DATAMATH CALCULATOR MUSEUM |
Lloyd's Accumatic 30 (Model EH-9036, Type 255)
Date of introduction: | July 1974 | Display technology: | Fluorescent |
New price: | Display size: | 8 | |
Size: | 5.8" x 3.5" x 1.25" 147 x 89 x 32 mm3 |
||
Weight: | 6.0 ounces, 169 grams | Serial No: | 5-29677 |
Batteries: | 4*AA | Date of manufacture: | mth 08 year 1974 |
AC-Adapter: | YA-7247 (120V AC), YA-7585 240V AC) | Origin of manufacture: | Taiwan |
Precision: | 8 | Integrated circuits: | Sharp LI2002 |
Logic: | Chain | Displays: | Futaba 8-CT-01A |
Memories: | |||
Program steps: | Courtesy of: | Joerg Woerner | |
Download manual: | (US/FR: 2.3M Bytes) |
This Accumatic 30 calculator sold by Lloyd's Electronics, Inc., Compton, California caught our attention when we were looking into the first single-chip calculator circuits manufactured by Sharp Electronics Corporation in Japan. To our surprise featured the first calculator we dismantled here at the Datamath Calculator Museum a Hitachi HD36290 brain and the second sample even a completely different Hitachi HD3633 chip.
Thirty-one calculators later, we were able to compile an overview of 12 different Accumatic 30 calculators manufactured between 1974 and 1976:
Product | Model | Type | Serial # Range |
Label Remarks | Manufacturing Site on Label |
Calculator Chip |
Display # of Digits |
Keyboard # of Keys |
Accumatic 30 | EH-9036 | 255 | 5-2xxxx 5-3xxxx |
EU-0010-0011 | Taiwan | Sharp LI2002 |
8 + LED | 19 |
Accumatic 30 | EH-9036 | 255B | 5-4xxxx | EU-0010-0011 | Japan | Sharp LI2002 |
8 + LED | 19 |
Accumatic 30 | EH-9036 | 255B | I2-0xxxx | EU-0010-0011 | Taiwan Republic of China |
Sharp LI2002 |
8 + LED | 19 |
Accumatic 30 | EH-9036 | 255C | 2-0xxxx 2-6xxxx |
EU-0010-0011 | Taiwan Republic of China |
Hitachi HD3633 |
8 + LED | 19 |
Accumatic 30 | EH-9036 | 255D | I5-1xxxx I5-3xxxx |
EU-0010-0011 | Japan | Hitachi HD36290 |
9 | 19 |
Accumatic 30 | EH-9036 | 255D | 5D-1xxxxx | Japan | Hitachi HD36290 |
9 | 19 | |
Accumatic 30 | EH-9036-0018 | 255D | 5E-1xxxxx | Japan | Hitachi HD36290 |
9 | 19 | |
Accumatic 30 | EH-9036-0018 | 255G | 5G-1xxxxx | Japan | NEC µPD940 |
9 | 19 | |
Accumatic 30 | EH-9036 | 260D | 5D-5xxxx 5E-5xxxx 5E-6xxxx |
Taiwan Republic of China |
Hitachi HD36290 |
9 | 19 | |
Accumatic 30 | EH-9036 | 260D | 5G-1xxxxx 5G-2xxxxx |
Taiwan | Hitachi HD36290 |
9 | 19 | |
Accumatic 30 | EH-9036 | 260D | 5J-1xxxxx | Taiwan | NEC µPD940 |
9 | 19 | |
Accumatic 30 | EH-9036-2 | 255E | I6-1xxxx I6-2xxxx |
Taiwan | NEC µPD940 |
9 | 22 |
It was during the peak of the "Calculator Wars" not unusual to switch various components during the life-cycle of a certain calculator, but what we discovered with Llody's Accumatic 30 was unheard-of. Switching the calculator brain, display or keyboard is not only a major redesign of the products, even sales brochures and user manuals are affected. So what was the reason to go through so many iterations with a single product? In most cases we identified one of three motivations:
• Reducing manufacturing costs to keep a profit margin while sales prices are dropping • Keeping production lines running during times of supply chain constraints • Adding functionality to stay competitive |
With this Lloyd's Accumatic 30 series, we suspect that the original Sharp LI2002 and the later Hitachi HD3633 chips had supply chain constraints and both the Hitachi HD36290 and NEC µPD940C were picked as second-sourceable replacements. The final design, dubbed EH-9036-2, was obviously driven by the desire to offer more functionality without increasing manufacturing costs.
Dismantling the featured Lloyd's Accumatic 30 (EH-9036, Type
255) calculator assembled by an unknown Original
Equipment Manufacturer (OEM) in August 1974 in Taiwan reveals a very compact
design based on a single-sided printed circuit board (PCB) for the main
electronics, a single-sided PCB for the keyboard and powered by four disposable
1.5 Volts batteries or an external power adapter.
The
Main-PCB is centered around a LI2002 single-chip calculator circuit manufactured
by Sharp and the few other remaining components on the PCB are mainly used to
generate the different supply voltages for the LI2002 and Vacuum Fluorescent
Display (VFD) and to bias the anodes and grids of the display with respect to
its filament.
To gain some knowledge about the differences
between the various single-chip calculator circuits used with the Lloyd's
Accumatic 30, we decided here at the Datamath Calculator Museum to give the featured calculator
a full "Teardown Treatment" and share our findings accordingly.
Calculating Unit:
The LI2002 located in the featured calculator seems to be one of the first
single-chip calculator circuits designed by Sharp. It features an integrated
clock oscillator and both its segment and digit output drivers are interfacing
directly with low-voltage VFDs up to 35 Volts. The two keyboard scanning inputs
can't withstand these higher voltages and need to be "isolated" with additional
diodes.
Display: The
featured Lloyd's Accumatic 30 calculator manufactured
in August 1974 makes use of an 8-Digit low-voltage VFD manufactured by Futaba and
known as Type 8-CT-01A. The display is soldered with its 18 wires directly to the
Main-PCB. A small red LED with long leads is soldered directly onto the Main-PCB
of the Accumatic 30 calculator to act as Minus "dot" for 8-digit numbers.
Display Driver: The term "low-voltage" Vacuum Fluorescent Display might
be misleading when used together with a calculator powered by four 1.5 Volt
batteries. Common VFDs used with portable electronic calculators are usually
operated around 30 Volts, significantly higher than the 10 to 15 Volts operating
voltage of single-chip calculator circuits used in the 1970s. While the first
generation of Texas Instruments TMS0100 single-chip calculator circuits lacked
any display drivers and left the choice of display technology to their
customers, focused the second generation products mainly on Light-Emitting Diode
(LED) technology. In or around 1974, most Western calculator designs still
relied on rather expensive LED technology but Japanese companies like Casio,
Sanyo, Sharp and Toshiba started to leverage the lower manufacturing costs of
VFDs, instead. Texas Instruments introduced in 1974 consequently with the
TMS0850 their first product series focused on battery operated VFD calculators
and modified the integrated segment and digit output drivers to withstand up to
-35 Volts. Sharp on the other hand entered the marked of single-chip calculator
circuits in 1974 and focused immediately on compatibility with VFDs. The LI2002 chips are manufactured in PMOS technology, meaning the
output transistors are "high-side" switching and the most positive voltage of
the chip is labeled VSS for 0 Volt, all other voltages in the
calculator are consequently negative with respect to VSS. Multiplexed
low-voltage VFDs need a voltage difference between its filament and the grids
and anodes of the numbers of around 30 Volts to light up and to avoid "ghosting"
while scanning, the deactivated grids and anodes should be slightly lower than
the filament voltage. An elegant and very common solution is found with this
Accumatic 30 calculator, too. The grids and anodes of the VFD are "pulled-down"
with 16 resistors (150k Ohm) to around -32 Volts, the filament is biased to
around -28 Volts (2.0k Ohm resistor) and the LI2002 switches the
relevant grids and anodes to around 0 Volt to lit them up.
Clock: The Lloyd's Accumatic 30 makes use of the internal clock
oscillator of the LI2002 chip, we identified a resistor with 120k Ohm connected
between Pin 4 (REXT) of the LI2002 and the negative
VGG power supply line, resulting in a clock frequency of about 71 kHz.
Power Supply: The Lloyd's Accumatic 30 calculator is powered with
four disposable AA-sized 1.5 Volt batteries or an external 6 Volt power adapter and uses a
complex DC/DC converter to
generate a total of four voltages:
• VDD - Negative supply for
LI2002 (-8.1 V) • VGG - Negative supply for LI2002 (-17.2 V) • VPP - Negative supply for VFD anodes and grids (-32.0 V) • VFIL - AC supply for VFD Filament (2.5 V) |
We measured the operating current of the featured Lloyd's Accumatic 30 calculator for three different cases:
Mode | Display | Current VBAT = 6.0 V |
Clock Frequency |
Calculating | 0. | 39 mA | 71 kHz |
Calculating | 88888888. | 46 mA | 71 kHz |
Calculating | 0. ⚫ | 46 mA | 71 kHz |
Calculating the power consumption at 6 Volts for the Lloyd's Accumatic 30 results in about 230 mW displaying a '0.', about 280 mW with all segments but the minus sign illuminated and around 280 mW displaying a "negative zero". A very interesting result, a Canon LE-84 calculator with a LED display and using four disposable 1.5 Volt Alkaline batteries and a DC/DC converter for its TMS0801 chip clocks in at around 100 mW displaying a '0.' and 320 mW with all segments lit; showing both an advantage and disadvantage of LED-based calculators versus their VFD-based counterparts:
• LED: Only illuminated segments draw current - advantage LED while displaying
'0.' • VFD: Filament uses always current, segment currents are almost negligible - advantage VFD while displaying '88888888.' |
And yes, the little red LED used as Minus "dot" draws about 7 mA, as much as 50 segments of the VFD.
Keyboard: The keyboard assembly of the
Lloyd's Accumatic 30 was manufactured by GICO and uses 19 spring-supported
plastic keys pushing small fingers on three stamped sheet-metal plates against
contacts etched on a single-sided phenolic PCB.
The three plates show nicely the
layout of the LI2002 keyboard scan-matrix, the numbers [1] to [9] are combined
with the first plate, all other keys but the [C] key are on the second plate,
while the [C] key is not included in the scan-matrix but uses an individual
contact instead. The keyboard module is connected with 13
pins (9 keyboard scan lines, 2 keyboard return lines, 2 contacts for the [C]
key) to
the Main-PCB.
Here
at the Datamath Calculator Museum we use
the DCM-50A Platform to
Characterize and
Reverse-engineer
Single-chip Calculator Circuits. Many designs of electronic calculators do not
use all features of their calculator brains and it would be difficult to unleash
the full potential of the calculator chips in these cases. Additionally are
electronic calculators "closed systems" with limited flexibility to measure
signals, change voltages or clock frequencies, provide additional input keys or
even change the display technology or specifications additional digits. Core
idea of the DCM-50A is providing a generic platform to access all features of a
single-chip calculator circuit and with the
DCM-50A (PLAYGROUND) we
increased the scope from Texas Instruments products to offerings from their
competitors in the 1970s, namely AMI, Cal-Tex, Commodore/MOS Technology,
Electronic Arrays, General Instrument, Hitachi, Litronix, Matsushita, Mitsubishi, Mostek,
National Semiconductor, NEC, Omron, RFT, Rockwell, Sharp, Toshiba, and Western
Digital.
One Calculator -
Four Brains:
The Accumatic 30 sold by Lloyd's mid of the Seventies went through many
redesigns within its lifecycle of around 2 years and we started to look into the
differences of the four different single-chip calculator circuits located in at
least 12
different design iterations. When the first Accumatic 30 (Model EH-9036, Type 255)
was introduced in July 1974, it was based on a rather unusual Sharp
LI2002 Chip, probably Sharp's first
"single-chip" design and competing with Texas Instruments' successful
TMS0800 Product Family introduced in Fall
1973. TI was looking back on two years experience with single-chip calculator
designs, their TMS1802 was introduced
already in September 1971 and consequently was the TMS0800 a very polished
product. When we started to
Characterize the LI2002, we noticed some quirks and bugs and decided to look
closer into the four different calculator chips used with the Accumatic 30 and
share our observations.
Sharp LI2002:
The first Accumatic 30 (Model EH-9036, Type 255) and second Accumatic 30 (Model EH-9036, Type 255B) are using an 8-Digit low-voltage Vacuum Fluorescent Display (VFD) manufactured by Futaba and an additional small red LED as Minus "dot" for 8-digit numbers. While the LI2002 supports 9-Digit Displays, does it output only a '-' for negative numbers in the extra digit. Instead of indicating error conditions like a calculator overflow or division by zero with 'C' or 'E', does it display instead an "8-Digit friendly" zero with all decimal points lit '. . . . . . .0. .'.
Observations about the Calculator Logic implemented with the Sharp LI2002:
Automatic Constant: Implemented for multiplication (1st number used as constant)
and division (2nd) with unusual behavior during division
[3] [x] [2] [=] [=] → '18.', [1] [=] → '3.'; [4]
[x] [=] → '16.'; [4] [x] [=] [=] → '64.'
[3] [:] [2] [=] [=] → '0.75', [1] [=] → '0.5.';
[4] [:] [=] → '0.25; [4] [:] [=] [=] → '0.0625'
Reciprocal Function: The reciprocal function is implemented using the [:]
key directly followed by the [=] key
[4] [:] [=] → '0.25'
Percent Function: The [%] key followed by the [x] key allows with the [+]
and [−] keys mark-up and discount calculations
[2] [0] [x] [5] [%] → '1.', [=] → '20.'
[2] [0] [x] [5] [%] → '1.', [+] [=] → '21.'
[2] [0] [x] [5] [%] → '1.', [−] [=] → '19.'
Known Calculator Logic Bugs:
Negative Zero Bug: Certain calculations result in displaying a negative zero
[1] [−] [2] [+] [1] [=] → '0.-';
[−] [0] [.] [0] [0] [0] [1] [:] [1] [0] [0] [=] [=] → '0.-'
Hitachi HD3633:
The third Accumatic 30 (Model EH-9036, Type 255C) is again using an 8-Digit low-voltage Vacuum Fluorescent Display (VFD) manufactured by Futaba and an additional small red LED as Minus "dot" for 8-digit numbers. Like the Sharp LI2002 does the Hitachi HD3633 Chip support 9-Digit Displays, but it outputs for error conditions like a calculator overflow or division by zero a 'C' or 'E' in the missing extra digit. This behavior is for the calculator user very confusing, the only obvious indicator of the error condition is the keyboard not responding to any other button then the [C] key.
Observations about the Calculator Logic implemented with the Hitachi HD3633:
Automatic Constant: Implemented for multiplication (1st number used as constant)
and division (2nd)
[3] [x] [2] [=] [=] → '18.', [1] [=] → '3.'; [4]
[x] [=] → '16.'; [4] [x] [=] [=] → '64.'
[3] [:] [2] [=] [=] → '0.75', [1] [=] → '0.5.';
[4] [:] [=] → '1.'; [4] [:] [=] [=] → '0.25'
[3] [+] [2] [=] [=] → '5.', [1] [=] → '1.'; [4] [+] [=] [=] → '4.'
[3] [−] [2] [=] [=] → '1.', [1] [=] → '1.'; [4] [−] [=] [=] → '4.'
Percent Function: The [%] key followed by the [x] key allows with the [+]
and [−] keys mark-up and discount calculations
[2] [0] [x] [5] [%] → '1.', [=] → '20.'
[2] [0] [x] [5] [%] → '1.', [+] [=] → '21.'
[2] [0] [x] [5] [%] → '1.', [−] [=] → '19.'
Known Calculator Logic Bugs:
Divide to Negative Zero Bug: Certain calculations result in displaying a negative zero
[1] [−]
[2] [=] → '1.-', [:] [1] [0] [0] [0] [0] [=] → '0.0001-', [=] → '0.-'
Negative Zero Bug: Certain calculations result in displaying a negative zero
[1] [−] [2] [+] [1] [=] → '0.-'
Hitachi HD36290:
The fourth Accumatic 30 (Model EH-9036, Type 255D) dropped the 8-Digit low-voltage Vacuum Fluorescent Display (VFD) with the additional small red LED as Minus "dot" for a 9-Digit VFD manufactured by Futaba. Error conditions like a calculator overflow or division by zero are indicated with 'C' or 'E' in the leftmost position of the display.
Observations about the Calculator Logic implemented with the Hitachi HD36290:
Automatic Constant: Implemented for multiplication (1st number used as constant)
and division (2nd)
[3] [x] [2] [=] [=] → '18.', [1] [=] → '3.'; [4]
[x] [=] → '16.'; [4] [x] [=] [=] → '64.'
[3] [:] [2] [=] [=] → '0.75', [1] [=] → '0.5.';
[4] [:] [=] → '1.'; [4] [:] [=] [=] → '0.25'
[3] [+] [2] [=] [=] → '5.', [1] [=] → '1.'; [4] [+] [=] [=] → '4.'
[3] [−] [2] [=] [=] → '1.', [1] [=] → '1.'; [4] [−] [=] [=] → '4.'
Percent Function: The [%] key followed by the [x] key allows with the [+]
and [−] keys mark-up and discount calculations
[2] [0] [x] [5] [%] → '1.', [=] → '20.'
[2] [0] [x] [5] [%] → '1.', [+] [=] → '21.'
[2] [0] [x] [5] [%] → '1.', [−] [=] → '19.'
Known Calculator Logic Bugs:
Temporary Divide to Negative Zero Bug: Certain calculations result in displaying a negative zero
[1] [−]
[2] [=] → '-1.', [:] [1] [0] [0] [0] [0] [=] → '-0.0001', [=] → '-0.', [:] [1] [0] [=] → '0.'
Negative Zero Bug: Certain calculations result in displaying a negative zero
[1] [−] [2] [+] [1] [=] → '-0.'
NEC µPD940:
The fifth Accumatic 30 (Model EH-9036, Type 255G) is using like the forth Accumatic 30 (Model EH-9036, Type 255D) a 9-Digit low-voltage Vacuum Fluorescent Display (VFD) manufactured by Futaba. Error conditions like a calculator overflow or division by zero are indicated with 'C' or 'E' in the leftmost position of the display.
Observations about the Calculator Logic implemented with the NEC µPD940:
Automatic Constant: Implemented for multiplication (1st number used as constant), division (2nd), addition (2nd), and subtraction (2nd)
[3] [x] [2] [=] [=] → '18.', [1] [=] → '3.'; [4] [x] [=] [=] → '64.'
[3] [:] [2] [=] [=] → '0.75', [1] [=] → '0.5.'; [4] [:] [=] [=] → '0.25'
[3] [+] [2] [=] [=] → '7.', [1] [=] → '3.'; [4] [+] [=] [=] → '12.'
[3] [−] [2] [=] [=] → '-1.', [1] [=] → '-1.'; [4] [−] [=] [=] → '-4.'
Percent Function: The [%] key followed by the [x] key allows with the [+]
and [−] keys mark-up and discount calculations. Using the [=] key following the
[%] key in multiplication leads to unexpected results
[2] [0] [x] [5] [%] → '1.', [+] → '21.', [=] →
'22.'
[2] [0] [x] [5] [%] → '1.', [−] → '19.',
[=] → '18.'
[2] [0] [x] [5] [%] → '1.', [=] → '20.'
[5] [:] [2] [0] [%] → '25.', [=] → '1.25'
Known Calculator Logic Bugs (Not relevant for Accumatic 30):
Negative Square Root Bug: Negative square roots are allowed and result in a negative number
[8] [√x] → '2.8284271'; [−] [8] [√x] → '-2.8284271'
Comparison Test
Spending countless hours with bread-boarding the Four Brains of the Accumatic 30 calculator, characterizing their clock generators, measuring their power consumption, analyzing their keyboard and display scanning, testing the embedded Logic and analyzing their manufacturing processes and architecture, one big questions needs to be answered:
And which single-chip calculator circuit is the best one?
Well, ask the OEM manufacturer of the Accumatic 30 - their sequence in updating from one chip to the next one - matched almost our own experiences here at the Datamath Calculator Museum:
• 4th Place -
HD3633: Very buggy Logic, large
40-pin package and confusing error condition signaling with 8-digit
display • 3rd Place - LI2002: Buggy, unusual implementation of Constant Functions • 2nd Place - HD36290: Still very buggy Logic, large silicon chip • 1st Place - µPD940: Bug-free, perfect implementation of Constant Functions, lowest supply voltages, innovative keyboard scanning, very small silicon chip. |
The µPD940 is winning this Comparison Test hands down and it is no surprise to us, that NEC won mid of the 1970s a big chunk of the single-chip calculator circuit business and was mid of the 1980s the World's largest semiconductor company by annual revenue. Well, Texas Instrument invented the Integrated Circuit, spearheaded the development of LSI (Large Scale Integration) chips for electronic calculators and took in Fall 1971 with the introduction of the TMS0100 Product Family the market of handheld calculators by storm - how about their TMS0850?
Both the NEC µPD940 and TI's TMS0850 were introduced in the first half of 1974, addressed precisely the same market and were manufactured in a very similar p-channel Metal–Oxide Semiconductor (PMOS) process. So we are not racing a VW Beetle against a Porsche 911:
Type | Prozess | Die Size | Power Supply |
ROM Size |
RAM Size |
Constant (M-D-A-S) |
Digits | Fixed DP | Rounding | Special Functions |
Known Bugs |
(6,7,9) Font |
Entry Overflow |
Calculating Overflow |
µPD940 |
7.5 um PMOS |
3.8 * 3.8 mm2 150 * 150 mils2 |
VDD -6.0 V VDD -12.0 V |
4,864 Bits | 144 Bits SAM |
1-2-2-2 | 8 + Sign | Float | None | [%] [+/−] [√x] [PI] |
√−x allowed |
|||
TMS0855 |
8.0 um PMOS |
4.1 * 4.0 mm2 160 * 155 mils2 |
VDD -10.0 V VDD -15.8 V |
3,520 Bits | 154 Bits SAM |
1-2-2-2 | 8 + Sign | Float | None | [%±] [√x] Timeout, Low V. |
√−x allowed |
On par!
If you have additions to the above article please email: joerg@datamath.org.
© Joerg Woerner, October 1, 2024. No reprints without written permission.