was supposed to have a 16-character LCD display, reduced functionality BASIC language, no Plug-in modules but full ALC I/O bus (peripherals) compatibility.

The ALC-LC never materialized, its spiritual successor could be found with the TI-74 BASICALC introduced in 1986. Nevertheless were the TI-74 BASICALC and its sibling TI-95 PROCALC, a keystroke Programmable Calculator, based on the hardware and software of the Compact Computer 40 Plus.

• ALC-A2 – SRP $250 (2k RAM, 32k BASIC Plug-in ROM)
   SRP $270 (2k RAM, 64k Pascal Plug-in ROM)
   SRP $270 (2k RAM, 64k FORTRAN Plug-in ROM)

Principally a multi-language version of the ALC-A with a ¼" thicker housing, the addition of a Plug-in (flat pack) programming language module, expansion of the language ROM space from 32k Bytes to 64k Bytes, and expansion of internal RAM capacity from 18k Bytes to 34k Bytes.

The ALC-A2 was realized with the Compact Computer 40, announced in January 1983 and available for the end user around June 1983 with multiple peripherals and application programs announced and/or made available.

• ALC-C/T – SRP $550/$650 (8k RAM, 32k BASIC Plug-in ROM

An expansion of the ALC-A2 to enhance the product for use in higher utility applications, specifically high level professional and managerial support. The ALC-C will be an ALC-A2 in a larger case with a 6-line by 40-character LCD display, full travel typewriter keyboard and built-in tape mass storage.

The ALC-T (Terminal) is a variation of the ALC-C and provides the added feature of a built-in 300 Baud modem and telecommunication software.

Texas Instruments was working on the Compact Computer 70 but as of today we have no evidence of a functional prototype of the device. Rumors are, that the first batch of Gate Arrays designed for the CC-70 were faulty.

• Sad Ending

Texas Instruments dropped out of the home computer market in March 1984 - after selling more than 2.5 million of the famous TI-99/4A - and production of the CC-40 was ceased immediately after. The CC-40 Plus never made it to the market and only a few prototypes survived.

• Legacy

While with the TP0530, TP0531 and TP0532 three of the Integrated Circuits used with the "Product X" aka TI Programmable 88 were designed for TI’s metal gate CMOS process, decided the team to go for the ambitious LCD III microcontroller, known as TP0485, a different path and decided for a polysilicon gate CMOS process.

Although the LCD III was with the cancellation of the TI-88 a complete write-off (its technology was considered to be not compatible with solar powered calculators – enter "Project Solar", survived its polysilicon gate CMOS process and had a great comeback with the TMC70C20 8-bit Single Chip Microcontroller family used with the Compact Computer 40, HX-1100 Video Interface, HX-2000 Wafertape Digital Tape Drive, HX-3000 RS-232 Interface, HX-3100 Data Modem, TI-74 BASICALC, TI-95 PROCALC, PC-324 Printer, Financial Investment Analyst FIA-10 calculator, Fixed Income Securities (FIS) calculator, and last but not least the Franken-Calculator Data Dimensioner.

• Thank you

Brian Green for locating and securing the "Project X" Binder from the estate of CB Wilson
Jon Guidry for scanning the "Project X" Binder and making it available
Sean Riddle for decapping dozens of TI Calculator Chips for our research
Ken Shirriff for reverse engineering the technology of TI Calculator Chips
Mike Sebastian for compiling a reference with TI Calculator Patents
Viktor T. Toth for operating a site dedicated to Programmable Calculators and Pocket Computers
Rick Furr for maintaining an accurate database about HP calculators
All the unsung heroes working between 1977 and 1982 on "Project X" and sharing their valuable insight with the Datamath Calculator Museum 
Many friends for sharing their Programmable Calculators and Pocket Computers with the Datamath Calculator Museum for our research
And last but not least an amazing community of Calculator and Pocket Computer Aficionados sharing their knowledge and...
.. and not to forget Google - the greatest Search Engine on the planet!


horizontal rule

If you have additions to the above article please email: joerg@datamath.org.

© Joerg Woerner, January 28, 2021. No reprints without written permission.