First-Hand:EPROM Cassette (1977) as a Predecessor of the USB Flash Drive

contributed by Remo J. Vogelsang, IEEE Life Member, President/Owner Remo Vogelsang AG Engineering Informatik from 1974 to 2015

Fitting technical information is available by clicking the ^[index] . Please click Smiley^{'s [index]} for background information or maybe a funny tale ! To return click ↑.

Evolution of Pluggable Memory Devices

Ubiquitous - from a few kilobytes to terabytes

Memory Cassette
The pluggable, non-volatile memory system of unmatched reliability and ruggedness with a capacity range from 4KB to 56KB for program and data storage in numerical control systems, communication processing, testing equipment etc.

Conceived already in 1977 and produced from 1979 till 1988 the Memory Cassette stands at the very beginning of the evolution of the pluggable and portable semiconductor memory devices. Products of comparable characteristics got commercially available only around 1990 - apart from a sales attempt made on a magnetic storage solution. Called Memory Card, Memory Stick, USB Flash Drive, Flash Drive, USB Stick - or known under a variety of other names - they still carry the same distinctive features as their forerunners, except for their tremendous capacity-size-price advantages, made possible by the incredible progression in semiconductor manufacturing. This historical succession represents a notable, most comprehensible example, illustrating the evolution predicted by Moore's Law:
A more than ten-million-fold capacity increase (64 KB to 640 GB) at a one-hundred-fold space decrease (240 cm³ to 2.4 cm³), resulting in a more than one-billion-fold capacity to size augmentation, taking place within 35 years, all with the benefit of a 10-million-fold cost decrease ($ 4.00 per KB to $ 0.40 per GB).

Features     These more and more appreciated and popular devices were gradually aligning to the available technology in parallel with the rising expectations of their users, still retaining their ever valid core attributes:

• Detachable and portable
• Non-volatile
• Ruggedness (no harm by surface scratches)
• Durable and reliable, no moving parts.
• Immune to electromagnetic interference (unlike floppy disks)
• Field-programmability
• Single supply voltage
• Hot swapping possible
• Tailored capacity range
• Opportune interface

Exploding Evolution     It started as a sort of a convenient template for reliably handling machining data under rough environmental conditions, replacing cumbersome paper tapes (punched tape) or avoiding flimsy floppy disks spinning in bulky peripheral drives. On the system level, it perfectly fitted the prevailing 8-bit and 16-bit microprocessor platforms. Over the years, gaining in capacity while shrinking in size, the pluggable memory progressed towards the now ubiquitous data storage devices, suitable for all sorts of applications, with possible storage capacities beyond anybody's expectations.

Memory Cassette	Bubble Memory Cartridge	Memory / Flash Card	USB Flash Drives	Type
1980	1981	1990	1998	Phase-in
4 to 56 KB	8 to 32 KB	64 KB to 240 GB	128 MB to > 1 TB	Capacity Range
160 x 100 x 15 mm	60 x 45 x 20 mm	32 x 24 x 2.1 mm (sample)	40 x 12 x 5 mm (sample)	Approx. Size
memory standard	proprietary	several	USB	Interface
none	none	several	USB mass storage	Standard

• Memory Cassette: Designed in the then popular Eurocard format, it featured all necessary attributes enabling the use of a succession in technology ranging from EPROM to EAROM to EEPROM to BBSRAM. Initially designed as EPROM/EAROM-Cassette, it contained from 4KB to 56KB of Erasable Programmable Read Only Memory.

• Bubble Memory Cartridge: Fujitsu Limited launched the 'Bubble Memory Cartridge' targeting an applications range similar to the one covered by the 'Memory Cassette'. The applied technology was based on using a thin film of a magnetic material to hold small magnetized areas - known as bubbles - as information storage. But it never became a serious contender against the steadily progressing market of non-volatile semiconductor memories. The cartridge could not be mounted like a printed circuit card into a card magazine, but required instead a special mounting socket. Applying it required the expenditure of a peripheral device.

• Memory Card or Flash Card: Designed in many different formats, sizes, interfaces and standards to cover the markets for cell-phones, digital cameras, industrial applications, etc., mounting is typically in a concealed fashion, fitting enclosures of the specific device. The technology used is based on the flash memory, the yet unchallenged, modern offspring of the EEPROM. A specialized version known as the SIM Card (Subscriber Identity Module) is used in mobile phones. Its functionally independent size ranges from a credit card to a small plate of 6x5 mm (for embedded-SIM). The meanwhile almost obsolete, credit card-sized PC Card was initially conceived as a memory device. It soon morphed to a series of popular peripheral interfaces, designed for laptop computers. It is still used as conditional-access module in television equipment.

• USB Flash Drives: The technology of this ubiquitous storage media of the day - mounted in the typically protruding stick-fashion - is also based on the flash memory. Apart from the clever choice of the Universal Serial Bus (USB) as the interface, the unique form of its connector - what a lucky strike - facilitates all sorts of form factors.

EPROM/EAROM Cassette

What a neat, novel, elaborate, handy gadget

An opened EPROM-Cassette VMC110-AF
Bottom half with embedded printed circuit board, assembled with 12 EPROM devices, protection diode arrays and additional interface circuitry.
Top half containing openings, admitting exposure to ultraviolet light for erasure.
Adhesive foil used for closing the openings in the unit as well as for its labeling. Shown is a fitting translucent labeling pocket with the inserted tag.

The Memory Cassette was built as a plug-in unit including a printed circuit board and an injection-molded housing comprising two halves. The housing had ribs (with latch notches) on its narrow sides, which were aligned with the printed circuit card. The unit was outfitted with a connector according to DIN 41612 with 2 x 32 contacts, fitting 0.1 inch (2.54 mm) grid pitch. As a whole - encasement inclusive - it was conforming to the norm of the Eurocard (printed circuit board) format of 160 x 100 x 15 mm, a most popular size at the time. Fitting into standard IEC 60297 racks, it was therefore facilitating the integration into given equipment.

Description    

The printed circuit board included two to fourteen semiconductor memory devices, a number of protection diode arrays, some necessary interface circuits and the connector. Inserted into the bottom half of the housing with its ribs, together they formed the 160 x 100 mm standard card size. The top half of the housing contained openings at locations corresponding to the quartz windows of the underlying memory devices, admitting - where applicable - exposure to an ultraviolet light source. Both halves of the housing were kept together by six resilient tabs, which were engaging when both halves of the housing got pressed together. Due to the solvable plastic material used, it was possible to make a permanent connection if the resilient tabs were wetted with solvent. The bottom half of he housing was fitted with a recess for posting an adhesive type plate. A suitable recess was also made to the top half, in order to facilitate the positioning of the adhesive foils used for closing the openings on the cover, as well as for appropriate labeling.

Originally, the VMC110-A/B Cassette was planned to contain from 4KB to 56KB of Erasable Programmable Read Only Memory (EPROM). However - in anticipation of further development - it was designed such, that its printed circuit board may also have got equipped with other types of memory devices, like EAROM, EEPROM, BBSRAM. Read access may have taken place at full memory speed for any of the device type. To exclude damage due to static discharge, special protection was provided to the relevant signal terminals. Provisions were made, that allowed cassettes to get interchanged without power interruption and ensuring proper cassette insertion. Necessary circuits for status check under program control were provided, allowing the detection of specific cassette types. Programming of any of the cassette types was done electrically via the appropriate function terminals. Operation required a single 5V supply only (except for programming). A UV-lamp, supplemented with a fitting tray, allowed erasing the entire data contents of the cassette at once.

The interface used was fully based on the access and control architecture of the then prevailing 24-pin semiconductor memory devices, which were known to any designer using the 8-bit and 16-bit microprocessor platforms. This turned out to be advantageous for using the Cassette Memory in various systems. Then our supplementary development set, containing hardware- software- and assembly elements, also provided valuable support. For the whole family of PDP-11/03^TM and LSI-11^TM microprocessor systems, we had a full set of mature, well-engineered, proprietary products for interfacing, programming, assembling and maintenance at our disposal.

Cassette Products    

Type	Capacity	Memory Technology	Storage Devices	Programming
VCM 110-A [VCM110-A]	4 … 28 KB	EPROM erasable programmable read-only memory	(2Kx8) [2716]	UV erasable / electr. progr.
VCM 110-B [VCM110-B]	8 … 56 KB	EPROM erasable programmable read-only memory	(4Kx8) [2732]	UV erasable / electr. progr.
VCM 110-C	4 … 28 KB	EEPROM electrically erasable progr. read-only memory	(2Kx8) [2816]	electr. erasable and progr.
VCR 110-A	4 … 28 KB	BBSRAM battery-backed random-access memory	(2Kx8) [MK48Z02]	regular random-access write

Are you somewhat confused about EAROM / EEPROM / FLASH memory? No surprise.

Application of EPROM-, EEPROM-, RAM-Cassette   ^{[translation]
from top-left to bottom-right}
- Intelligent Terminal Programming Station
- Handling and Operation
- Teleprinter Programmiing Station
- Communication Processor
- Integration into LSI-11 Systems:
- Numerical Control System
- Erasing by UV-light Exposure
- System Scheme
- Integration into Euro-Card systems

Application    

• The classic: To reliably handle template-like machining data in numerical control systems, sample data in testing equipment, collect data in data loggers under rough environmental conditions commonplace in machine shops or on outdoor sites.
• Storage of application and/or customer oriented program and data portions in so-called embedded systems - often operating under real-time computing constraints - with dedicated functions within a larger electrical or mechanical system. Here field-programmability was a very desirable feature, as it could reduce turnaround time and cost for exchanging modified firmware.
• Any application requiring portable memory under severe environmental and handling conditions.
• Simplifying servicing and repair work (in case of presumed or real fault) by allowing the exchange of firmware by unskilled personnel.

System-Information    

• In true read-only applications - i.e. a programmed cassette served as a data source - interfacing was simply done through galvanic connections between the appropriate terminals of the cassette and the microprocessor memory bus. No additional electric components, such as line drivers/receivers, were needed. The interfacing task was reduced to arranging a clever mechanic connecting scheme.
• For off-line programming - i.e. data programming of the cassette was done externally from the target equipment or a dedicated programming station - any brand of those off-the-shelf EPROM programmers construed to program one single memory chip at a time (almost everybody had one like that at his disposal), was good enough. Provided one afforded undergoing all those laborious programming procedures necessary. A simple gadget made of fitting connectors, a rotary switch and some wiring was required as a supplement.
• Equipment, such as a data logger or a cassette programming station, which absolutely required built-in programming facilities, needed a dedicated, built-in programming module with adequate data access to whatever source required.
• As mentioned before, there existed a whole set of hardware and firmware to efficiently merge the cassette system with the family of PDP-11/03^TM and LSI-11^TM microprocessor systems: a) The EPROM/RAM memory system VMC11-A which may have contained up to 32 KB/(28 KB) of EPROM and/or RAM, also providing all the electrical and mechanical means to readily interface to the VCM110-A/(B) EPROM Cassettes. b) The EPROM-Programmer VMP11-A, an option enabling programming of the VCM110-A/(B) EPROM Cassettes as well as the EPROM/RAM memory system VMC11-A.

Development

Seize the chance

Particular requirements of a customer of mine initiated the development of this pluggable memory. For his planned custom machine-tool he needed a numerical control (NC, CNC) with properties beyond the capabilities of common, off-the-shelf solutions available at that time. He also expected some appropriate device - sort of a template - for handling machining data reliably under the rough environmental conditions prevailing in his machine shops. He was very reluctant, if not defiant in accepting paper tape (punched tape) or floppy disk (8-inch and 5¼-inch emerging on the market then). He had strong arguments for sure, although fitting solutions did not exist yet. However, he signaled to provide adequate compensation for following through his viewpoint.

Conception     It was spring 1977. The notion of the pluggable memory took form on a hastily jotted sketch during the train ride home from an inspiring meeting with this customer, located in the western (French speaking) part of Switzerland. It showed a number of EPROMs, placed on a fittingly encased printed circuit board, the size of the then widely utilized Eurocard. Openings, placed in the encasement at corresponding locations to the quartz windows of the underlying EPROM-devices, allowed exposure to ultraviolet light for erasure. Finally, a removable label holder, normally covering these holes, carried the identification.

Choosing a practical interface was as crucial as it was simple - regarding the prevailing context. Its characteristics are best understood by consulting the description in the ^[2716] EPROM datasheet. At that time, developers were familiar with the simple scheme: Read-access required eleven input signals to select the address of the particular cell, eight multiplexed outputs to read its content and one control signal to select the memory device. One additional input was available to control programming (write). By suitably arranging these signal lines, supplementing them by appropriate power lines and fitting cassette identification signals and by properly protecting connections against damage by electrostatic discharge, an unpretentious, well defined, widely acceptable interface got created - avoiding red tape.

Design and Implementation     Already in mid 1976 we had developed our own Q-bus compatible 8 KB EPROM Memory Card ^[VME11-A] , based on the then popular 1K x 8 EPROM circuit ^[2758] , in conjunction with an appropriate Programmer Card ^[VMP11-A] , to be able to accomplish our goals in projects based on PDP-11/03^TM and LSI-11^TM microprocessor platforms.

Meanwhile - in order to fulfill rising demands and to keep up with progress in memory technology - the development of the 32 KB Memory Card ^[VMC11-A] , based on the newly available 2K x 8 EPROM circuit ^[2716] got under way, featuring four-fold capacity and versatile usability. One such card covered 32 KB, one half of the intrinsic memory space (16-bit address) of the LSI-11, also allowing to freely mixing EPROM with SRAM in 4 KB segments. Anticipating the emergence of electrically erasable PROM - fitting EEPROM did not exist on the market yet - we included necessary functions, allowing to switch among EPROM and EEPROM.

The availability of this innovative set of mature, well-engineered, proprietary memory products, in conjunction with our proficient team in designing, prototyping and debugging such devices, had unintentionally propelled us into an advantageous position to jump-start this cassette project and to implement it efficiently and successfully. Development efforts and costs remained on a limited scale, since interfacing to the cassette is simply done through galvanic connections between the appropriate terminals of the cassette and the internal memory bus of the Memory Card. So the interfacing task boiled down to arranging a clever mechanic connecting scheme. Then, the providently designed Programmer Card - using the features of the LSI-11 On-line Debugging Tool (ODT) - needed only a few minor extensions at the firmware level to serve the additional requirements. Refraining from a compromise using some sort of a makeshift cover, I decided for the professionally produced, injection-molded plastic encasement, which required an expensive tool and a long lead time.

Conceived during spring 1977 and the customer's order coming in towards the end of 1977, we had completed the cassette design in the middle of 1978. Since fall 1978 the cassette system was operational in several ongoing projects (using makeshift encasements). We had postponed the design release of the encasement for as long as possible to ascertain that no detail got overlooked; early 1980 the first regular batch was produced.

Marketing and Patent Issues

Never knew a prospector yet that died rich

Marketing Efforts     Engineering was our business model - not production. So it was ok to design and manufacture technically attractive components - such as modules - on a small scale. In other words we would pick opportunities to rapidly close certain demand gaps. After a couple of years, when the applications technology had jumped to these novel opportunities, we would abandon such products. However, the cassette product offered qualities, worth to consider a longer term commitment. Fact was, the proliferation of solid-state memory with its incessant increase of capacity, the emergence of the EEPROM - a predecessor of flash memory, combined with the eventual prospering of the surface-mount technology (SMT) would pave the way for a vast market of portable solid-state memory. The launching of the 'Bubble Memory Cartridge' by Fujitsu Limited in 1981 - targeting an applications range similar to the one covered by the 'Memory Cassette' - raised hope, that we could profit from their advertising such memory products. Confident about our product, we were looking forward to compete. But soon, Fujitsu's initially impressive promotion ceased.

Right during conception of the cassette, I became aware, that it might make sense to up the stakes in this case and to venture for a marketing trial. However, having never underestimated what it takes, to market a novel product, I was cautious not to squander our resources. We constrained such efforts to a modest promotion, periodically distributing ^[leaflets] and void Cassette ^[specimen] . Over the years (until about 1988), a fair quantity of cassettes and cassette systems got manufactured to cover our own requirements and to supply our regular customer base. The cassette system proved to be of extraordinary reliability and versatility, to the benefit of many applications.

Patent Applications     Realizing, that aiming to exploit a presumably vast market of portable solid-state memory would be too big a world, I considered it advisable, to at least apply for patents, thereby preserving the rights for possible licensing. Due to pertinent experiences, I was cautiously optimistic regarding patent procedures and feasible benefits. I was mindful too, that besides the USA also Japan would become of great importance. Buying and using EPROMS in respectable quantities over a number of years - it turned out, that US manufacturers adhered to standards of acceptable quality limit (AQL), not suitable for dense semiconductor circuits. Customers had to carry the burden of intolerable quality losses. Apparently aware of this handicap, the emerging Japanese producers were shipping zero defects (ZD) parts - therefore rapidly gaining market share. Eventually I applied for patents in four countries:

Country	Appl. Date	Appl. Nr.	Patent Nr.	Status
Switzerland	22.09.1980	7 083/80-4		first response 14.01.1985, abandoned
Germany	04.11.1980	P 30 41 596.4	DE 30 41 596 C 2	granted 15.12.1983
USA	14.09.1981	S.N. 301,859		fruitless, abandoned
Japan	22.09.1981	148932/'81		no official response, abandoned

The patent procedures went on at a sluggish pace. The results, after long waiting times, were meager. Extending expedient efforts - i.e. diligently instructing examiners - would have become prohibitive. Worse, considering tardy patent procedures versus the speed of change in the semiconductor field, the cassette in its original form would have soon become outdated. I decided to abandon this path. You can't win them all!

Round-Up

What will be, will be

For us doing engineering, the goal is to successfully fulfill an urgent demand by efficiently applying current and near future technology, avoiding red tape. Usually there is little time left for contemplating circumstances. In retrospect, I am trying to make a fair assessment of work done and of historical facts:

• What began in 1977 on a limited scale for special purpose applications, had developed - not to say exploded - to become a ubiquitous article of daily use, produced in gigantic numbers.
• For sure, the Memory Cassette was a first portable memory. It took years until a comparable follow-up emerged. Due to the almost non-existent marketing, its perception was undervalued.
• Patent attorneys and examiners were struggling to appreciate the intrinsic idea. At that time - personal computers were years away - few people were knowledgeable about computer technology. It appears that the appraisal of the mechanical aspects of the cassette prevailed. Patent efforts and money spent returned no apparent benefit.
• This story may serve as a fitting example to highlight Moore's prediction, written in his seminal 1965 paper.
• Right now, it feels that memory sticks are small enough, and that their capacity is abundant; but then, how many times did we fool ourselves, predicting saturation regarding bulk memory! Let the expert speak: "… It eventually saturates out. I guess I see Moore's Law dying here in the next decade or so, but that's not surprising." - from Gordon Moore in an interview, published in IEEE Spectrum's April 2015 issue's Special Report '50 Years of MOORE'S LAW'

References

To get to the bottom of things ….

Behind the Scenes

Tales, some serious some funny !

Copyright © 2015 by Remo Vogelsang