Aalborg Universitet

The Turing-850 Project

Developing a Personal Computer in the Early 1980s in Mexico
Ortiz-Arroyo, Daniel; Rodríguez-Henríquez, Francisco; Coello-Coello, Carlos

Published in:
IEEE Annals of the History of Computing

DOI (link to publication from Publisher):

10.1109/MAHC.2010.68

Publication date:
2010

Document Version
Early version, also known as pre-print

Link to publication from Aalborg University

Citation for published version (APA):

Ortiz-Arroyo, D., Rodríguez-Henríquez, F., & Coello-Coello, C. (2010). The Turing-850 Project: Developing a
Personal Computer in the Early 1980s in Mexico. IEEE Annals of the History of Computing, 32(4), 60-71.
https://doi.org/10.1109/MAHC.2010.68

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

- Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
- You may not further distribute the material or use it for any profit-making activity or commercial gain
- You may freely distribute the URL identifying the publication in the public portal -
Take down policy
If you believe that this document breaches copyright please contact us at [email protected] providing details, and we will remove access to
the work immediately and investigate your claim.

Downloaded from vbn.aau.dk on: octubre 25, 2023

 [3B2-14] man2010040060.3d 8/11/010 10:56 Page 60

The Turing-850 Project: Developing

a Personal Computer in the Early
1980s in Mexico
Daniel Ortiz-Arroyo
Aalborg University, Denmark

Francisco Rodrı́guez-Henrı́quez and Carlos A. Coello Coello

Center for Research and Advanced Studies (CINVESTAV-IPN), Mexico

In response to the increasing popularity in the late 1970s of afford-

able, general-purpose, microprocessor-based personal computers in
the US, several countries attempted to create indigenous personal
computer industries. The Turing-850, a general-purpose PC designed
and built in Mexico in the early 1980s, shows that Mexico had the
technological capabilities to develop its own such industry, but diverse
factors prevented its creation.

On 8 June 1958, the National Autonomous In 1970, the Mexican government created
University of Mexico (UNAM) bought what the National Council for Science and Tech-
appears to be the first computer ever to oper- nology (CONACyT), the institution responsi-
ate in Mexico (or anywhere else in Latin ble for the policies and investment in science
America)—an IBM 650.1 To have a proper and technology. The initial research grants
home for the IBM 650, UNAM’s authorities CONACyT awarded were primarily used to
created the Electronics Computer Center help expand higher education in Mexico. As
(CCE), located in the heart of its main cam- a result, a small community of researchers
pus.2 Soon after its creation, the CCE began and scientists was created in the country.6
disseminating knowledge about applications By the late 1970s, active research and de-
for this novel computer technology, includ- velopment in computer systems was being
ing an annual workshop under the name conducted in a handful of Mexican univer-
‘‘Computers and Their Applications.’’ Re- sities such as UNAM, IPN, and the Benemér-
markably, the third conference in the series, ita Universidad Autónoma de Puebla
held in 1961, featured MIT professors John (BUAP).7 These efforts, however, were iso-
McCarthy, Marvin L. Minsky, and Harold V. lated, without an articulated national strat-
McIntosh as the keynote speakers.3 egy.6,8 This situation was symptomatic in a
In 1963 and 1964, an analog computer country like Mexico, where the government’s
called UNIKORNIO was built at UNAM, the investments in science and technology were
first of its type in Mexico. Shortly after as well below the era’s international standards.9
an experiment, Manny Lemman helped Furthermore, private funding for research
UNAM construct the digital computer and development was scarce, and most Mex-
MAYA, based on the design of the Sabre com- ican-owned companies were importing all
puter at the University of Israel.1 the technology they needed. Lastly, coopera-
A few years later, Mexican universities tion between industry and academia was
began offering educational programs in practically nonexistent.6,10,11
computer science and engineering. The first In an effort to create an indigenous infor-
Mexican bachelor’s program in computer mation technology industry, the Mexican
engineering was offered as far back as 1965 government decided to launch a greenhouse
by the National Polytechnic Institute strategy. To this end, on August 1981 the
(IPN)4,5 and was soon followed by a few Promotion Plan for the Electronics and Com-
other universities. puting Industry (PFIEC) was announced.12

60 IEEE Annals of the History of Computing Published by the IEEE Computer Society 1058-6180/10/$26.00  2010 IEEE
c
[3B2-14] man2010040060.3d 8/11/010 10:56 Page 61

The general goal of the PFIEC was to develop at BUAP, undertook the project to design
a national computer and electronics industry and build the Turing-850 personal computer
that could help the country reach a level of system. The aim was to create state-of-the-art
autonomy in information technologies.13 In- indigenous computer technology that could
formation technologies were regarded as be readily manufactured by the national in-
strategic, helping to reduce the tremendous dustry. The project attempted to create a co-
technological dependence from abroad. In operation channel with Mexican-owned
the specific case of computers, PFIEC companies by transferring the technology
imposed the restriction that every computer to interested parties.
sold in the country should have a high per- The Turing-850 became, to the best of our
centage of components made in Mexico. knowledge, the first general-purpose, per-
Later on, this policy was changed to allow sonal computer system entirely designed in
companies to import all computer compo- Mexico. This article provides a detailed ac-
nents they wanted under the condition that count of the Turing-850 project and gives
these companies had at least 51 percent of some insights into the issues commonly con-
Mexican capital. Likewise, the PFIEC man- fronted by researchers carrying out techno-
dated multinational companies to invest be- logical projects in developing countries. As
tween 3 and 6 percent of their gross sales in part of the historical framework within
R&D, as well as to include a fraction of Mex- which the Turing-850 was conceived, we
ican manufactured components in their briefly describe other similar computers
systems.3,9,13–16 designed in Mexico and outside the US dur-
Nevertheless, soon after the PFIEC an- ing the same period.
nouncement, a major Mexican financial crisis
was unleashed, causing profound economic Project antecedents
consequences across the 1980s. From 1981 In 1980, National Cash Register (NCR) cre-
to 1989, in a high-inflation scenario, the ated an R&D department at its Puebla manu-
Mexican peso exchange rate against the US facturing plant.18 The goal was to further the
dollar went from an average of 24.5 pesos development of the NCR 2140, one of NCR’s
per dollar to an average of 2,461 pesos per most popular cash registers at that time.
dollar.17 The R&D department was headed by Luis
A second factor that also negatively af- Medina-Vaillard, a Mexican researcher who
fected the nationalistic policies promoted had graduated from the California Institute
by the PFIEC was the introduction of the of Technology. Nine Mexican engineers—all
IBM PC and its open architecture. This recent graduates from Mexican universities—
event ignited the standardization of micro- were hired by NCR to join its new R&D de-
computers around the IBM architecture, leav- partment. Their first project was to adapt the
ing little room for other design alternatives.15 NCR 2140’s hardware and software to control
As a consequence of these factors, the fuel supplies at gas-pumping stations. How-
Mexican government changed its policy ever, just before reaching the first year of oper-
and promoted liberalization instead in ation, the entire R&D department was shut
1985, when IBM asked for permission to pro- down because of NCR’s financial problems.
duce PCs in Mexico. After a lengthy negotia- Luis Rivera-Terrazas, one of the pioneers
tion, the permission was finally granted in of modern astronomy in Mexico19 and presi-
1987. The PFIEC was finally abandoned in dent of BUAP from 1975 to 1981, was a
1990. That year, the only Mexican market pro- committed scientist who understood the im-
tection that was still in place was a 20 percent portance of computers, both in science and
tax on hardware, which fell to 12 percent by education and as an effective means of
1994 and was definitely removed in 1998 in improving Mexico’s social development. He
the context of the North American Free was a supporter of the consolidation of Mex-
Trade Agreement (NAFTA).15 ican science and technology to achieve tech-
Another consequence of the Mexican fi- nological independence from abroad.20
nancial crisis in the 1980s was that the few When NCR closed its R&D operation,
scientists the country had been able to foster Rivera-Terrazas personally offered the entire
started to emigrate. As a result, only a few R&D team a position in the Department of
small groups within academia had enough Microcomputer Applications at the BUAP
funding and human resources to continue (DMA-BUAP).
developing technology.1 One of those aca- Shortly after, six members of the original
demic groups, an engineering team working NCR’s R&D team submitted the Turing-85

October–December 2010 61
 [3B2-14] man2010040060.3d 8/11/010 10:56 Page 62

The Turing-850 Project: Developing a Personal Computer in the Early 1980s in Mexico

project to Rivera-Terrazas, who committed by the end of 1985, with the hope that the
BUAP’s resources to finance the project’s ini- Mexican economy would have recovered
tial phase and promised to look for the extra by then.
funds necessary to support it thereafter. The The Mexican commercial and education
project’s main goals were to assess the feasi- sectors were the target markets envisioned
bility of creating state-of-the-art computer for the Turing-850. The Turing-850’s design
technology within a Mexican public univer- was conceived to compete in this market
sity and to design a computer system using with other similar computers of the era in
technology that could be easily transferred terms of its features, cost, and performance.
to interested Mexican-owned companies for The machine was also planned to fulfill
mass production. BUAP’s computing equipment needs, to-
Together with the project leader Medina- ward a sustainable self-production of
Vaillard, the Turing-85 project participants microcomputers.
were Daniel Ortiz-Arroyo (a coauthor on
this article), Gregorio Arenas-Muñoz, Carlos Project development
Blanco-Salinas, Sergio Guevara-Rubalcava, Phase 0 of the project started in the spring
and Francisco Serrano-Osorio. of 1982, with the goal of delivering a fully
functional 8-bit personal computer system
Turing-85 project phases by late 1983.
The Turing-85 project’s original goal was The restrictions imposed by the govern-
ambitious: design a 32-bit multiuser minicom- ment through PFIEC, together with the
puter system. The NS16000 microprocessor, high retail cost of the original IBM PC24
manufactured by National Semiconductors,21 (and other computers manufactured abroad),
was initially selected as the computer’s main had created good market opportunities for
processor. The project contemplated design- low-cost personal computers fabricated in
ing most components, from the power Mexico. However, to ensure the Turing-850
supply, motherboard, terminals, and I/O a share in the incipient Mexican market for
interfaces to the operating system and com- personal computers, it was critical to deliver
pilers able to support at least two computer it on time because the personal computer
languages. To carry out an enterprise of this manufacturing sector in the US was growing
magnitude, the original project’s budget rapidly. To better maintain control of the
included the hiring of numerous researchers project, the design team enacted and
and engineers from Mexico and abroad. The enforced policies to write, discuss, and get ap-
feasibility study and preliminary specifica- proval for the functional specifications of
tions of the computer were completed by each subsystem. Design inspections were car-
January 1982. (See related work for a ried out and periodic individual progress revi-
detailed specification.22) sion evaluations were performed for the
Unfortunately, the Mexican peso was sud- duration of the project. Additionally, a strict
denly devalued on 18 February 1982.17,23 project schedule was set up. However, despite
That event triggered a Mexican financial applying all these procedures, the project suf-
crisis that virtually canceled the original fered a series of delays caused by internal and
project. Nevertheless, amid the economic external factors.
turmoil and with a substantially reduced On the one hand, the local industry had
budget, a more modest proposal was no experience in supporting the develop-
promptly specified: design a low-cost, high- ment of technological projects. Some reluc-
performance 8-bit personal computer system. tant providers were convinced to participate
Furthermore, because of the limited human in the project only when it was promised
resources and budget available, all efforts they would supply the components to mass
were concentrated exclusively on designing manufacture the computer. Moreover, all
the hardware. In honor of Alan Turing, the imported parts had to be acquired through
new project was renamed Turing-85 phase the appropriate university channels that, on
0, or Turing-850. The project was then occasion, were slow in processing the pur-
planned to be delivered in two stages. In chase orders. The continued depreciation of
phase 0, an 8-bit machine would be built; the Mexican peso against the dollar during
the target in phase 1 would be the construc- those years also caused the project to run
tion of the 32-bit minicomputer system orig- out of funds much earlier than anticipated.
inally planned. In addition, the goal was set Ultimately, the computer was only finished
to complete the design of the 32-bit machine thanks to donations obtained from some of

62 IEEE Annals of the History of Computing

[3B2-14] man2010040060.3d 8/11/010 10:56 Page 63

a
designs had to use components manufac-
tured in the country or otherwise easily
obtained from abroad. Some essential com-
ponents such as the keyboard and floppy
disk units that represented a large percentage
of the total system cost had to be imported.
To complicate things even further, because
of the financial crisis, special import permits
only allowed for academic institutions and a
few public and private companies had to be
obtained from the government.
b In its final design, approximately 65 per-
cent of the Turing-850’s parts were from
Mexican providers. This percentage of com-
ponent integration from national providers
was achieved in some cases with significant
difficulty. For instance, few Mexican compa-
nies owned the technology to manufacture
double-layer printed circuit boards. This
technology was largely unknown in Mexico,
Figure 1. The Turing-850 (a) frontal and (b) lateral
yet was essential in the design of the com-
views.
plex layout of the Turing-850’s motherboard
containing two processors. At that time,
the Mexican company that manufactured
the companies involved and a contribution Turing-850’s motherboard was still testing
of BUAP’s Semiconductors Department. and adjusting its manufacturing process to
Because of these factors, a fully functional produce double-layer printed circuit boards.
Turing-850 computer prototype was not This situation produced numerous errors in
delivered until late 1984.25 the first versions of the motherboard that
The estimated project budget for the were difficult to find and fix.
Turing-850 was approximately $23,146,000 The Turing-850’s design featured a high-
Mexican pesos—about US$137,938 at the performance architecture based on a dual-
1984 exchange rate.17,26 Out of this budget, processor organization and a light pen
97 percent was spent on salaries and the input system that enabled free drawing on
rest on materials and other expenses. The the computer screen. Other more standard
limited budget forced the development features of the Turing-850 were a dual 5-1/
team to look carefully at the design options 4-inch floppy system, one parallel port
because there was no room for based on the Centronics standard, and one
experimentation. RS-232C serial port. The Turing-850 also
The original plan designed to advertise the included support in hardware and software
Turing-850 within the national industry to connect a hard disk drive compatible
included three configurations as possible sell- with Winchester technology.
ing options to the public. The basic configu- CP/M was the only operating system that
ration, with 64 Kbytes of RAM and no floppy was widely available in Mexico at that time.
disk units, had an estimated manufacturing For this reason, Zilog’s Z80 microprocessor
cost of approximately US$941. The typical was chosen for the Turing-850, not only be-
configuration with one floppy disk unit and cause of its compatibility with CP/M, but
the operating system cost approximately also for its low cost and wide availability
US$1,445. Finally, the fully configured op- within Mexico.
tion, which included an extra floppy disk The Turing-850’s design was organized
unit and a dot-matrix printer, had an esti- into six main modules: the software (mainly
mated cost of approximately US$2,047 at the basic input/output system, or BIOS),
1985 prices. motherboard, motherboard’s printed circuit
Figure 1 shows frontal and lateral views of layout, case, power supply, and video moni-
the Turing-850. tor. The design of each module was assigned
Because one of the main project goals was to a single developer. However, the team dis-
to ease the manufacturing of the computer cussed the design options proposed for each
in Mexico, the mechanical and electrical module, generally reaching consensus on

October–December 2010 63
 [3B2-14] man2010040060.3d 8/11/010 10:56 Page 64

The Turing-850 Project: Developing a Personal Computer in the Early 1980s in Mexico

the diverse design alternatives selected. This simultaneously generated an interrupt signal
practice had the beneficial effect of making to the PP. In turn, the PP received the inter-
all participants knowledgeable of and com- rupt, performed the indicated I/O operation,
mitted to the project. and once it was completed, generated a new
interrupt signal back to the CP by writing
Turing-850 motherboard and BIOS the result into the interprocessor communi-
The Turing-850’s architecture consisted of cation port. The PP employed 24 Kbytes of
two Z80 processors interconnected in a mas- static RAM to store its internal data.
ter-slave configuration. The goal of this archi- Figure 2 shows a block diagram of the
tecture was to create a high-performance main components in the Turing-850’s
computer by splitting the I/O tasks and stati- architecture.
cally assigning them to each processor. Addi- The Rainbow 100 personal computer from
tionally, because networks of personal Digital Equipment had a similar architecture
computers were becoming more popular, it as the Turing-850, with two processors inter-
was planned to use the slave processor to connected in master-slave fashion.27 The
manage the network functions in future main differences were that the Rainbow 100
models. An evaluation of the trade-off be- model employed a Z80 processor together
tween cost, performance, and ease of design with an Intel 8088. This architecture let the
convinced the design team that this architec- computer execute either an enhanced ver-
ture was the most effective option, given the sion of MS-DOS or CPM-86/80. Additionally,
relatively low cost of the processors. How- contrary to the Turing-850, interprocessor
ever, the dual-processor architecture required communication in the Rainbow 100 was
a more complex motherboard layout with a implemented through shared memory using
design that required laborious, manual work an arbitration mechanism built into the
because there were no automated tools avail- hardware. (A detailed description of the Rain-
able for the task. bow 100 model was published in 1984.27)
The Turing-850’s master or central pro- The Turing-850’s design enabled the possi-
cessor (CP) was a high-speed 6-MHz Z80B ble connection of extra peripheral devices
microprocessor. This processor was assigned through the peripheral bus in the mother-
to execute the operating system and user pro- board. The bus was made accessible by
grams and manage the hard/floppy disk using a special external connector designed
drives. The design also included a direct for that purpose. Finally, the parallel and se-
memory access (DMA) channel to enable rial port designs were compatible with the
mass storage devices’ direct access to memory Centronics and RS-232 standards. Figure 3
without requiring the central processor to in- shows the Turing-850’s motherboard.
tervene. The main memory system designed The firmware written for the Turing-850
for the central processor consisted of 64 Kbytes consisted of an interface layer in the BIOS
of DRAM. that enabled the Turing-850 to boot and exe-
A slower slave Z80A processor running at cute the CP/M OS with all the user applica-
4 MHz managed the rest of the I/O devices. tions. Additionally, diagnostics routines
This peripheral processor (PP) was also used were written to test the memory and I/O
to control the serial and parallel I/O ports system during initialization. Because no
and the computer terminal.25 microprocessor development system was
The communication between the central available, a special monitor program was
and peripheral processors was implemented written to facilitate program debugging. The
with a special parallel interprocessor commu- BIOS was written directly in the Z80 assembly
nication port that enabled the processors to language.
exchange data and control signals. The coor-
dination of the two processors was estab- Computer monitor, power supply,
lished through a handshaking protocol. The and case
central processor fetched a program’s instruc- Designing the monitor for the Turing-850
tion stream from memory and determined represented a challenge because there were
which operations would be either executed no specialized companies capable of manu-
locally or sent directly to the PP. All opera- facturing computer monitors in Mexico at
tions corresponding to the serial/parallel the time. Therefore, the most viable option
port or the video terminal were sent to was to employ a monochromatic CRT fabri-
the PP by writing the request into the inter- cated by one of the local TV set manufactur-
communication port, an event that ing companies, such as Admiral or RCA.

64 IEEE Annals of the History of Computing

[3B2-14] man2010040060.3d 8/11/010 10:56 Page 65

Interrupts

Interprocessor Peripheral Processor

Central Processor
Communication
Z80B Port Z80A

i8255

ROM ROM

DRAM SRAM

DMA Controller Serial

Dual FloppyDisk
Communication
Controller i8257 Interface

I/O Port Keyboard

i8255
Line Printer
Interface

Video Controller
Monitor
i8275

CentralBus Peripheral Bus

Figure 2. A block diagram of the Turing-850’s architecture. On top, the i8255 was used as the parallel
interprocessor communication port between the central and peripheral processors.

power company occurred frequently in

many regions of the country. These large var-
iations made it necessary to use costly exter-
nal AC voltage regulators to protect all the
electronic equipment. To cope with this situ-
ation, the team decided to design a special
power supply that did not require any exter-
nal protection equipment. At that time,
the technology of switching-mode power
supplies was relatively recent. Furthermore,
because the local industry lacked the tech-
nology needed to manufacture specialized
Figure 3. The Turing-850’s motherboard. The image high-frequency magnetic components, de-
shows the central and peripheral processors and signing a switching-mode power supply was
the rest of the circuitry. not a feasible option. Instead, the Turing-
850’s power-supply design employed a con-
ventional linear voltage regulator with a
However, to control the CRT, special ampli- special feedback circuit connected from
fiers had to be designed to fulfill the more the secondary to the primary winding of the
demanding specifications, in terms of band- transformer through opto-couplers.28 The
width, compared to the ones required by TV switching feedback circuit controlled
sets. Designing the control circuitry for the the power supplied to the voltage regulator
CRT took a considerable amount of time in the transformer’s secondary winding
given the system’s nonlinear characteristics. with a thyristor placed in the AC input line.
During the early 1980s, fluctuations in the The power supply design for the Turing-850
voltage supplied by the Mexican national could tolerate variations of ±25 percent in

October–December 2010 65
 [3B2-14] man2010040060.3d 8/11/010 10:56 Page 66

The Turing-850 Project: Developing a Personal Computer in the Early 1980s in Mexico

voltage (90–150 VAC) operating at 45 C microprocessor-based computers were

without requiring any external voltage designed and built at the DMA-BUAP. For ex-
regulator. ample, a multiuser system called SMU based
The mechanical design of the Turing-850’s on an S-100 bus design was fully functional
case was inspired by some of the computer from 1979 until the late 1980s.35 This system
terminals manufactured by NCR at the was able to serve up to 10 dumb terminals
time. The design included some of the ergo- connected to the S-100 bus. Another example
nomic recommendations for computer was the CP-UAP, designed with a NEC-V20
equipment from the International Organiza- microprocessor on an STD-bus.4,36 These
tion for Standardization (ISO). A thermody- microcomputer systems were built with
namic analysis of Turing-850’s case interior two main goals in mind. First, it was desirable
was performed to identify the optimal num- to have a practical platform for executing and
ber and location of the ventilation holes. As hosting the compilers for REC and Convert,
a result, an aesthetic slim case was produced two programming languages based on regu-
containing all the components, except the lar expressions and pattern matching, respec-
keyboard. The case was fabricated locally tively, that were designed by DMA-BUAP
from metallic sheets because it was difficult researchers.4,37 Second, these computers
to design and assemble the mold required were used as tools for teaching programming
for a lighter plastic case at a reasonable cost. skills to computer science and engineering
Several case prototypes were constructed in students. Given the limited human resources
wood and metal sheets to test the physical available, all these projects concentrated
layout of all the boards and the monitor. exclusively on designing the motherboards,
whereas the rest of the components were
Similar computers of the era imported.
Some other computer systems designed In IPN’s Center for Research in Comput-
and built outside the US during the early ing Technology (CINTEC), the first prototype
1980s had capabilities similar to those of of the computer Almita II was completed in
the Turing-850. Although we do not intend August 1984. The Almita II’s chief designer
to provide an exhaustive list of such com- was the Mexican engineer Miguel Lindig-
puters, these examples help illustrate similar- Bos. In its first version, Almita II had 256
ities and differences. Kbytes of RAM memory, two 360-Kbyte
floppy disk units, and an intelligent terminal
Other Mexican computers powered by an Intel 8031 processor. The
Arguably the earliest digital computer ever main processor was a state-of-the-art 16-bit
designed in Mexico was the Heterarchical Intel 80186 running at 8 MHz. Reportedly,
(AHR) parallel-processing machine,29 which this microcomputer’s processing speed was
was built at UNAM between 1979 and 3.4 times faster than that of the original
1983.30–34 The AHR was a special-purpose IBM PC.38,39 The first prototype of Almita II
computer designed for executing Lisp was a proof of concept; it did not include a
natively. The AHR employed a front-end case, and all the boards were assembled
minicomputer, where users could edit and using the wire-wrap technique.40
debug the Lisp programs while they were Lindig-Bos did not receive the required
being executed by the AHR back end in paral- budget for building 10 microcomputers
lel. The computer could host 5 to 64 Z80 from the IPN authorities until July 1986.
microprocessor chips. This task was successfully completed in
The Micro-SEP 1600 computer was April 1987.41 By that time, the design of
designed around 1985 by the Mexican Minis- those 10 microcomputers had some differen-
try of Public Education (SEP). This microcom- ces with the original Almita II, most impor-
puter was a slightly modified redesign of the tantly that the main microprocessor used
Radio Shack TRS-80 color computer contain- the Intel 80188. Partially because of this
ing a Motorola MC6809E microprocessor. change, the new design was called the IPN
Micro-SEP’s main purpose was to supply a E-16.39 Furthermore, the IPN authorities
low-cost computer to public junior high launched a mass-production program of the
schools. IPN E-16 in hopes of reaching a sustainable
The DMA-BUAP, headed by Harold V. self-production of microcomputers. By the
McIntosh, was one of the pioneer Mexican end of 1993, 1,189 IPN E-16 computers and
centers in the development of computer sys- its descendants were operating in most IPN
tems based on microprocessors.4 Several faculties.42 The model IPN E-16 is a rare

66 IEEE Annals of the History of Computing

[3B2-14] man2010040060.3d 8/11/010 10:56 Page 67

case of relative success in the history of Mex- Brazilian computers

ican computers. The Cobra consortium was formed in
There were a few other computer design Brazil in 1974. This consortium was a state-
efforts going on in Mexico during the owned company formed with the participa-
1980s. For example, the Digital Technology tion of the British company Ferranti and
Research and Development Center (CITEDI- several Brazilian private companies and pub-
IPN) developed a multiuser system (up to lic universities.46,47 Cobra’s purpose was to
three users) based on the Z80 processor and help create a national company capable of
a proprietary operating system in 1984.31 An- designing computer systems equipped with
other example was the digital computer its own technology. In the early 1980s, the
IMPetrón, designed by a team of four engi- company produced the Cobra-210 personal
neers in the Mexican Institute of Petroleum computer system. Cobra-210’s design
(IMP). IMPetrón was an 8-bit microcomputer included an 8-bit Z-80B microprocessor run-
developed around 1985; its main application ning at 5.85 MHz, 64 Kbytes of RAM, and
was the simulation of digital control two floppy disk units.48 To control periph-
designs.43 Because so little documentation is erals independently of the main processor,
available on these systems, we will not Cobra-210’s design included an Intel 8257
discuss them any further. DMA controller. Moreover, the machine had
the optional capability of including a floating-
British computers point coprocessor to speed up the execution
The Amstrad PCW 8256 was a personal of numerical applications. The Cobra-210
computer manufactured by the British com- was compatible with the computer’s previous
pany Amstrad,44 which started its commer- models: the Cobra-300 and Cobra-305.
cialization in 1985. The Amstrad PCW 8256 In 1984, Microtec Sistemas brought to the
model included a high-resolution monochro- Brazilian market a compatible PC-XT.49
matic monitor, 3-inch floppy disk drives, and Thanks to a governmental policy of protect-
a printer controlled through a proprietary ing the national IT industry, by 1986, the
protocol. Its design included an 8-bit Z80 Brazilian computer industry was able to sup-
microprocessor running at 4 MHz. The Z-80 ply up to 86 percent of its internal market.50
was capable of addressing up to 256 Kbytes However, in the beginning of the 1990s the
of memory using a paging technique. This Brazilian government changed this policy,
computer was designed as a word processor, allowing the importation of computers from
but it was also capable of running CP/M abroad. This produced a crisis in the national
and all its applications. computer industry, forcing it to look into
The BBC Micro was a personal computer other market segments where it could be
built by the British company Accorn in more competitive. The Brazilian computer
1982 for the BBC of London. It was a low- industry decided to reconfigure their new
cost computer design based on an 8-bit designs to target the commercial bank and
6502 microprocessor manufactured by MOS services sectors. Brazilian information tech-
technology running at 2 MHz. The most ex- nology companies such as Procomp, SID,
pensive models of the popular BBC Micro and Itautec still have an important share in
(models B and Bþ) came with 32 and 64 this market today.51
Kbytes of RAM memory and cost £399 (ap-
proximately US$764) and £499 (about Comparison
US$953) in 1982 and 1985, respectively. A The goals and scope of the Turing-850
phase alternate line (PAL) encoding video project had important differences with the
modulator inside the computer allowed it other Mexican computers we have described
to use an external TV. A special interface let here. As opposed to the Micro-SEP and the
the BBC Micro connect an external Z80 special-purpose, research-oriented AHR, the
microprocessor to execute CP/M. The extra Turing-850 was an entirely new design; it was
processor cost was £894 (approximately general purpose and had a high-performance
US$1,707). The BBC Micro project’s initial hierarchical dual-processor architecture.
objective was to design a computer that Furthermore, contrary to the rest of the
could be used to educate people about com- DMA-BUAP computers, the project included
puter technology. A special TV show from the design of the entire computer system,
the BBC called ‘‘Making the Most of the not just the motherboard. In comparison,
Micro’’ promoted the use of this computer the IPN E-16 employed a single, more
within the UK.45 advanced processor but was delivered almost

October–December 2010 67
 [3B2-14] man2010040060.3d 8/11/010 10:56 Page 68

The Turing-850 Project: Developing a Personal Computer in the Early 1980s in Mexico

Table 1. Comparison of the Turing-850 with other similar computers.

Feature Turing-850 IPN E-16 Amstrad PCW 8256 Cobra-210

Operating system CP/M MS-DOS CP/M Proprietary,
compatible CP/M
Central micropro- Z80 at 6 MHz 64 Kbytes 80188 at 8 MHz Z80 at 4 MHz Z80B at 5.85 MHz 64
cessor RAM and 8 Kbytes ROM 256 Kbytes RAM 64–128 Kbytes RAM Kbytes RAM
Peripheral Z80B at 4 MHz 28 Kbytes No secondary No secondary An optional floating-
microprocessor RAM and 4–8 Kbytes ROM processor processor point processor DMA
DMA channel, Intel 8257 channel, Intel 8257
Monitor B/W, 24  80, 12’’ RGB color, Monochromatic green Monochromatic green
640  400, 12’’ 32  90, 12’’ 27  80, 12’’
Floppy disk Two 5 1/4-inch units Two 5 1/4-inch Two 3-inch units Two 8-inch units
units
Serial port RS232C RS232C * RS232C
Parallel port Centronics interface Centronics Proprietary interface Proprietary interface
interface
Case Metal, single (CPU, video) Metal Plastic, single (CPU, Plastic, (CPU, video)
plus keyboard video) plus keyboard plus keyboard and
floppy disks
Year 1984 1987 1985 1983

* Data not available.

three years after the Turing-850. Finally, un- After a CONACyT’s initiative, the Turing-
like most Mexican computers of the era, spe- 850 system was also exhibited in some im-
cial care was given in the Turing-850’s portant industrial expositions. At the time,
design to ease its manufacturing and opera- CONACyT had created a special fund called
tion within Mexico. the Trust Fund for Shared Risk53 to foster co-
Table 1 shows a feature comparison be- operation between academia and industry in
tween the Turing-850, IPN E-16, the high-tech projects. Under this scheme, CON-
Amstrand PCW 8256, and the Cobra-210. ACyT provided half of the investment needed
The Turing-850 and IPN E-16 were devel- to produce the technology developed within
oped within academia by a small group of academia. Despite the advantages offered by
developers with little previous experience in this initiative, the restrictions imposed by
computer design. In contrast, the Amstrand PFIEC to promote the creation of a national
PCW and Cobra-210 were produced by com- computer industry, and all the efforts made
panies with ample experience. Despite those to publicize the project, no Mexican company
limitations, Table 1 shows that the Turing- showed interest in producing the Turing-850
850 was a competitive machine when com- or any other computer designed by Mexican
pared to similar computers of the era. researchers. In late 1985, the next stage of
the project, consisting in the design of a
Conclusions 32-bit minicomputer, was definitively can-
The Turing-850 and other similar Mexican celed and the original design team dis-
computers showed that despite the limitations banded. A few years later, some members of
faced and the adverse economic conditions the original Turing-850’s design team formed
prevailing in the country at that time, it was a start-up company to design and manufac-
feasible to develop state-of-the-art technology ture electronic devices. The rest pursued aca-
for personal computers within academia. demic careers in Mexico and abroad.
After the project was finished, the Turing- The Turing-850 project failed to reach the
850 computer was demonstrated in diverse goal of transferring its technology to a Mexi-
academic and industrial meetings. Specifi- can-owned company for its widespread produc-
cally, the project was presented at CONACyT, tion. Furthermore, the design of the more
UNAM, IPN, the University of Las Americas, ambitious 32-bit computer was canceled prema-
and the Arturo Rosenblueth Foundation.52 turely. These failures occurred for several reasons.

68 IEEE Annals of the History of Computing

[3B2-14] man2010040060.3d 8/11/010 10:56 Page 69

First, the economic and political condi- participate in joint ventures with state-
tions prevailing in the country when the owned companies.15,47,54
project was initiated changed drastically dur- The Turing-850 and other Mexican com-
ing its development. The Mexican govern- puters developed in the early 1980s showed
ment shifted its original policy from a that Mexico had the technological capabil-
greenhouse scheme to promoting liberaliza- ities to initiate the development of a national
tion in the computer sector. Furthermore, computer industry. Unfortunately, the finan-
government policies on technology failed to cial crisis of the 1980s together with other
establish clear goals and coordination mech- factors prevented this from happening.
anisms at a time when the development of
computer technology required sophisticated Acknowledgments
capacities and links between industries, gov-
We thank Trevor Misfeldt, Francisco Serrano-
ernment, and universities. Another factor
Osorio, and the anonymous reviewers for
was the lack of development of the few Mex-
their valuable comments and suggestions,
ican companies that were manufacturing per-
which greatly helped to improve this article.
sonal computers at that time. Instead of
Francisco Rodrı́guez-Henrı́quez acknowl-
using the protective mechanisms imple-
edges support from CONACyT through the
mented by the government to develop their
CONACyT project number 60240.
own technology or seeking the collaboration
of the few research centers in the country,
those companies decided to acquire the tech- References and notes
nology from abroad. A representative exam- 1. L.A. Lomnitz and L. Cházaro, ‘‘Basic, Applied
ple of this trend was the Mexican firm and Technological Research: Computer Science
Printaform. This company acquired the rights and Applied Mathematics at the National Au-
from Columbia Data Systems in the US to tonomous University of Mexico,’’ Social Studies
manufacture and sell, with some success, of Science, vol. 29, no. 1, 1999, pp. 113–134.
the personal computer Columbia Printaform, 2. UNAM’s central university city campus (known
which was similar to the original IBM PC-XT. as C.U.) was officially inaugurated in 1954. On
Finally, the appearance of the IBM PC-AT 2 July 2007, C.U. was included in the Unesco
in August 1984, together with the success of World Heritage list.
MS-DOS and the IBM PC-compatible com- 3. A. Cantarell and M. González, eds., Historia de
puters, made CP/M-based computers such as la Computación en México: Una Industria en
the Turing-850 obsolete very quickly. Desarrollo [History of Computer Science within
In contrast to Mexico, Brazil and other Mexico: A Developing Industry], Hobbiton
countries, such as the so-called Four Asian Ediciones, 2000.
Tigers (Hong Kong, Singapore, South Korea, 4. G. Cisneros, ‘‘La computación en México y la
and Taiwan), were investing heavily in tech- influencia de H. V. Mcintosh en su desarrollo’’
nological development during the 1980s. [Computer Science in Mexico and the Influence
Brazil is especially significant because it of H. V. McIntosh on its Development], Reunión
shares with Mexico a similar level of eco- Nacional de Matemáticos en Homenaje al Dr. José
nomic and industrial development. Adem, ‘‘Medio Siglo de Matemáticas en México:
Brazil achieved relative success in creat- Estado Actual y Perspectivas,’’ 1991; http://delta.
ing an important computer industry capable cs.cinvestav.mx/~mcintosh/oldweb/pothers.html.
of producing its own technology, in part due 5. CENAC-IPN, ‘‘Antecedentes históricos del centro
to the protective mechanisms implemented nacional de cálculo (CENAC)’’ [Historical Antece-
by the Brazilian government at the right dents of the National Calculation Center]; http://
time. Combined with crucial support since www.cenac.ipn.mx/Conocenos/Historia.html.
the early 1970s of both public and private 6. C. Gonzalez-Brambila, J. Lever, and F. Veloso,
sectors, this fostered the development of ‘‘Mexico’s Innovation Cha-Cha,’’ Issues in
the computer sector. However, in Mexico, Science and Technology, vol. 24, no. 1, 2007,
similar protective mechanisms did not pp. 51–58.
work in the same way. A possible explana- 7. Formerly called the Universidad Autónoma de
tion is that, contrary to Mexico, the Brazil- Puebla (UAP), its name was changed on 1 April
ian banking sector supported the creation 1987 to Benemérita Universidad Autónoma de
of a national computer industry because it Puebla (BUAP).
foresaw the importance of computers to 8. R. Casas, ‘‘Ciencia y tecnologı́a en México.
their business market. Finally, multinational antecedentes y caracterı́sticas actuales’’ [Science
companies in Brazil were convinced to and Technology in Mexico: Background and

October–December 2010 69
 [3B2-14] man2010040060.3d 8/11/010 10:56 Page 70

The Turing-850 Project: Developing a Personal Computer in the Early 1980s in Mexico

Current Characteristics], Revista Mexicana de 23. On 18 February 1982, the Mexican government
Sociologı´a, vol. 45, no. 4, 1983, pp. 1323–1334. announced an official devaluation of the Mexi-
9. M. Trevino, ‘‘Regulation of Technology Transfer: can peso from 28.50 to 46 per US dollar. The
The Mexican Experience,’’ J. Technology Trans- exchange rate of the Mexican currency reached
fer, vol. 1, no. 4, 1989, pp. 46–51. 80 pesos per dollar by the end of that year.
10. E. Calderón, ‘‘Desarrollo de la computación en 24. The original IBM PC had a retail cost in the US
México’’ [Development of Computer Science in of approximately US$3,000 with 64 Kbytes of
Mexico], INNOVA, Semana Nacional de Innova- RAM and one 5 1/4-inch floppy disk.
ción y Calidad de Administración Pública, 1980; 25. G. Arenas-Muñoz et al., ‘‘Computer Turing 850.
http://turing.iimas.unam.mx/~remidec/difusion/ Project Turing 85: Development of a General
textos/Calderon-Alzati-Comput-Mex-2003.pdf. Purpose Microcomputer. Final Report,’’ tech.
11. R. Prieto-Dı́az and S. Willson, ‘‘The Impacts of report, Dept. Microcomputer Applications,
Computers on the Latin American Countries,’’ SIG- Sciences Inst., Autonomous Univ. of Puebla,
CAS Computing Soc., vol. 11, no. 2, 1981, pp. 2–9. 1985 (in Spanish); http://cs.aaue.dk/~do/
12. Plan de Fomento a la Industria Electrónica y de publications/mypapers/turing85.pdf.
Cómputo, in Spanish. 26. The exchange rate average value for 1984 was
13. A. Borja, El Estado y el desarrollo industrial: la of $167.8 Mexican pesos for US$1.
polı´tica mexicana de cómputo en una perspectiva 27. K. Olsen et al., ‘‘Digital’s Personal Computers,’’
comparada [The State and Industrial Develop- Proc. IEEE, IEEE Press, vol. 72, no. 3, 1984,
ment: Mexican Computer Politics from a Com- pp. 283–299.
parative Perspective], M.A. Porrúa,, 1995. 28. D. Ortiz-Arroyo, ‘‘Fuente de poder con control
14. L.A. Gómez-Gómez, ‘‘La información y su con- de consumo de potencia’’ [Power Supply with
texto social’’ [Information and Its Social Con- Power Consumption Control], Proc. XIV Int’l
text], Boletin Informativo de la Direccion General Conf. Electronic Eng. (ELECTRO 91), Chihuahua
de Bibliotecas, UNAM., vol. II, no. 2, 1999. Inst. of Technology, 1991, pp. 21–25.
15. J. Dedrick et al., ‘‘Economic Liberalization and 29. The term heterarchical was introduced by this
the Computer Industry: Comparing Outcomes computer’s designers to indicate that all pro-
in Brazil and Mexico,’’ World Development, cessors in the architecture were organized in a
vol. 29, no. 7, 2001, pp. 1199–1214. horizontal fashion, as opposed to a hierarchical
16. J.A. Fox-Lozano, ‘‘Polı́tica informática en pers- arrangement.
pectiva comparada’’ [IT Policy in Comparative 30. A. Guzmán and L. Lyons, ‘‘La computadora
Perspective], bachelor’s thesis, Computer Eng. AHR: Construcción de un procesador con LISP
Dept., ITAM, Mexico, 1996. como su lenguaje principal’’ [The Computer
17. Centro de Estudios de las Finanzas Públicas, AHR: Construction of a Processor with Lisp as
Cámara de Diputados, H. Congreso de la their Main Language], tech. report AHR 80 10,
Unión, ‘‘Estadı́sticas históricas indicadores IIMAS, UNAM, 1980; http://www.cic.ipn.mx/
macroeconómicos 1980–2006’’ [Statistics in aguzman/sourcepubli.html.
Historical Macroeconomic Indicators 1980– 31. A. Guzmán, ‘‘A Hierarchical Multiprocessor
2006], 2009; http://www.cefp.gob.mx/intr/ Lisp Machine,’’ Proc. IEEE Workshop Computer
e-stadisticas/copianewe_stadisticas.html. Architecture for Pattern Analysis and Image
18. NCR’s R&D department in Puebla was one of Database Management, IEEE Press, 1981,
the first R&D departments ever created by a pp. 309–317.
transnational company within Mexico. 32. K. Norkin and A. Guzmán, ‘‘Diseño y construc-
19. G.M. Montes, ‘‘La trayectoria académica de Luis ción de una máquina paralela heterárquica:
Rivera Terrazas’’ [The Academic Background of Reporte final del proyecto AHR’’ [Design and
Luis Rivera Terrazas], Boletı´n de la Sociedad Mexi- Construction of a Parallel Heterarchical Ma-
cana de Fı´sica, vol. 13, no. 2, 1999. chine: AHR Project Final Report], tech. report
20. Rivera-Terrazas founded the Department of AHR 82 21, IIMAS, UNAM, 1982; http://www.
Semiconductors and the Physics Institute that cic.ipn.mx/aguzman/sourcepubli.html.
now carries his name at BUAP. 33. A. Guzmán-Arenas, ‘‘Diseño y construcción de
21. C. Hunter and E. Farquhar, ‘‘Introduction to the computadoras paralelas mexicanas’’ [Design
NS16000 Architecture,’’ IEEE Micro, vol. 4, no. 2, and Construction of Mexican Parallel Com-
1984, pp. 26–47. puters], Descubrimientos y Aportaciones Cientı´fi-
22. G. Arenas-Muñoz et al., ‘‘Project Turing 85: De- cas y Humanı´sticas Mexicanas en el Siglo Veinte,
velopment of a General Purpose Microcomputer. S. Estrada, ed., Academia Mexicana de Ciencias
Preliminary Description,’’ tech. report, Dept. y Fondo de Cultura Económica, 2008.
Microcomputer Applications, Sciences Inst., Au- 34. Interview with A. Guzmán-Arenas, chief de-
tonomous Univ. of Puebla, 1982 (in Spanish). signer of AHR, Mexico City, 20 Jun. 2008.

70 IEEE Annals of the History of Computing

[3B2-14] man2010040060.3d 8/11/010 10:56 Page 71

35. Interview with H. Garcı́a-Monroy, chief designer 53. Fideicomiso de Riesgo Compartido, in Spanish.
of DMA-BUAP microcomputers, Puebla, Mexico, 54. P. Bastos-Tigre and A.J. Junqueira-Botelho, ‘‘Bra-
20 June 2008. zil Meets the Global Challenge: IT Policy in a
36. A. Licona et al., ‘‘La computación en el instituto Postliberalization Environment,’’ The Information
de ciencias de la UAP’’ [Computer Science at Society, vol. 17, no. 2, 2001, pp. 91–103.
the Science Institute of UAP], 1985; http://delta.
cs.cinvestav.mx/~mcintosh/oldweb/pothers.html. Daniel Ortiz-Arroyo is an
37. H. McIntosh and G. Cisneros, ‘‘The Program- associate professor at the De-
ming Languages REC and Convert,’’ SIGPLAN partment of Electronic Sys-
Notices, vol. 25, no. 7, 1990, pp. 81–94. tems and a member of the
38. M. Lindig-Bos, ‘‘Una microcomputadora mexi- Computational Intelligence
cana de 16 bits compatible con la IBM-PC’’ [A and Security Laboratory at
Mexican 16-Bit Microcomputer Compatible Aalborg University, Denmark.
with IBM PC], Proc. 2nd Int’l Conf. Electrical His research interests include
and Electronics Eng., 1985, pp. 76–79. computational intelligence, computer architecture,
39. M. Lindig-Bos, R. Espejo, and M. Partida-Tapia, machine learning, information retrieval, social net-
‘‘La microcomputadora IPN E-16’’ [The Micro- work analysis, and security. Ortiz-Arroyo has a PhD
computer IPN E-16], Informes Técnicos del IPN, in computer engineering from Oregon State Uni-
Mexico City, Mexico, July 1988. versity. Contact him at [email protected].
40. M. Lindig-Bos, ‘‘Antecedentes e historia del
CINTEC’’ [Background and History of CINTEC], Francisco Rodrı́guez-
Informes Técnicos del IPN, Mexico City, Mexico, Henrı́quez is an associate pro-
Dec. 1988, p. 61. fessor in the Computer Science
41. Interview with J.-C. González-Robles and E. Department at the Center for
Rodrı́guez-Escobar, two members of the IPN Research and Advanced Studies
e-16’s design team, Puebla, Mexico, 30 May 2008. (CINVESTAV-IPN), Mexico. His
42. M. Lindig-Bos, ‘‘Informe final de actividades’’ research interests include cryp-
[Final Report of Activities], Informes Técnicos tography, computer arithmetic,
del IPN, Aug. 1993. and efficient implementation of algorithms on
43. Interview with A. Partida-Romo, IMPetrón’s de- reconfigurable hardware devices. Rodrı́guez-
sign team, Mexico City, Mexico, 30 May 2008. Henrı́quez has a PhD in electrical and computer engi-
44. D. Pountain, ‘‘The Amstrad PCW 8256,’’ Byte, neering from Oregon State University. He is a mem-
vol. 11, no. 3, 1986, pp. 333–340. ber of IEEE, the Academia Mexicana de Ciencias,
45. C. Whytehead, ‘‘BBC Microcomputers,’’ 2009; and an alumni member and research associate of
http://acorn.chriswhy.co.uk/Computers/ the Information Security Laboratory at Oregon
BBCMicros.html. State University. Contact him at francisco@cs.
46. E. Luzio, The Microcomputer Industry in Brazil, cinvestav.mx.
Praeger/Greenwood, 1996.
47. E. Adler, ‘‘‘Ideological ‘‘Guerrillas’ and the Carlos Artemio Coello
Quest for Technological Autonomy: Brazil’s Do- Coello is a professor at the
mestic Computer Industry,’’ Int’l Organization, Center for Research and Ad-
vol. 40, no. 3, 1986, pp. 673–705. vanced Studies (CINVESTAV-
48. Clube Old Bits, ‘‘Outras linhas’’ [Other Lines]; http:// IPN), Mexico. Coello Coello
www.cobit.xpg.com.br/micros/cobra305.htm. has a PhD in computer science
49. A. Santos, ‘‘Personal Computers Are Hot,’’ IEEE from Tulane University. He
Spectrum, vol. 33, no. 7, 1996, pp. 34–39. serves as an associate editor
50. Computing and Informatic Museum (MCI), for the IEEE Transactions on Evolutionary Computa-
‘‘História dos computadores no Brasil’’ [History tion, Evolutionary Computation, Journal of Heuristics,
of Computers in Brazil]; http://www.mci.org.br/ Pattern Analysis and Application, and Computational
micro/index.html. Optimization and Applications and as a member of
51. A. Botelho, ‘‘From Industry Protection to Indus- the editorial boards of the Soft Computing, Engineer-
try Promotion: IT Policy in Brazil,’’ Center for Re- ing Optimization, and International Journal of Com-
search on Information Technology Organizations putational Intelligence Research. Contact him at
Technical Report, 1999; http://crito.uci.edu/ [email protected].
papers/1999/brazil-case-10-99.pdf.
52. Interview with F. Serrano-Osorio, member of
Turing-850’s design team, Puebla, Mexico, 11
Jul. 2008.

October–December 2010 71