Yeno OrdiMagic alias Epoch Alpa-1

April 25, 2025

For young children in the 1970s and 80s, (compact) cassette tapes with narrated stories (basically children audio books on tapes) were a popular pasttime, not least because it did not require parental presence, because cassette players were relatively inexpensive, and because content (i.e. audio book cassette tapes) was cheap or could even borrowed from friends and libraries.
If you were a manufacturer of educational toys in the 80s, and you wanted to sell a toy that combined the advantages of narrated stories and that could e.g. verify the answer to a question posed to the child in the story, you could not go completely digital.

The problem was not so much the educational part (some keys and a small microprocessor do the trick), but the audio part and the synchronization between audio and educational element. For example, the audio should stop when the user is supposed to enter an answer and only after an answer the system should continue.

Playing audio in acceptable quality digitally was (at that time) very costly in terms of memory and computation. There are estimations that 30 minutes of uncompressed cassette audio need about 30 MB of memory. 30 MB for a computer of say, 1986, would not only mean a RAM size that would exceed the typical main memory size of almost all models (a 1986 Atari ST would maybe have 1 MB of RAM), but also that there would be no affordable mass media to store all this data on. The 1987 HD Floppy has only a capacity of 1.44 MB. Ideally, you would like to use some sort of audio compression to bring the 30 MB down to something more processable. Unfortunately, MP3 is not standardized until the end of 1991 and chips that can decode MP3 in realtime exist only some years after that.

But engineers are trained to overcome such difficulties, so what alternatives does one have? Well, you could use speech synthesis and license Texas Instruments’ Linear Predictive Coding technology. TI’s own 1978 “Speak & Spell” calculator toy does that in order to read the spelling of words. Unfortunately, this system basically allows for only one, very robot-like voice and one has to pre-compute the needed data using TI’s R&D department (i.e. it cannot be done on the toy in realtime).

The other alternative is much more low-tech: you use a stereo audio cassette and have the (mono) audio on one track and the data needed for the educational part on the other track. In order to read the data track, you need a small modem circuit in your toy, and voila: good (mono) audio quality (for the time) and all the possibilities of a small computer.

This is what a large Japanese toy manufacturer must have thought in 1986 when it decided to bring out a toy on exactly this technical basis. This company was the Japan-based “Epoch” which still exists today and which is well-known for Japan’s first successful programmable console video game system, the 1981 Cassette Vision (here Cassette means cartridge), and for its Doraemon and Sylvanian Families toy and video game productions.

Interestingly, I did not found any mention that this toy was released in Japan itself (although that is possible, maybe even likely). Instead, there was a version in the UK, named the Epoch Alpa-1, and a version in France called the Yeno OrdiMagic. Yeno was a French game electronics reseller that sold products under its own name produced by Epoch, Sega, and others. “OrdiMagic” seems to be a word made out of the French word for computer (ordinateur) and the English word magic (I guess the more French “OrdiMagique” was already taken).

It is obvious (when one thinks about it) that Epoch’s toy needed to be modified for the corresponding market. On the one hand, the audio content needed to be recorded in the corresponding language, on the other hand, the keyboard needed to contain all language-specific characters. This might be a good opportunity to show the device and all of its interaction elements. Let’s start with the French version.

Image
Yeno OrdiMagic

The lower part of the device consists of a keyboard with 40 keys. 26 are italic small caps characters, 10 are digits which also double as keys for special french characters, one key is labelled “vrai” (true), one “fault” (false), one has a sad face (which is the correction key) and one has a happy face (which is the Enter key). Above the keyboard there is a switch labelled “Musique” and “Programme”, an LED indicating power (“MARCHE”), one switch called “PASSE”, and a green LED labelled “JUSTE” and a red one labelled “FAULT/COMMENCER”. The upper part has a volume knob, a loudspeaker and a large decal showing the product name and four rabbits in colorful attire. The device has a collapsable handle on the top and the opening to slide in a cassette sideways on the right side. Next to the opening there is a single button for fast forwarding the cassette or ejecting it. On the bottom there is an opening with a sliding top for 4 “D” cells. That’s it. No power switch or any other interface.

The device is switched on by inserting a cassette. The switch on the top allows to chose either the normal cassette player function “Musique” or the educational mode “Programme”.

The following paragraph is a little bit of speculation based on the analysis of an original audio cassette. Still, it should be quite accurate. However, the audio cassette was (due to an imperfect copy process) not runnable on my device. Here is what I assume to happen:
In the “Programme” mode, the device plays the audio track and listens to the data track. If there is data on the data track, the playing stops and an entry with the keyboard is waited for. If the entry corresponds to the data, the “JUSTE” LED is lit, else the “FAULT” is lit. If the answer was correct, the audio part is continued until the next data is found.

The software/audio cassette catalogue contains 13 titles, and it seems there was never another title published. These 13 cassettes are described on the Alpa-1 box and there was a plastic box for Alpa-1 cassettes which exactly holds 13 copies. These titles are:

  • Introductory Cassette: A: Visit to Murky Manor, B: Meet the Stars
  • 8501 FUN WITH LETTERS
  • 8502 WORD MAGIC
  • 8503 LET’S EAT
  • 8504 BUILDING WORDS
  • 8505 ALL ABOUT ANIMALS
  • 8506 BEGINNING CONCEPTS
  • 8507 THE WORD DETECTIVE
  • 8508 ON LAND, SEA, AND AIR
  • 8509 ONCE UPON A WORD
  • 8510 NOTHING BUT NUMBERS
  • 8511 FUN WITH ARITHMETIC
  • 8512 MEET THE EXPERTS

The differences of Yeno’s OrdiMagic to Epoch’s Alpa-1 seem to be superficial.

Image
Epoch Alpa-I

The case says “Epoch” and “Alpa-1”, the picture on the sticker is different. Instead of small caps italic characters, the keyboard contains big caps letters, and there are no special characters on the row with the digits. The “PASSE” button is missing.

We have analysed the (very, very bad copy of the) audio for “Murky Manor”, the A-side of the introductory cassette, probably included in every Alpa-1 box. The MP3 copy of the audio can be found here:

MP3 version of the “Murky Manor” side of the introductory cassette

It is basically a series of spelling tasks woven into the story of the spooky Murky Manor. The first word to spell by the user (at around 4:48) is the word “cat”. On the left channel of the audio there is some data activity from 4:51:92 to 4:52:42, for around 0.5 seconds (you cannot hear that on the MP3 copy because this is removed by the encoding). We have cut this part, and slowed it down 10-times to 5 seconds and increased the volume by a lot. Here it is.

0.5 seconds of the data track slowed down 10-times

The trick of using a stereo compact audio tape with mono audio on the one channel and data on the other channel is not new. The 1985 Teddy Ruxpin toy, a talking bear could move his eyes and his mouth based on signals on one channel of a built-in cassette tape, thus “speaking” with more facial expression (although I think these signals were not logic commands, but directly controlled the corresponding motors).

The hardware of the OrdiMagic is very simple and cheap. Apart from the cassette drive, you have a small PCB with a NEC D7506C microcontroller, and two audio ICs:

Image
Yeno OrdiMagic PCB

The D7506C is a 4-bit microcontroller with 1 kB of ROM and 32 bytes of RAM. It probably runs at a frequency of 400 kHz.

The keyboard on the device is really, really awful. The keys look like calculator keys, but they are not.

Image
Keyboard Caps

If one opens the case one finds that every key has a rigid part on the top (which is simply part of the keyboard cap matrix which is a single, large molded part). A key can therefore only move to the bottom where a protrusion on the bottom (also part of the mold) presses against some sort of sensor matrix mat on the PCB. The rigid top tends to break (given the age) and the the key cap is quite lose in its opening. My OrdiMagic tends to “eat” the tapes, therefore I use an analog-to-cassette adapter for a car cassette player to play tapes.

There you have it, an early solution to the question of how audio cassette tapes and a small computer can be used for educational purposes, the Yeno OrdiMagic or the Epoch Alpa-1, respectively. This toy is very, very rare today as it probably was not sold in substantial numbers at the time.

Technical Data

CPU: NEC [email protected]
RAM: 32 bytes
ROM: 1k
Markets: UK, France
Released: 1986
Original Price: 500FF

References

  • TILT JEUX ET MICRO: LE GUIDE 86, pp. 22+23

Sony Librié EBR-1000EP – The first Device with an eInk Display

January 4, 2025

The device I want to talk today about is not a computer in the sense that you can run your programs on (although it probably has all the technical means to do so), but a humble
ebook reader. It’s not even the first ebook reader (that honor goes to the Rocket eBook). But it is not only the first ebook reader with an eInk display, it is the first ever device with an eInk display. Meet the 2004 Sony Librié EBR-1000EP.

Image
Sony EBR-1000EP without its cover

From the viewpoint of today, the ebook reader is not very exciting: 6″ eInk (non-touch) display. It has a large number of small keys (including a complete keyboard), and it runs of 4 AAA batteries (which is great from a collectors point of view because then you don’t have to chase down a power supply). It’s age shows in using Sony Memory sticks and offering only a mini USB interface, but apart from these details, it looks like many other readers today.

However, from a 2004 viewpoint, the eInk display is a top-notch innovation. Nowadays, a manufacturer can get a complete eInk display from E Ink Corporation and integrate it into a device. There are mainly only a few types of display models to chose from and they differ only in size and product generation. When E Ink Corporation brings out a new display, it is typically incorporated in the next generation of devices across manufacturers. Until 2004, there was no complete eInk display you could buy. E Ink Corporation was around since 1997 and the media regularly reported on their development of the eInk technology, but you could not buy a device containing eInk. From what I understand, E Ink Corporation at that time could sell you the electronic ink itself (i.e. the spheres that contain black and white bally), but not much more. If you wanted to create a device with that technology, you had to create the display using these spheres yourself. And that’s exactly what Sony did.

It was such a complex undertaking that Sony did not even do it by themselves. A company called Toppan Printing put the spheres on a substrate, Philips created the display control electronics, and Sony integrated the display into a device. This collaboration took three years to suceed.

The resulting device was the Sony Librié EBR-1000EP. “Librié” seems to be one of these Japanese marketing terms that were termed by manufacturers to have a name for a product range that sounds cool and possibly European. “EBR” is the Sony product code for “e-Book Reader”. The weight of the unit including case and batteries is 300gr, intendly the same weight as an average Japanese paper book.

As there was no established ebook market, there was also no established ebook format, and, subsequently, no established infrastructure to create and sell ebooks in such a format. Therefore, both had to be created for this product.

On the ebook format side, Canon and Sony developed a format called Broad Band eBook (BBeB) in order to compress the memory need of a book and to enforce a DRM system for these books. BBeB-formated books are stored as sources as LRS files (an XML format). These LRS sources are then compressed and encrypted as LRX files, or just compressed, but not encrypted as LRF files. An average 250-page book required about 500 kB compressed. LRX and LRF files can then be distributed and loaded to the reader either via a PC and USB (by an application) or via Memory Sticks. BBeB was even multimodal, and there were books with an audio track. Sony readers supported BBeB until 2010 after which this format was replaced by EPUB (a format that was initially released in 2007).

On the ebook distribution side, Sony and some large Japanese publishers created a joint venture called “Publishing Link” where users could purchase books from their PCs (not from the reade itself because there was no Wifi). I use “purchase” here in a loose sense as all DRM-protected books expired 60 days after they were created. In this sense, the users could not purchase books, but loan them. Readers could choose books from seven groups (or “clubs”) ranging from novels to business books and had to pay either ¥315 for a single book or join a club and get access to up to five books a month for ¥210 each. The EBR-1000EP came pre-loaded with three Japanese dictionaries and an encyclopedia.

The EBR-1000EP was replaced in 2006 by the PRS-500 with 64 MB of storage, and in 2007 by the PRS-505, the first Sony ebook reader with a conventional E Ink display. While the EBR was sold only in Japan, the PRS models were sold all over the world. This marks the transition of the ebook market from a technical novelty to a (relatively boring) mass market.

Because it was only sold (at least officially) in Japan, because it was not very successful, and because devices like ebook readers are not very emotional devices (and are more often thrown away than kept), EBR-1000EPs are very rare today. Funnily, they are currently also not expensive (I bought mine for the princely sum of about 8 Euros).

Technical Data

Device type: ebook reader
Introduction: 2004 (Japan)
Supported ebook formats: BBeB
Memory: 10 MB
OS: Linux-based
CPU: Motorola DragonBall MX1 (ARM920T core)
Display: 6″, 800 x 600 pixels, 4 grey scales
Interfaces: Memory Stick, DC in (proprietary 3 prong), mini USB, headphones
Power: 4 AAA batteries (enough for 10,000 pages)
Weight: 190 gr without batteries and case, 300 gr with batteries and case
Initial price: ¥41,790 (~ $260 USD in 2004)

References

RANDOC Podcast created by an AI

November 8, 2024

I recently learned about Google’s Notebook LLM, a Large Language Model that allows you to create prompts on documents you upload to the system. It can do many things, but one thing it can do quite excellently is to create a 2-person podcast episode on the uploaded documents completely automatically. As you can also use a web page as the document, I, of course, immediately had it create a podcast episode on this blog. I’m note sure whether it is worth listening to completely, but maybe you can marvel at what is possible today for a few minutes. Here is the audio:

pro_VME VMEST: The rarest of the contemporary Atari ST Clones

April 11, 2024

If you ask people what the first Atari ST-family clones were, most will answer with names like Hades (1996, 68040) and Milan (1998, 68040). Very few will mention the Medusa T40 (1994, 68040, although one could argue that this is not a clone, but just a very invasive accelerator) or the spectacularly rare GE-Soft Eagle (1994, 68030). Regular readers of this blog will of course know the 1988 IBP 190ST as the first Atari ST clone (which aimed at bringing the ST architecture to the industrial world).
Therefore, I was very excited when I was offered a 19″ Atari ST computer at the VCFB 2022 that I did not know anything about. I received it recently and it turns out to be a previously unknown Atari ST clone, possibly even the only one existing today.
“Previously unknown” unfortunately also means that there is very little information available and we have to speculate more than usual to fill in the gaps. In such cases, the best approach is to examine the corpse the computer first and then start from there. The computer came in a (not so handy) 19″ rack, clocking in at a total over 30 kg. Here it is:

Image

The ST would have been the first module on the left (it is out in this picture), then some other, 3rd party modules, a little bit wider panel with nothing behind in the middle, an FDD-and-HDD module to its right (also missing), and then a black panel with a number of interfaces on the right. Two of these interfaces say “Midi” and “ACSI”, but these are the only hints to a possible ST inside. Now, the rack also came with two peripherals which scream “Atari ST”: an SM124 monitor and a ST-style keyboard which connect to the (non-ST-standard) interfaces on the ST module.

The modules of the rack which are interesting to us bear a company name: “proVME”. This company (whose official name is “pro_VME”) played a small role for the Atari ST as it was the first company to introduce a really compatible accelerator for the Atari ST, the “Hypercache ST” from 1989. It consisted of a replacement PCB for the 8MHz Motorola 68000 CPU of the ST. This PCB hosted a 16MHz version of the same CPU and an 8kB instruction cache and a 8kB data cache in order to cope with the rest of the 8Mhz Atari. They followed this model up with the Hypercache ST+ which used two IDT7174 chips as the cache (interestingly, this and the following accelerator, the hyperCache 030, are later on sold and further developed by the same company GE-Soft that creates the Eagle in 1993/1994…).
Ok, now it is time to take the ST module out of the rack and have a look at it.

Image

First, we can see the mainboard. There are a lot of chips we recognize from a Mega ST:

  • 2 D27C1001A-12 EEPROMs with a label “TOS143DLY”
  • 1 Yamaha YM2149F (sound chip)
  • 1 C100110-001 Atari ASIC (DMA)
  • 2 C025912-38 (MMU)
  • 1 MC68901P (MFP)
  • 2 MC68B50P (ACIA)
  • 1 empty socket (for an 68881 FPU?)

The CPU is a MC68HC000FN16 (which actually runs at 16 MHz). Apart from the faster CPU, the only special thing is that there are two MMU chips, and it is unclear to me why. Also the TT has two MMUs (one for the ST RAM and one for the TT RAM), but I doubt that we have two types of RAM here, too.

The EEPROMS probably contain a modified version of TOS 1.043, but I did not check that. There are two more empty ROM sockets.

And then there are some chips we would normally not expect in an ST:

  • 2 IDT 7174 8kx8 SRAM&Address Comparator
  • 1 MC68681P (DUART)

The IDT chips are the same as in the Hypercache ST+ accelerator, so in conjunction with the faster CPU it seems as if the accelerator is here already integrated into the architecture. Sweet!

We are also missing some chips:

  • the entire RAM
  • the Shifter
  • the GLUE

Additionally, we have four connectors on the board:

  • 2 x DIN 41612, type C, 96-pin (one VME interface and one proprietary interface)
  • a 74-pin bus on the top
  • a 70-pin bus on the bottom

The rest of the space is filled with discrete logic, GALs, PALs, and smaller chips.
There is a 2.4V NiCd battery with 30 mAh (which we removed briefly after this picture was taken because it already started to leak).

We will find the RAM and the Shifter on the daughterboard, but the GLUE is nowhere to be seen.
My hypothesis is that they could not use the GLUE chip because of the differences of this computer to a standard Mega ST and because the VMEbus has some needs, too. If this is true, probably the colorful selection of PALs, GALs, and discrete logic contains the needed logic that is normally packaged in the GLUE chip.

Finally, at the front, there is a 9-pin sub-d keyboard and a 15-pin sub-d display interface as well as a 2-direction switch labeled Reset on the top and xxxx on the bottom.

The back of the PCB is empty with 5 locations having patches:

Image

Oh, and yes, finally: the name of the model can be found on the back of the mainboard as well:

Image

It’s the VMEST (this names makes a lot of sense, because it is an ST for a VMEbus system).

On the top of the mainboard, there is a daughterboard:

Image

The PCB says PRO12, it is unclear to me what this means. The board contains four rows of chip space for 8 1Mbit chips, i.e. a maximum of 4 MB of RAM which fits perfectly to the ST architecture. As you can see, only two rows are populated, so this board has 2 MB of RAM. As I said before we can also find the Shifter chip on the daughterboard, as well as a 32 MHz clock chip. The daughterboard also contains a row of connector pins. The bottom of the daughterboard is boring:

Image

We will now have a look at the other elements of the rack that belong to the VMEST.

HDD-AND-FD-module

This is simply a 19″ module that contains a double-sided, double density floppy disk drive and a SCSI harddisk drive:

Image

The HDD is a Fujitsu M2622SA 330MB SCSI-2 drive and the FDD is a TEAC FD-235F 112-U.

Image

Breakout Panel and ACSI-SCSI Converter

The VMEST mainboard has a second 96-pin connector (the one on the bottom). This connector basically provides signals of all of the peripherals that can be connected to a Mega ST. You can see the backplane side of the connector here:

Image

The cable of this connector ends up in three cables. One of these cables leads the breakout panel, the other one to a PCB which more or less dangles in the back of the rack:

Image

The breakout panel is also a PCB that looks like this:

Image

This is the front of the panel:

Image

This panel contains mainly 25-pin d-sub interfaces for a printer, a modem, MIDI, probably two more serial interfaces (named T3 and T4), ACSI, and a floppy interface. The MIDI interface is also a 25-pin d-sub plug, the other interfaces are the same interfaces as on a Mega ST.

Let’s talk about the mysterious PCB inside the rack. It is connected to the second cable coming from the VMEST board. Here is a (bad) picture from it:

Image

It is a pro_VME development (with another ProXX name on it) and it ends up in a cable leading to the (SCSI) HDD: it is an ACSI-SCSI converter.

The third cable from the VMEST mainboard finally is a floppy cable that can be connected either to the FDD or to the connector on the breakout panel.

Monitor and Keyboard

Image
Image

These are quite boring. Standard Mega ST keyboard with a standard cable that has one end with the proprietary VMEST connector. Standard SM124 display (newer version) with the proprietary VMEST connector on the display cable. I hope to report on the pin assignment soon, but then, also, who can use this information except me? Both keyboard and monitor boast a pro_VME sticke

The VMEST on the Internet

Finally, something that does not take long writing about it. There is one(!) mention of the VMEST that I could find. It is a short snippet from a Serbo-Croatian computer magazine from January 1990:

Image

Also, there is an indirect mention of what could be a VMEST in the German “ST Computer” magazine (1993/10): “Normen had his first contacts to the company proVME at the Atari Fair 1989 that then showed the first modification of the ST to 16 MHz plus cache in form of a VMEbus card which was meant as a low-cost graphics terminal for industrial applications.” (“Auf der ATARI-Messe 1989 knöpfte Normen dann die ersten Kontakte zur Firma proVME, die damals den ersten Umbau des ST auf 16 MHz mit Cache gezeigt haben, und zwar als VME-Bus-Board, gedacht vor allem als preiswertes Grafikterminal für Industrieanwendungen.”) Finally, there is another mention in a German Atari ST magazine from 1989, but it gives no details and list “Becker Ing.-Büro” as the distributor, which is possible, but also not that likely.

Why aren’t there more VMESTs out there?

Difficult. I have the feeling that the VMEST is a (hopefully working) prototype and that the machine was never really sold. Maybe the VMEbus version of the IBP 190ST was available earlier or did cost less. If you happen to also own a VMEST, I would certainly like to hear from you!

Conclusion

There you have it: The VMEST from pro_VME is an industrial clone of the Mega ST for VME racks. It consists of three pieces: The motherboard that also includes a 16MHz accelerator and offers sockets for up to 4MB of RAM, an ACSI-SCSI converter and a breakout box for more interfaces. This is the only machine of this type I have ever heard of and the information on the Internet about it are virtually non-existent. It might be a prototype and probably never hit the market.

From a collector’s point of view…

the VMEST is a dream. Probably the only one in the world. Take that, mass-market IBP 190ST with your sales in the dozens 🙂

Technical Data

Manufacturer: pro_VME
Model: VMEST
CPU: Motorola 68000@16MHz
RAM: 2 MB
ROM: 256 kB
The rest of the data is the same as an Atari Mega ST.
Initial Price: 4160 DM
Introduced in: 1989
Produced devices: unknown

References

Sony PTC-500,-550,-300,-310 Family of Early PDAs

January 28, 2024

In the late 80s, early 90s, mobile computing still meant traditional text-and-keyboard devices although the computers used for that purpose were already quite small. You could get portable versions of WIMP desktop computers (e.g. the Macintosh Portable), but these computers required you to sit down in order to use them.

Mobile devices for casual use while standing required new paradigms like the pen-based computers from 1991 onwards (yes, there was the 1989 GRiDPad 1900, but this was a too rare model and the Operating System was basically an unaltered MS-DOS).

One of the difficulties of pen-based computers was that you really need some sort of handwriting recognition in order to exploit the full potential of such an approach. The problem is that you need a lot of compute power to do that, more than what was available for a small computer at that time that can be hold in one hand if you wanted to have a battery life of more than two or three hours. This problem would haunt almost the entire first generation of devices that wanted to do so until mobile CPUs become more powerful (or until Palm came up with its “Graffiti” input method which simplifies the problem from recognizing any handwriting to recognizing exactly one handwriting that the user had to learn).

This problem of entering text into a mobile computer is much worse if your language is not alphabet-based, but consists of several thousand symbols like Chinese or Japanese. For these languages, in the early 80s even displaying all characters was not trivial, not to speak of designing a keyboard that allows one to efficiently enter the needed symbols. For these languages the use of a pen or brush is not only convenient, but a necessity.

And so, finally, we are in the time, use case domain, and country where Sony saw a need for a pen-based PDA (Personal Digital Assistant) in Japan in 1990 (a PDA is an electronic appliance that aims to be a digital multi-tool for all your Calendar, Address Book, and Notebook needs, and that ideally, can be synced with your desktop computer). And that’s exactly what Sony designed: it’s family of “PTC” PDAs, which came out only in Japan (PTC means PalmTop Computer and you find Sony’s “PalmTop” logo all over these products).

Image
Sony PTC-500

The first model was the PTC-500 released in 1990 for 198000 Yen (or about 3200$ in 2023). It was quite large (205x45x158mm closed) and heavy (1.3 Kg), but had a pen and could recognise more than 3500 Japanese and Western characters. The (electrostatical) pen still needed a cable to the computer, and you could not simply write anywhere on the display (although you could scribble everywhere), but had to fill pre-defined boxes with a single character at a time. In order to achieve this technical marvel (and it was a marvel at the time), Sony had to use the latest in algorithmic wizardry: Fuzzy Logic!

Image

Fuzzy Logic expands the classical 2-value Boolean Logic (knowing only TRUE and FALSE) to a number interval between 0 and 1, with the ends encoding classical logic, but also being able to represent a logical level of e.g. 0.5. Fuzzy Logic values are able to represent vagueness and imprecise information. Applied to the problem of recognizing symbols I guess this allows for some level of tolerance when entering these symbols compared to the expected standard pattern of this symbol.

The PTC-500 was clearly targeted towards business people as depicted by the ads for that device.

Image

It did not only look (when closed) like a black calendar, it also offered the applications that would be needed by traveling business men, e.g. the management of contact information. Showing a vaguely handset-like indentation on the upper half of the case tells you where to put the handset of your (fixed line) telephone because it can (touch-tone) dial a number from the contact list (as an accessory you could even get one of these external acoustic couplers which used a microphone and a small loudspeaker to receive and send data over a telephone line having a handset strapped on the top). Business-like was also the price of the PTC-500 (remember, the equivalent of $3200 nowadays). It had a large 512×342 backlit, monochrome LCD display with a diagonal of xx inch. Apart from a 2 MB ROM it also featured 320 KB of RAM.

The software of the PTC-500 falls into 3 main categories: Planning, Directory, and Idea Filing.

Image

For planning, users can define “actions” that have a time period associated to them. These actions can be managed as To-Do lists or as appointments in a calendar.

Image

The directory allowed to store contacts with many possible data fields like birthdays, address, phone numbers, etc. Different views on this data allowed to e.g. have a phone directory or a list of customers in a certain area.

Image

The “idea filing” tool is a cross between a note taking and a presentation tool. It allows you to note down ideas and informations on cards and to categorize them, and also to create presentations that can be shown to customers.

A calendar, a world clock, FAX software and a tool to exchange files completes the software.

Data Modules

Image

It was not easy to exchange data with a PTC PDA. There is no wireless (or wired) network that it can connect to and the device does not have a standardized card slot as the PCMCIA standard was only published in 1990. There is the proprietary extension interface, but if offers only a 2″ floppy diskdrive and an acoustic coupler as (optional) communication equipment. Therefore, all PTCs have an integrated (proprietary) solid state memory card slot for battery-buffered SRAM cards. The available sizes are 64 KB (PTM-064) and 256 kB, but I never saw the latter. Funnily, there was one other Sony device that could use these memory cards and that is the (Japanese) Video Titler XV-J777. This machine had a PTM-064 card included and (being basically an MSX2 machine) is interesting in itself (but this is a story for another day).

Other Models

The PTC-500 was replaced in 1991 by the PTC-550.

Image
PTC-550

Instead of the (mainly useless) opens-like-a-book form factor of the 500, the 550 was a classical slate wrapped in a plastic cover. Therefore, it was a lot smaller when being used and about the same size when stowed away. The 550 weighs a little bit less and also costs a little bit less (30kYen less)

Also in 1991, a low-end model completed the line-up.

Image
PTC-300

The PTC-300 halved the weight and almost divided the prize by 3. For that money, you got half the resolution, and a resistive instead of a capacitative pen (at least it did not need a cable anymore). You lose the extension interface, but gain an infrared communication port that lets you talk to other PTC-300s). One reason for the weight loss is that the device now uses a smaller, less widespread battery (which also can be replaced by 2 AA batteries in a holder). The PTC-300 has a lid on the side which contains also the pen when not used.

Finally, in 1992, the last model of the line is the PTC-310.

Image
PTC-310

This is simply a PTC-300 without the lid but enclosed in a leather-like binder for a more upmarket look.

One problem of the PTCs is the usage of the early soft paint that many mobile products featured at the time. The problem with this generation of paint is that is turns into a sticky gooey mess after some time that costs one quite an effort to touch if not treated. The seller of my device has treated this problem, but even so e.g. the pen of my PTC-300 does not really want to leave its notch in the lid.

Sony PTC PDAs are, in my opinion, very cool, early devices, but you cannot develop your own software for it (I think). Being borderline appliances, not computers, and having being sold only in Japan (having a Japanese-only user interface and documentation), there are probably not many collectors attracted to these devices. The PTCs are quite rare even in Japan with the PTC-310 maybe being the rarest.

Sony did not produce further PTC models, but from 1994 re-entered the PDA market with their (2 model) line-up of Magic Cap devices, the PIC-1000 and PIC-2000.

Image
PIC-1000

Especially the PTC-550 speaks design-wise a very similar language to the PIC-1000 from 3 years later. Also the PICs use a Sony videocamera battery, they also have a resistive pen (like the PTC-300s), it is a landscape format device, and a display with about the same resolution. The differences are that the PICs only weigh half so much, are much more graphically oriented, and are meant more for communication, also integrating a modem.

Regarding PDAs, finally Sony also produced their own versions of Palm-powered PDAs from 2000 to 2005 called the CLIÉ devices.

Technical Data

Model: PTC-500
CPU: Motorola 68HC000@8MHz
RAM: 320 KB
ROM: 2 MB
Mass memory: 320 KB battery-buffered RAM
Display: 512 x 342, backlit
Pen technology: capacitative with cable
Batteries: NP-55 (NiMH), about 6h runtime
Interfaces: power, 2-inch memory card adapter, proprietary 28-pin extension interface
Size: 205 x 45 x 158 mm closed, x 298mm opened
Weight: 1.3 Kg
Initial Price: 198 kYen
Introduced in: 1990


Model: PTC-550
CPU: Motorola 68HC000@8MHz
RAM: 320 KB
ROM: 2 MB
Mass memory: 256 KB battery-buffered RAM
Display: 512 x 342, backlit
Pen technology: capacitative with cable
Batteries: NP-55 (NiMH), NP-77H also possible
Interfaces: power, 2-inch memory card adapter, proprietary 28-pin extension interface, proprietary 26-pin interface
Size: 215 x 25.5 x 190 mm
Weight: 970 g
Initial Price: 168 kYen
Introduced in: 1991


Model: PTC-300
CPU: Motorola 68HC000@8MHz
RAM: 288 MB (battery-buffered)
ROM: 2.5 MB
Display: 320 x 256, backlit
Pen technology: resistive
Batteries: Sony BP-2PT (also BP-2EX and BP-2X might be usable), but there is also a holder for 2 AA batteries included
Interfaces: power, 2-inch memory card adapter, infrared
Size: 210 x 105 x 20 mm
Weight: 435 g
Initial Price: 65 kYen
Introduced in: 1991


Model: PTC-310
CPU: Motorola 68HC000@8MHz
RAM: 288 MB (battery-buffered)
ROM: 2.5 MB
Display: 320 x 256, backlit
Pen technology: resistive
Batteries: Sony BP-2PT (also BP-2EX and BP-2X might be usable)
Interfaces: power, 2-inch memory card adapter, infrared
Size: 210 x 105 x 20 mm
Weight: 360 g without / 560 g with cover
Initial Price: 68 kYen
Introduced in: 1992

References

Small update on the tv-computersystem 6800

January 10, 2024

I recently brought the tv-computersystem to a vintage computer meeting in Stuttgart, Germany (click here if you are interested) and learned a little bit more about this machine.

The computer works when switched on (after around 48 years…) without even having to replace capacitors. We could not find the channel on a TV that gave us a stable or even unstable picture (although one could see that there was something out there somewhere).

So, a participant (thank you so much, Joachim Oswald!) proposed to try to convert the signal that goes into the tv modulator into a composite monitor signal and started on this endeavour right away (my electronics “knowledge” is so limited that I cannot only sit next to a hardware wizard and bring coffee if requested). It turns out there is already such a signal available on the PCBs! All you have to do is to

  • connect the (inner) pin of the cinch plug to Ground
  • connect the (outer) ring of the cinch plug to the “MO” pad on the power-supply-and-modulator PCB (the brown PCB)
  • Voila! Instant composite signal of a very good quality.
Image
My computer connected to an 80s Philips Green Monitor
Image
Here you can see the very good picture quality a little bit better

We experienced that the light pen did not work in the beginning, but increasing the intensity on the display did the trick (we used a green display where the bright parts might not be that bright as on a TV). Remember that the light pen will only work on a CRT display.

And now, for the first time, I could see a working tv-computersystem! I entered the first example program from the manual (adding two numbers) and it worked once I entered all needed values and selected “run”. The system is quite easy and smooth to use. The light pen works even a few millimeters above the CRT surface, and you can glide it easily from cell to cell. Together with the command to fill an entire memory page with the same value (that’s the first command on the top), entering values is quite fast (especially as you can enter hex digits), I imagine much faster than using switches and lamps.

I was now also able to (roughly) weigh the computer. It comes at about 3.5 kg in total.
weight

The last open question is about the clock speed the system uses. A M6800 of that era in a standard configuration would run at 1 Mhz, but I cannot confirm this yet. There is one quartz oscillator with 27 Mhz on a PCB (which seems to be a common sight at that time, especially for a device connected to a TV), but this cannot be simply divided down to 1 MHz by powers of 2, so this still is some sort of a mystery. There are circuit diagrams out there, I just did not have seen it yet. In the end, it does not matter that much on a device intended for teaching you how to program.

Franz Morat KG tv-computersystem 6800

January 3, 2024

Recently, there was a computer on ebay.de which I never heard about before which turned out to be the quirkiest, most interesting and innovative model I learned about in a long time. No one else bid on it, so I got it for a low price. It is now the oldest computer I own. Meet the 1976 “tv-computersystem 6800” of the Franz Morat KG.

Image


No, this is neither a power supply nor measurement equipment although the very generic steel case (in a fitting 1970s orange) and the very clean front plate (one on/off switch and a thing on a cable is all you get) certainly look like that type of devic

It is made by a small German company called Franz Morat KG from the High Black Forest muncipality of Eisenbach with about 2000 inhabitants. The company specializes in mechanics, custom drives, and plastic products. The muncipality and the entire region is well-known for their watch industry (think cuckoo clock if you are into stereotypes). So, in about 1975 someone from this industrious company decides that this is the right time for a computer product.

At that time, computers were a very uncommon sight in Germany. Big companies might have some very expensive, very big computers to do all this data processing witchcraft, but home computers will not really arrive in Germany until 1978 (with the Commodore PET 2001). Engineers can amuse themselves already with computer kits, but they require at least soldering abilities. The standard interface for a small computer these days is a long row of switches and lights that show the content of a single memory unit in binary. When you want to have a bit flipped, you literally flip the corresponding switch.

But, being the crafty German engineer that you are, you decide to make a very simple, relatively cheap computer that you can produce easily in-house and that people will buy because they want to learn to program computers. Keyboards are quite expensive and cannot be produced in-house. So, what about a purely electronic input device, let’s say a light pen? What, you don’t know what a light pen is? Meet the first broadly used direct input device for “graphical” user interfaces. It consists simply of a photo diode mounted on the tip of a pen connected to the computer. When the diode sees a light it means that the electrode beam of your CRT just passed underneath the position of your pen on the screen and showed a white pixel. The computer is very aware of this position because it controls this beam in order to produce a picture on the screen. Light pens are used since the 1950s and were common on the graphics terminals of the 1960s. Using a light pen also means using a CRT display. Hmm, that’s a cost issue. Computer monitors are very expensive (at that time). Well, there is one CRT display already 93% of all households in Germany own in 1975: a TV. So, lets produce an antenna signal for a TV to display.

Now, using a TV to show the output of a computer in 1975 or 1976 is a big innovation. From 1977 onwards, most home computers use a TV in order to save cost. But in 1976, only games consoles connect to TVs and until November 1976 when the Fairchild Channel F hits the (US) market, game consoles do not even have microprocessors. I’m sure there is a computer model out there that connected to a TV out-of-the-box before the tv-computersystem, but I could not find it. So, I claim that this model is the first and wait for the angry comments to prove me wrong 🙂

The resulting computer therefore has only 3 leads: One light pen, one power cable, one TV antenna cable, all permanently connected to the computer so you cannot lose them. Inside the box, the architecture is quite simple: you have a microprocessor, some small RAM (1kB), a 0.5 kB ROM (the initial brochure notes it down as the OG Intel 1702, the first available EPROM), a simple power supply, and a lot of discrete logic. Everything is housed on 5 PCBs, 4 of them connected at the edges with what would be nowadays would be edge connectors and ribbon cables (or maybe a backplane), but what is actually a long line of single wires connecting each PCB in a bus manner.

Image


The PCBs slot in slits of simple pieces of plastic on the bottom and on the top: voilà the finished computer, at least the basic version. The basic version would have cost you 1294 DM, roughly a 1/10 of the price of the (then new) VW Golf GTI. If you spent 650 DM more, you could buy the luxury “C” version of the tv-computersystem that gave you a cassette interface and and external 37-pin bus interface.

To set this price in relation: in the same issue of a German electronics magazine that announces the tv-computer another article describes the private build of another 6800-based computer with a TV interface with less than 1kB of RAM but a keyboard imported from the US for a component cost total of 2000 DM.

Let’s talk a little bit about the logical architecture of this computer. What might astonish you (it certainly did astonish me) is that the basic model does not have any ROM whatsoever. None. Ok, you might say, where is the firmware? Is it a “clean computer” like some Sharp models that even has to load the firmware from a cassette? The answer is no: there is no firmware (at least in the basic model). All basic functionality of the computer is done in hardware. The user interface? Done in hardware. The light pen management? Done in hardware (actually there is an article in a German electronic magazine aptly named “Elektronik” from September 1975 that describes this latter circuit and it is probably written by the man who designed at least the circuit for the tv-computersystem). The microprocessor is not so much an enabler in this computer, it is the “system-under-test”, or, more correctly, it is, together with the RAM, the model of a computer system the device exposes so the user can learns its ways. From this point of view, the choice of the concrete CPU model is not really important, and now it does not surprise us that the manufacturer intended to offer different CPUs for it.

Now you might say, but, wait, I just saw you talking about a 0.5 kB ROM, what is all this no-firmware nonsense? Ah, you see, this ROM exclusively contains the library for using the cassette interface in the extended “C” model. And it is only available as a library. When you want to store to or read from a cassette tape, you need to write a small program calling the corresponding library subroutine and start it.

Ok, enough about the architecture. Let’s talk about the user interface (and remember, we are still talking about hardware). Here it is:

Image


That’s all you get. The UI (basically a classical monitor) allows you to see the content of the RAM and to control the CPU, all of which happens by using the light pen. On the left side on the majority of the screen you can see 32 bytes of memory either as hexadecimal numbers or as bits. Each of these values has to fit into a 16×16 grid of cells. Each cell has a dot at the top left corner which is the target for the light pen if this cell is to be selected by the user. Next to the dot is the space for a single character, e.g. an hex digit. The rest of the controls can be seen on the right: a column with the 16 hex digits (each one in the same cell structure as in the values table) and a second column with the controls, also in cell structure. The controls column allows to scroll the values table through the memory, to set the value of one of the values table to a certain value, to fire an interrupt, to start, stop, and continue the execution of a program, and to single-step through a program.

Let me iterate once again that all the things that you see are generated by circuits without the help of the microprocessor. There is no bitmapped graphics and no text memory. When the content of a RAM page changes and if this page is the one you have selected for display, you will see this update very quickly.

Because the UI is hard-wired, you can also not change it. There is no command to display something on the screen. When your program wants to output something, you change the value in a memory cell and hope that the user is currently watching this page (or, less sensationally formulated, for any program you tell the user on which page he/she can see the results. This is done e.g. for the example programs in the [Book].

So, does this mean that a program cannot output anything graphical? Well, you can, in a way. Remember that I told you that the UI can display values either as hex digits or in binary (and that I, sneakily, showed you only a picture of the hex mode)? Here is a picture of the binary mode:

Image
Binary Ui (Ignore the box on the bottom right corner)

As you can see, now the thing with the 16×16 value matrix makes sense. In hex mode, you need only four cells per value (because a 16-bit value can be expressed as four hex digits), but in binary, you need 16, and the entire cell space is displayed either black (being a 0) or white (being 1). You could argue that this is some sort of bitmap graphics with the sensational resolution of 16×16 pixels and two colors and that you have even 32 user-selectable pages of these graphics (and that you have a text mode with the small restriction that is can only show 10 digits and 6 (big caps) characters in a 4×16 grid. But it would be a bitmap and text mode where the output memory is also the code memory, and it is a bitmap where the user can also change the bitmap interactively. In every case, in my opinion, it is ingenious and certainly saves a lot of components that do not need to be designed for the computer and not being paid for by the buyer.

Because of all of that, programming on the tv-computer is rather bleak. Of course, there is no higher programming language, not even a symbolic assembler. You have to convert your assembler program (e.g. by using the provided paper programming forms) into numbers yourself which you then put into the memory using the UI. When you were ready, you select “run” on the screen and hope for the best. One advantage of the used Motorola 6800 processor is that it dumps its registers into memory after each interrupt, so in single-step mode, you can directly see the results of your assembler commands. The other aspect of being bleak is that you really get a deep insight into the 6800 machine code, that the computer is (very relatively) cheap, and that more comfort at that time comes with a hefty price tag, and many computer users are still accustomed to using computers with switches and lights. The Altair 8800 computer was released only a year prior and so was Microsoft Basic.

The tv-computersystem came with a German 250-page manual ([Book]) that describes usage of the computer and 6800 machine code in detail and contains many example programs. BTW, I scanned a copy:

Models

It is difficult to really tell what models really exist. The brochure talks about 6800-, 8080-, and SC/MP-based versions, but I never saw anything else than the 6800-based one. The brochure talks about two models, the “G” one (G = Grundausführung = basic model) and the “C” one. The “C” one contains a DIN (audio) cassette interface and an external, 37-pin extension bus whereas the “G” version does not have these luxuries. The pricelist also refers to 8080 “G” and “C” versions (having the same price as their 6800 counterparts). In reality, I saw two versions. My version is an orange, steel case “C” version with a single (power) switch on the front. The homecomputermuseum.de model has a nice wood case, and an additional switch at the front (marked “RAM”) and seems to be a “G” version where the PCBs on the inside are fixed by a single screw, which does not exist in my model.

Image
Source: Boris Jakubaschk, Link: https://www.homecomputermuseum.de/sammlung/detailansicht/comp/Computer/show/tv-computersystem-6800/

I believe the tv-computersystem was sold really only in Germany, but in the references you can find an one-page ad for it in the UK-based magazine Electronics Today from 1977 from a company called Rotex which might have been located in the Netherlands.

ICs inside my tv-computersystem

The CPU is of course an original gold-capped Motorola M6800. The RAM consists of 8 2102-1PC 1 KB x 1 bit chips of different manufacturers. There is one MMI 6341-1 PROM (512 bytes) in a socket next to another, empty socket. There are 2 MMI 6331-1J chips which seems to be NiCR PROMs with the gigantic capacity of 32 x 8 bit each, so 64 bytes in total. The rest seems to be discrete ICs. No other Motorola 68xx family chips can be found. The 64 bytes of PROM solve a problem I was wondering about for some time. If your base, “G”, model has no ROM how do you produce the hex digits on the screen? The digits are in a 5×5 pixel matrix and you need 16 of these digits. This means you need to represent at least 25 * 16 bits or 50 bytes. This fits very well into a 64 bytes space…

From a collectors point of view, this computer, in my opinion, is a dream. It is an early, quirky computer with innovative, unusual features that is very, very rare, even in Germany. It is a self-contained unit where every significant cable is fixed to the unit and cannot get lost. It can be connected to a TV. It is virtually unknown in Germany, let alone any other country. It has extensive documentation (even if it is completely in German). As it is so unknown, the prices are (currently) quite low.

Technical Data

Manufacturer: Franz Morat KG
Model: tv-computersystem 6800
CPU: Motorola 6800@maybe 1MHz
RAM: 1 KB
ROM: 64 bytes + 0.5 KB (“C” model only)
Resolution: 16 x 16
Colors: monochrome
Interfaces: power cable, tv antenna cable, lightpen cable, 37-pin extension bus, cassette interface (DIN 5 pin)
Size: 255mm x 115 mm x 190mm
Weight: ~3.5 Kg
Initial Price: 1295 DM (“G”) or 1954 DM (“C”)
Introduced in: 1976
Produced devices: 2000 (estimated)

References:

Entex MAC

March 4, 2023

In this blog I try to talk only about computers with the understanding that computers are devices a user can execute programs on that haven’t been contained on this computer before (I am aware that this definition of a computer helps only to distinguish between, let’s say, dedicated word processors and computers in our sense. Using this definition, an Amstrad PCW is a computer, a Magnavox VideoWriter is not).

But there are some devices that stretch that definition, I must admit. One of these devices is the very cool, but barely-a-computer Entex M.A.C. that I want to explore in this post.

Image


Entex Industries was an US-American toy and electronic game manufacturer that existed from 1970 to the early 1980s. Their toy lineup included a Lego-like system called Loc Blocks and model kits. As its high times it wasn’t a too small company either with sales exceeding $100 million in 1980. The “Entex” company name derived from NTX, which were the initials of two of the company’s founders. The company logo was an Royal Air Force bullseye with a smiling face overlayed on it.

The electronic games Entex produced (mainly handheld and tabletop electronic games, often mimicking arcade games) were adressed at the more high-end market. The three highlight products for me from these products are the Bike Computer (not a game, but indeed a computer for your bike if you don’t mind to have a medium sized tabletop calculator sized thing on your bike handle), the Adventure Vision tabletop console (exceedingly rare, especially in working condition), and the “Multi-functional Advanced Computer” or MAC Mini Computer (and no, Apple could not sue them, Entex had this name much earlier and was probably bust anyway by the time the Macintosh came around).

Image

The package promises “The First Fun Home Computer for Kids and Adults with Game, Music, Math and Programming Capability”. It reveals a slick, red plastic device with a calculator-like keyboard, a display with a 4×4 (full-size) LED matrix, and a (VFD) 8-digital numeric display beyond. The right side of the case contains a series of small black cards with holes and some description on it.

But before we dive into the software aspects, let’s look at what makes this device interesting to me: the use of 2 4-bit chips of the family of the world’s first microcontrollers: The Texas Instruments 1000 series. This series of 4-bit chips were first used in 1972, being beaten by Intel’s 4004 CPU chips by only a short time (the 4004s were no microcontrollers, but required quite a number of chips for a complete system). The 1000 series was first used in TI’s own calculators before they were sold to everyone from 1974 on. The 1000 series contained ROM, RAM, counters, timers, and I/O interfaces and was used in a plethora of toys and calculators.

The 1000 series models Entex used in the MAC was one TMS 1600 with 4000 bytes of ROM, and 64 bytes of RAM (yes, bytes, not kilobytes), and a TMS 1170 with 2000 bytes of ROM, and another 64 bytes of RAM, and the interface to run a VFD display (therefore, the 1600 in the MAC dealt with making music and the programming, and the 1170 did the calculator and the games). The ROMs are mask-programmed, this means that the ROM content is added during manufacturing of the chips and cannot be changed later on. This makes for smaller production cost, but also maximum inflexibility, because if you want to change the software in the ROMS, you have to produce new controller chips. For good measure the two microcontrollers are complemented by a TMS 1024 “Interface Expander” that deals with addressing the 4×4 LED matrix.

Let’s now look at what the MAC came with and what it could do.

The Manual

Normally, we do not talk much about manuals because they either state the obvious or are, in the best case, boring but useful references. For a toy, the expectations are normally even lower.

However, this isn’t a suspicious leaflet by an Asian manufacturer translated by the half-blind daughter of someone who used other Asians manuals to learn English, but a very decent 50-page (at least in German) brochure by a knowledgeable expert in the field, probably the engineer who designed the entire device. Therefore, the MAC manual surprises with accurate, detailed information, giving the big picture of how computers change the world (at the time), and concrete assistance for the user. This is partly really needed because there is so much non-obvious functionality, and partly it is way more information the user really needs because he/she cannot use it anyway.

The “punch cards”

As I wrote above, the MAC comes with a series of black cards that ressemble a little bit computer punch cards, and one asks oneself how the MAC reads these holes. The disappointing answer is, it doesn’t. The cards can be inserted in a slot on the top of the display in order to hide some of the LEDs for some applications. If you play the Tic-Tac-Toe game for example, it reduces the 4×4 matrix to the 3×3 one needed for the game and for some applications it also adds some inscriptions to the LEDs so you know which button to press to address this LED position.

The Calculator Mode

The calculator is a very basic run-of-the-mill 4 species calculator with one memory slot and 8 digits (my machine displays 00000000 when the number overflows, this seems strange to me). Together with the VFD display and the calculator keyboard, this functionality feels very much like the spritual home of the hardware. 10 of the 50 pages of the manual explain the calculator using many examples.

The “Music” Mode (a.k.a Beep Machine)

You have two modes for making music (or noise): piano and organ. The tone does not differ in both modes, the only difference is that in organ mode the tone is held as long as you press a key. You can either play “live” or record and play back songs. You have two octaves of tones, but you can play only monophonically. In record mode you can also access the half tones. In this mode you can even vary the length of tones and add pauses of different length. I don’t know much about music, but what I know is that the tones are played VERY LOUDLY.

The Games and Applications

There are 5 games and applications contained in the MAC. There is a 2-player Tic-Tac-Toe game where (inexplicably) the users have to decide themselves whether someone won the game, a game called Tactics that I do not really understand, a game called Concentration that I do not really understand, a tool that can calculate the local time in a number of cities given an entered time and time zone code, and finally some sort of slot machine (which I do not really understand). All in all, nothing exciting.

Programming Mode

The other thing that really interested me was the programming mode. You can have up to 55 commands and the P1 key starts a program and halts it (and then you can re-start it again by P1). When a program reaches the end, it simply starts over again. I was really baffled though when I learned the commands. You have commands to play tones, to switch on the LEDs for an amount of time, and to switch off all lights. That’s it. That’s not exciting at all. Err, wait, and then there is an example program in the manual that gives you a digital die. How can you program a random die with only these commands??

This is the place where you could try to figure it out yourself. I have given you all the information you need to know. It took me a while to understand it.

Ready? Ok, the randomness comes from the user, i.e. the point in the program execution when he or she presses P1. The program consists simply of the sequence of all numbers on the die that are displayed on the LEDs one after another, and the one that you roll is the one that is currently displayed when you press P1. As the computer runs through the program so fast you can neither distinguish the numbers on the LED matrix nor react fast enough, you cannot control which number the program stops in. Genius!

Of course the lack of any flow control commands means that all your programs have to fit into this scheme or they are programs that need to be always the same. You have a metronome program as an example and you can play melodies or have pretty patterns using the LEDs.

My MAC

Some of the Entex electronic products were also available under different names in some countries. The MAC was also sold under the name “Multifunktions-Spiel-Computer” by the giant German mailorder company “Quelle” (their main catalog was over a 1000 pages and was printed in 8 million copies). This Quelle version might have been sold the most because nowadays this is the version that seems to be available more often.

I have the Quelle version, which means a (sometimes not very knowledgable) translated German manual and German cards (and because the German card descriptions would need more space, most of the cards instead just refer to the corresponding manual page).

Image


The battery compartment of my MAC is so mucky that I did not want to clean it. Therefore, I can report that the MAC runs just fine on a 6V universal power supply with a 2.5mm plug (plus on the outside, minus on the inside). The MAC package says the device needs 250 mA.

Image


When opening the MAC and looking at the PCB (and you need to unscrew basically everything for that, undoing some 15 screws, including two tiny ones holding the on/off siwtch), I found that two of the TI chips have Japanese-style markings on them indicating that maybe at least the PCB was soldered in Japan (see Photo).

Also, as expected by the use of the microcontrollers, there is not much else going on on the PCB component-wise, some resistors and capacitors, a few transistors, the three chips, the LCD, the VFD, and that’s it. This is, of course, very appropriate for a toy.

Conclusion

The Entex MAC is an early toy computer with interesting hardware and an attractive esthetics. It has an unexpectedly good manual, but is limited in its capabilities due to the restrictions of what could be delivered at the time in a toy. It cost $80 in 1981 at a time when e.g. a 16kB 8080A-based Interact “R” home computer was advertised for $249 (because the remaining inventory of the closed original manufacturer was sold off).

From a collector’s point of view the Entex MAC is a relatively rare thing, but maybe not many people want to have one because it falls a little bit in the void between a game and a computer.

Technical Data

1 TMS 1600 N2LL Microcontroller
1 TMS 1170 NLHL Microcontroller
1 TMS 1024 NLL Input/Output Expander
with together 6000 Bytes Masked-Programmed ROM
and 128 Bytes RAM
4×4 red LEDs
8-digit VFD numeric display
27-key keyboard
Uses 4 C batteries or a 6V, 250 mA power supply with a 2.5 mm plug, plus on the inside
Initial price: $80 ($260 in today’s money)

References

https://www.handheldmuseum.com/Entex/MAC.htm
https://www.handheldmuseum.com/Entex/index.html
http://www.computersammler.de/der-erste-mac-mini/
https://www.handheldmuseum.com/Entex/BikeComputer.htm
https://www.handheldmuseum.com/Entex/AV.html

The Fujitsu Micro 8 and its Bubble Memory Drive

January 7, 2022
Image
Fujitsu Micro 8

At the VCFB 2021 in Berlin I presented a recent purchase of mine, a Fujitsu Micro 8 computer complete with its Bubble Memory drive option.

If you do not want to the read following post, you can simply watch the video.

Ok, so obviously, you want to read about it. Here we go!

Let’s start with the manufacturer. Fujitsu is is a Japanese multinational information and communications technology equipment and services corporation which was established on June 20, 1935. This makes it one of the oldest operating IT companies after IBM and before HP. Fujitsu manufactured in 1954 Japan’s first computer, the FACOM 100 mainframe, and launched in 1961 its second generation computers (transistorized), the FACOM 222 mainframe. Fujitsu was a major chip manufacturer until 2014 when they spun off their chip manufacturing business. Think of it as a Japanese IBM if you do not know the Japanese market.

At the beginning of the 1980s, personal computers become a hot topic, and Fujitsu decides to enter this market (at this stage they (as other Japanese manufacturers) have alread released their first micro computer, the LKIT-8 kit). In contrast to IBM, Fujitsu wants their first real entry to the personal computer market to be at the very front of the state of the art. Also, they aim both at private owners and at small businesses. So, the list of features of this model, the Fujitsu Micro 8, released in May of 1981 is very impressive:

  • 2 x 6809 CPUs (one Fujitsu 68A09 processor at 1.2 Mhz as the main CPU and one Motorola 6809 @ 1 Mhz as the graphics processor)
  • 64 kB RAM (first usage of 64 kbit DRAM chips in a microcomputer)
  • 48 kB VRAM
  • 44 kB ROM
  • 2 x Bubble Memory drive as an option (first such drive in a microcomputer)
  • graphics with maximally 640×200 pixels in 8 colors
  • many interfaces: cassette, digital RGB, Composite, TV, cartridge, expansion, RS232, printer, analog in

Unfortunately, this list of features does not only wow the consumer, it also makes him/her notably poorer as it requires Fujitsu to ask for a price for the basic model of 218 kYen then which is about 290 kYen today which corresponds to $1982 or 1690 EUR today. This price does not include the many available options as we will see below.

Internal Options:

MB22003 Kanji ROM (2965 JIS Level 1 Kanjis), 16×16 pixel 30‘000 Yen
MB22002 Non-Kanji (figures, symbols) ROM 10‘000 Yen
MB22401 Z80 card 11‘700 Yen
MB28011 8088 card (set with CP/M-86)
MB22601 Bubble holder unit 85‘700 Yen
FBM43CP Bubble cassette (32KB) 35‘000 Yen

External Options:

MB27603 Standard floppy disk drive (8 “)
MB27605 / MB27607 Mini Floppy Disk Drive (5.25 “)
MB27606 / MB27608 Thin mini floppy disk drive for expansion (5.25 “)
MB22605 128KB Bubble Holder Unit
MB27301 12 inch high resolution color CRT 188‘000 Yen
MB26001 System Expansion Unit
MB22603 FDD interface Card 17‘000 Yen
MB27601 5 inch FDD 313‘000 Yen

Operating Systems Options:

SM07217-F023 F-BASIC Ver. 2.0 (So-called DISK BASIC, but it works in RAM without using the main ROM)
SM07217-F061 F-BASIC Ver. 2.2 (128KB bubble cassette)
SM07217-M041 OS-9 / 6809 Level 1
SM07217-M021 FLEX
SM07217-M033 UCSD Pascal
SM07217-M011 CP/M-80 (with Z80 card installed)
CP/M-86 (when 8088 card is installed)

One option we might have to explain non-Japanese readers is the “Kanji ROM” option. What is it and why was it so expensive?

You have to understand that written Japanese consists out of three different character sets. Two of them, Hiragana, and Katakana use relatively simple signs and consists of about only 50 characters each. They can be represented easily using a “normal” 8×8 pixel grid, and they do not require much more memory than Latin characters (also, they do not have upper and lower case). The problem is the third character set, the Kanjis. These are basically Chinese characters (in principle, the reality is more complicated). These are both graphically complex and there are many. They are so complex that a normal 8×8 grid cannot be used because you could not see the differences between the characters. There are also so many that they do not fit into a standard amount of ROM. Therefore, the Kanji option switches the character grid to 16×16 pixels (and quarters the amount of characters on screen), and it provides enough ROM space for the additional almost 3000 different(!) characters. To that end it is delivered in 128 kB consisting of 16(!) additional ROMs for which there are also sockets on the PCB.

Image

My Micro 8 came with the Bubble Memory drive option, the internal Z80 card and the Kanji option. It likes to throw its 1.5A fuse when it gets more than 100V, e.g. 110V…

Let us conclude the view on the core model with a note on its two CPUs. It sounds very promising to have a computer model with two of these components. Does this make the machine twice as fast? Well, no. See, as I mentioned, the 2nd CPU was the chip responsible for the graphics, and this model does not have another dedicated graphics chip. Also, it acts as a complete independent system. It has its own boot ROM, its own memory, and runs its own program. It communicates with the main CPU exclusively using a 128 byte memory area. So the main CPU writes into this shared area (while the graphics CPU is stopped), and afterwards the graphics CPU is re-started. This is not a very efficient way to communicate, and in the beginning it was almost impossible to write graphics intensive programs. Fortunately, one of Fujitsu’s BIOS programmers left several hidden commands in the BIOS. These commands permit to transfer large binary programs to a reserved memory space and JMP. Only after discovering this hacking technique, this model had enough graphics performance for fast games…

Now, the one option that makes the Micro 8 quite unique is the Bubble Memory drive (there are only two other (and later) computer models out there that use Bubble Memory as a standard). So what is Bubble Memory and why did you possibly never have heard about that?

Bubble Memory is a type of non-volatile computer memory that uses a thin film of a magnetic material to hold small magnetized areas, known as bubbles, each storing one bit of data. Non-volatile means it keeps its data even if the power is switched off, like e.g. Flash memory or a magnetic tape. Bubble memory started out as a promising technology in the 1970s, offering memory density of an order similar to hard drives but performance more comparable to core memory while lacking any moving parts. This led many to consider it a contender for a “universal memory” that could be used for all storage needs (i.e. main memory like RAM or mass memory like a hard drive).

Because of this appeal, by the mid-1970s, practically every large electronics company had teams working on bubble memory. Texas Instruments introduced the first commercial product that incorporated bubble memory in 1977. By the late 1970s several products were on the market, and Intel (later MemTech) released their own 1-megabit version, the 7110. Other companies that produced such chips were Sharp, Hitachi, Rockwell, and Fujitsu. This technology was even produced in the USSR.

In the first half of the 1980, companies offered Bubble Memory for computers, either as an option (here we have e.g. our Micro 8 in 1981, the “MPC BUBDISK” card for the Apple II in 1982 (128 kB for $895), or HP’s 9825A card in 1984 (128 kB for $1900)), or as built-in mass memory in portable computers (like the 1982 GRiD Compass or the 1983 Sharp PC-5000). Another use was Konami’s Bubble System arcade video game system, introduced in 1984. It featured interchangeable bubble memory cartridges on a 68000-based board.

Bubble Memory works as advertised, even if it requires a “warm-up” time before it can be used because it can be operated only if it is at least 30 °C warm. This initial warm-up time can be up to the 85 seconds the Konami system needed after a cold start (and this time “cold” is not meant figuratively).

So, why don’t we used Bubble Memory today? There are two main reasons. First, although Bubble Memory is as fast as core memory, it is much slower than (chip) RAM (which came later). Second, chip RAM and hard disks became cheaper than Bubble Memory quite soon. As a consequence, in 1981 major companies working on the technology closed their bubble memory operations.

Bubble Memory was used also after 1981 due to the only technical advantage it had left: it can store data even without power and it is immune to even severe movements because it does not have moving parts. It was the Flash memory of its day (Flash was not on the market before 1987) and was therefore used in mobile applications until the entire technology disappeared entirely by the late 1980s.

How does Bubble Memory work?

As I mentioned before, the bubbles, small magnetized areas exist on a thin film of magnetic material. This material is arranged to form a series of parallel tracks that the bubbles can move along under an external magnetic field. The bubbles are read by moving them to the edge of the material where they can be read by a conventional magnetic pickup, and then rewritten on the far edge to keep the memory cycling through the material. There is a very instructive animation(?) of such bubble moving through some material which can help imaging you this mechanism:

Taken from https://www.gsalmasi.com/newpage/bubbles.html

What does a Bubble Memory chip consist of?

Image
Taken from a TI report entitled TIB0203 Magnetic-Bubble Memory and Associated Circuits from November 1978

See the above picture. First, you have the die with the magnetic area itself. It is wrapped in two layers of orthogonal coils. This package is sandwiched between two permanent magnets, and the entire package is housed in some tin case. For a real-life “unwrapping” of a Bubble Memory chip I can recommend Wolfgang Robels corresponding web page (http://www.wolfgangrobel.de/museum/bubble3.htm) which has a series of very nice photos.

Let’s come back to the Bubble Memory drive in the Micro 8.

Image
Micro 8 Bubble Memory Drive

The optional Bubble Memory drive inside the Micro 8 has two slots and can therefore use up to two Fujtsu cartridges. These cartridges have a capacity of 32 kB each (later on, 128 kB cartridges were available, but I am not sure whether they can be used in this drive or only in the later, external drive). The cost of the option and the cartridges is enormous: the drive costs 85 kYen, the cartridges 35 kYen each. This made the configuration above cost 155 kYen (or $1400 today)…

As I said at the beginning, the Micro 8 was a very expensive model and it did not sell well. As a reaction, Fujitsu splits the Micro 8 only a year later into the FM-7 for the mass market at a much lower price (and without a Bubble Memory option) and the FM-11 which is aimed at businesses. The FM-7 becomes a very popular model in Japan and is of the “Popular 3” personal computers of that era.

From a collector point of view, currently you can buy Micro 8s at a not too high price in Japan, but one with the Bubble Memory option is very rare and is typically very pricey. In every case, they are quite cool machines.

IPC MagicWriter

December 29, 2021
Image

Recently, I had the rare joy to open the still sealed box of a PDA computer. Normally, one would think twice about opening a still sealed box, but this is a model that is so rare that there is almost no demand for it. I got it for 20€, postage included.

The manufacturer’s name on the box is IPC, a very well-known (in fact, the largest) Singapore computer manufacturer at its day. During the Asian financial crisis at the end of the 1990s its computer ventures dwindled and IPC nowadays is into property investing.

The model name is the “MagicWriter”. Does not ring a bell? Don’t worry, it did neither for me nor probably for any other person outside the handful of people that were involved in its development.

So, what is the IPC MagicWriter?

As I mentioned, the MagicWriter is a PDA or Personal Digital Assistant, a moniker for a relatively small, light-weight pen-based mobile computer (typically without keyboard) that usually included applications such as a notebook, an address book, a calendar, and the like. One can distinguish between a dedicated PDA which can exclusively execute the contained applications and a PDA computer, which also offers a built-in programming language or at least some way to execute programs that can be loaded into the computer. Unfortunately, the MagicWriter seems to be a dedicated PDA (because I like the computer variety more).

The MagicWriter was sold from 1994 or 1995. It was designed by a Singapore company, Imagique Computer Design Pte Ltd. It was manufactured by IPC. I read a rumor that can be read that 10,000 devices were sold for deaf and mute people in Japan, but I cannot confirm that.

Another source mentions that Com 1 (French company) designed a PCMCIA GSM/telephone/fax card (maybe also the software?) for the MagicWriter, and a corresponding hint to the existence of such a card can be found also in the manual.

The MagicWriter came with some applications bundled with it: a File Explorer, a Database, a Spreadsheet/Calculator, a Calendar, a Schedule, a Memo, and a Phonebook/Address Book. You could also search in the data. Due to its nature of being a pen-based PDA without a keyboard, the MagicWriter has some basic handwriting recognition in the sense that you can enter single characters in some pre-defined boxes when input is required. A virtual keyboard could also be used.

Quite a mystery for me is the Operating System of the MagicWriter. As we will see later on, the PDA is basically a mobile PC hardware-wise. It has a PC chipset, a (Phoenix) BIOS, 1 MB of RAM and 2 MB of (Flash) mass memory. Still, neither the package nor the manual (or any other source for that matter) mentions an Operating System, PC-wise or otherwise. There is also no GUI on the screen, just a sea of text. The functionality of a dedicated PDA also does not require an explicit Operating System, and the PC BIOS supplies already a base layer of I/O and other functions. Maybe the applications directly sit on the BIOS.

The sealed MagicWriter package contained:

  • the PDA itself
  • 3 AA batteries (partially spilled out, but as they were still sealed separately, no harm was done)
  • a (dumb) pen
  • a faux leather 6-hole ring binder with some paper in it where the PDA could fit in
  • a PC 3.5″ floppy disk titled “Windows Application”
  • a 40-page, loose-leaf manual to be inserted into the ring binder (but without holes)

Now, let’s have a look on the inside of the MagicWriter:

Image
PCB

The PCB seems to be quite clean at first, but after a closer look one can find a major revision done to it (using even an own small PCB, hand-connected to the major PCB components). This seems to indicate that a) there was a major problem to the PCB revision and b) that the device was priced high enough that just tossing the original device into the bin was not an option.

The used CPU is a 1994 Chips & Technologies F8680A SoC which unites the CPU, RAM management, power management, and CGA video. It is said to be 80286 compatible through emulation on the chip which is a enhanced 80186 clone. It runs on 14 MHz. It was used in all sorts of subnotebook and embedded computers. There is also a predecessor, the 1991 8680 running on 10 MHz. It was used in the 1992 Gateway 2000 “HandBook” DOS (but not Windows) subnotebook.

Major chips found on the PCB

  • C&T F8680A (CPU SoC)
  • C&T F87000 (Multi-Mode Peripherial Chip, handles PCMCIA cards) x2
  • Toshiba TC518512FTL-80LV (PSRAM) x2
  • UM62256DM-70LL (128kB SRAM)
  • Intel E28F008SA-120 (1 MB Flash) x2, one having a
    • PhoenixPICO BIOS PRODUCT and one having a
    • Phoenix PCM+ sticker

As I said, the box was in the original shrink wrap, and the content of the box clearly never taken out. Still, when took the PDA out, there were some loose screws rattling in it (which I tossed out before switching it on). Unfortunatly, I could not convince the thing to switch on, neither by using batteries, nor by attaching a power supply to it. I did not find any popped condensators nor batteries inside, so either the thing did never work, or it deteriorated in an silent way over time.

Why was the MagicWriter not succesful? We do not know the initial price or whether there was a major flaw preventing the usage of this device, but let’s assume it did work and that the price was not too high. What chance would such a device have had in 1995? Large, tablet-sized pen computers in numbers existed since 1991 (NCR 3125). They were heavy, and they were pricey, and they were something for the professional user. In 1993 some mid-sized and quite portable pen-based machines were released: the Apple Newton, and the Tandy Z-PDA. The Newton promises recognition of cursive handwriting, but basically falls short of this promise with the first models. In 1994 General Magic and its conglomerate of Sony, Motorola, and other heavy-weights give the PDA idea a new twist by putting the focus on communication, allowing some of their devices to even communicate wirelessly. All these devices have some sort of GUI, the General Magic devices even a very graphic one. From 1996 USR will revolutionize the PDA market by making the devices cheap, very lightweight and small. Pocket-sized DOS PCs (even if the did not have a pen) exist since the Atari Portfolio in 1989. So, in 1995 bringing out a text-based PDA that is not even DOS-compatible seems like a recipe for failure (and it probably was).

From a collector’s point of view, dedicated PDAs are not very interesting, and there are many makes. This model, however is at least very rare (my MagicWriter seems to have the serial number 461), and it uses a quite rare, interesting chip set. The Operating System is a mystery. Maybe there is a way to execute an arbitrary program.

Technical Data

CPU: Chips&Technologies F8680A@14MHz (80286-compatible)
RAM: (PSRAM) 1 MB
Mass Memory: 2 MB Flash
OS: unknown
Interfaces: 2 x PCMCIA type II slots, RS232 (proprietary connector)
Batteries: 3 x AA Alkaline
Size: 210x135mm
Weight: 455g
Released in: 1995
Number of produced machines: unknown
Initial price: unknown

References