Epson EHT-10

October 26, 2020

Epson created the first “true laptop computer” with the HX-20. It was released in July 1982 and weighed 1.6 Kg. The successor to the HX-20 was in 1984 the 2.3 Kg PX-8 which was now capable to run CP/M due to its Z80-compatible CPU. The next model, the 1985 PX-4 combined features from the PX-8 and the HX-20, was also Z80-based and offered CP/M. I could not find the weight of the PX-4, but it was probably in the range of the HX-20.

The HX-20, the PX-8, and the PX-4 were all laptop computers, i.e. they required a desk or at least a lap to sit on so you can use it with your hands. But there are usage scenarios that require a more mobile device, i.e. one that you can hold in one hand while the other one uses the device. It is technically possible to use e.g. a HX-20 for this purpose, but it’s not comfortable at all.

So if you are a device manufacturer like Epson and want to serve the market that needs that higher mobility, what do you do? You design a version of the PX-4 that is lighter and that you can hold in one hand. Voila the Epson Handy Terminal (EHT)-10 family from 1986 onwards.


The (original?) model, the EHT-10 is a (from a today’s point of view) massive beige brick with an on/off switch at the top and an enormous 7-inch black-and-white display. It is mainly a touchscreen that you use with your finger, no pen required.

The basic usage scenario of the EHT-10s is mobile data entry. But you do not need to develop your application on the EHT-10, in fact, you can’t. You develop it instead on a desktop CP/M machine, and then simply transfer it to the EHT-10. To that end the EHT-10s are equipped with both CP/M 2.2 and the basic runtime environment that allows to run Epson Basic programs (e.g. coming from a PX-4). All you have to take into account are the different display sizes and input possibilities. This does not mean that you can get a CP/M shell on the EHT-10 or an editor to enter Basic programs; the EHT-10 is purely an execution device. You can transfer your programs to an EHT-10 either by serial connection, or by “IC Card”, or burn it on an EPROM and stick it into the conveniently located socket next to the main battery. So, you cannot program on the EHT-10, but you can connect the EHT-10 to your development computer by using a “Development Cartridge” for debugging purposes.

The main battery of the device is a proprietary 4.8V, 700 mAh NiCd module which gives the device a runtime of 8h. In case you have to change the battery, there is a secondary RAM backup battery (4.8V, 45 mAh). Using the stock power supply it takes 10h to load the main battery and a whopping 45h to load the secondary one it was completely empty.

On the sides of the case, at the bottom, there are two massive buttons to which you could strap a ribbon, so the device could go around your neck, thus freeing also the other hand if needed.


The other two models of the family are, first, the EHT-10/2b which looks like an oversized calculator with its 4-line display and its 34 keys (there is even a CALC key which puts the device in basic calculator mode). Apart from that, it’s innards are the same as the EHT-10’s.
The second model is the EHT-10/2 which is a EHT-10/2b without a backlight.

I received my two EHT-10/2bs (plus a printer) together with a simple 5.6V, 0.4A travel power adapter with a connector crudely welded to it. I have a 64 kB and a 128 kB RAM model. Using this power adapter I could never get the 128 kB model to run for more than a few seconds even connected to the power adapter, but the 64 kB model worked fine. And if you think about it, this makes complete sense. There was no main battery in any model (because they would be quite proprietary NiCd modules), but the backup cells were still in there. No, if you load the backup cells, the (battery-buffered) RAM immediately run on the cells, and even if they are full, the power needed by 128 kB is seemingly larger what the backup battery can deliver when switched on. The 64 kB model need less power and can run on my power adapter. The stock adapter had 6.0V anyway.

Epson produces many printers, and there is a clip-on dot-matrix printer model that fits the top of any EHT-10 device (it even uses a tiny ribbon cassette which you find even today on ebay).

There are almost no Internet sources on the EHT-10. The only one is listed below. It contains some photos (actually showing the very devices I ended up with), some (but unfortunately not all) manuals and the development software meant for CP/M or MSDOS desktops.

Epson sold these devices from 1986, but they were not the first company to target the ultra-mobile market. The first company probably was DVW (later on Husky Computers) with their Husky line of devices from 1981. The competition to the EHT-10 at the time was probably the 1984 Husk Hunter 2 which weighed 1.15 Kg, and had to be held in landscape mode, where as the EHT-10 was below 1 Kg, 50% smaller, and could be used hanging from the neck. The CPU and OS on both the EHT-10 and the Husky Hunter 2 was very similar.

Another competitor would have been the much lighter and cheaper Psion Organizer II from 1986, but it was much more proprietary in terms of CPU and OS, had much less RAM, and was much slower than the EHT-10.

Progress cannot be stopped and Epson later on gave up the EHT-10 family for the 1991 EHT-20, EHT-30, and EHT-40, which were PC- and DOS-based devices for the same usage scenario (all looking quite similar to the original EHT-10). The final model was the 1995 EHT-400c which was a small color-screen pen-based Pen Windows 3.1 tablet.

Epson EHT devices are probably quite scarce today because they were probably never distributed over normal computer stores (but to system integrators) and they were very pricey (which they could be in the beginning because there was not much competition). As a piece of industrial equipment, though, only a few people are interested in them.



  • CPU: Z80@3.68 MHz (CMOS)
  • RAM: 64-256 kB
  • ROM: 128 kB System ROM, up to 128 kB Application ROM
  • OS: CP/M 2.2
  • Batteries: main: backup: NiCd cells
  • Interfaces: RS232C, Barcode Scanner, IC cards
  • Dimensions: 21.3 x 9.3 x 3.8cm
  • Weight: 600g without main battery
  • Display: 12 x 14 characters (5 x 14 characters out of which can be used as a touch screen), 84*154 pixels


  • same data as the EHT-10/2b, but without the backlight


  • same data as above except the
  • Display: 20 x 4 characters, 120 x 32 pixels, with backlight
  • Keyboard: 34-key-calculator keyboard

Reference (yes, this time it’s only one)

Vintage Computing Festival Berlin 2020

October 8, 2020

In contrast to some other events, the VCFB 2020 will actually take place. This weekend (depending on when you read this), on October 10th and 11th, it will take place via a Wiki, BigBlueButton, and some streaming.

Find all important information on

Most exhibitions, presentations and workshops are in German, but there are also some English things:

I will also hold a presentation (on “Collecting Computers as an hobby: An Introduction“), but it will be given in German (as some information relates to the German situation).

Also, I will moderate an event on “Technology Museums and Collectors: Question Time“, but again, in German, because these are German museums.

Siemens Nixdorf PCD-3 Psl/20

October 8, 2020

Recently, on Ebay I discovered a variant of the NCR 3125 that I did not know: the Siemens Nixdorf PCD-3 Psl/20.

I did not buy it, but from the picture it seems quite clear that this is, in fact, a NCR 3125. Also, the name seems to hint to the 3125′ 80386SL CPU at 20 MHz…

VTech Logo Computers

August 5, 2020

I am always fascinated about computers with a programming language other than Basic in ROM. You might have heard of computers with Forth (e.g. the Jupiter Ace), maybe even APL. But do you know any computers with LOGO as their first language? No? Then you probably also do not know the company with the largest lineup of LOGO-computers. Any guesses? No, it’s VTECH!

You might pretend that you don’t know VTECH, but that would be very unlikely given that they are in business since the mid-70s and that their products are still sold in a big shop near you. VTECH is a Hongkong-based electronics company and my impression of them was always that they started selling computers in the 80s, and cleverly never changed their architecture afterwards, just their target market. However, that’s not really true.

VTECH (short for Video Technology) started their business in the mid-70s with a PONG console (what else?). They followed this up with LED-based games (looking like old calculator displays), then LCD-based handheld games. In 1982, they released their own games console, the CreatiVision. We might know VTECH for their 1980s line-up of home computers (and similar devices), the Laser family of computers (i.e. the Z80-based Laser 50, 100, 200, 210, 300, 310, 350, 500, and 700, and the 6502-based Laser 128, 2001, and 3000).

Next to their “serious” home computer line, VTECH also started in the 80s to build up a line-up of educational devices targeted at kids that soon imitated the looks of serious computers, either desktop models, or, later on, notebook models which makes one think of them as toy computers, especially when a tiny, black-and-white LCD display is mounted in a normal-size notebook lid, resulting in a huge bezel. Initially, these models were also based on Z80 CPUs.

Now, lately, I watched Nostalgia Nerd’s Youtube episode on VTECH (educational, toy) Laptops (, and I was very surprised to learn that there was a model from 1998 which even had the programming language LOGO built in.

This puzzled me because there is no other computer with a built-in Logo in ROM (as far as I know). Sure, there were quite some Logo modules and interpreters that could be added to all sorts of computers, but no manufacturer chose to provide this as the initial choice and in ROM.

So, I started to explore this road a little bit. What is the CPU Logo is interpreted on? How many models did VTECH produce with Logo? How good is the interpreter? There is not much information about these aspects on the Internet (probably you can find all of it in the references section). So, here is what I could find out.

All the models with Logo were a single generation of VTECH educational computers that were produced between about 1997 and 2001. This generation was probably all based on the same or at least a similar System-on-a-chip (SoC) architecture plus a RAM chip (the Black Magic CX had 128 kB of static RAM) that was completed by a conventional masked ROM chip that had up to 2 MB of ROM memory.

Unfortunately, we do not know the exact SoC model(s) that were used for sure, but the MAME code trying to emulate some of these computers use “CR16B” emulations. CR16B is a version of National Semiconductors 16bit “CompactRisc” architecture that can address (you guessed it) up to 2 MB of code/data. We don’t know for sure because, unfortunately, the SoC is bonded directly on the (very small) PCB, with a drop of epoxy on top of it, and no helpful labels on the PCB.

Main PCB of the Black Magic CX

However, the friendly decapping nerds from Team Europe decapped the SoC of a 6600 CX (see References) and produced a picture of the chip underneath the epoxy. It looks as if the die has a label that reads “NSC1028”. They also cite a press release from 1999 saying: “National Semiconductor designed the Geode NSC1028 processor specifically for VTech’s new email appliances. The system-on-a-chip integrates a powerful 16-bit RISC processor, keyboard and printer ports, LCD display controller and speech synthesis circuitry into a single piece of silicon. This custom integrated processor represents the first step in a partnership between VTech and National to bring attractively priced, easy-to-use information appliances to the consumer market.”

BTW, it seems to me that this line of Geode-branded SoCs has nothing to do with the x86-compatible line of Geode processors based on Cyrix technology that was sold later on to AMD.

The Logo is a nameless, quite basic version of the programming language. It is rudimentarily explained on 2.5 pages in the manual of the device, and consist of 25 drawing commands (turtle graphics), 15 mathematical functions, 9 “word and list” commands (don’t forget that Logo is a descendent of Lisp), 3 logical functions, 17 “other” commands, and 11 “flow control” commands, resulting in 80 commands overall.

If we compare this to e.g. to Commodore’s C64 Logo (by Terrapin), we find 37 drawing commands, 20 mathematical operations, 16 word and list commands, 7 flow control commands, and 66 other commands, 146 commands in total. The differences lie mainly in the larger range of devices that can be addressed on the C64.

Words and lists (in the Logo sense!) must not have more than 20 characters. Apart from that it is not known how much memory is available to the Logo interpreter, or whether there are other restrictions to the language. It seems to me that the interpreter is very, very slow, at least when graphics are involved.

The VTECH devices do not only offer the Logo interpreter (and editor mode), but also three groups of sample programs, ready to be edited if needed. When using the editor, Logo programs can be stored to either the internal (battery buffered) memory or a storage cartridge if you have one.

The different VTECH computer models had partially different displays, varying in size and resolution. The basic model (e.g. the 5005 X) had only a 5.4″ display with a resolution of 130 x 50 pixels (still giving you 4 or 5 lines of text). The deluxe Black Magic CX has a whopping 7.1″ with (about) 237 x 138 pixels. Unfortunately, because VTECH needed to appear modern, the usable area for Logo programs is much smaller as there is a window border around the area, and command icons on the right and the bottom. As a result, e.g. on the Black Magic CX, you can use maximally 202 x 86 pixels under Logo.

Apart from the English version, there was also a German version of Logo that is completely localized language-wise even down to the command names. So if you have e.g. the 5005 X model which features both English and German, you get also an English and a German Logo! Apart from German I did not find another language Logo was localized to (although I could check this only for the Genio 6000 model which has Spanish and English software. The Spanisch software offer also the English Logo.
For the later, deluxe model (the Black Magic CX) the Logo part of the manual was even extended heavily. Now, you are introduced through 10 pages into Logo (plus the extended explanation of the commands), quite a nice change.

Now to the question of which models of this generation do have Logo on board. This question is a little bit hard to answer as

  • there is no comprehensive, worldwide catalogue of VTECH devices
  • different language versions of basically the same machine have different names

So it’s up to lists of machines where the Logo feature we could ascertain. Here is my current list:

Power Zone 2000

Year: 1997
Format: Desktop with separate keyboard, see-through case
Display size: large, probably 18.1 cm = 7.1 inch, 237 x 138 pixels, with background light
Languages: English

PC Endeavour

Year: 1998
Format: Notebook
Display size: small, probably 5.4 inch, 130 x 50 pixels
Languages: English

Power Zone Edge

Year: 1998
Format: Desktop with separate keyboard, see-through case
Display size: large, probably 18.1 cm = 7.1 inch, 237 x 138 pixels, with background light
Languages: English

Power Zone Plus Notebook

Year: 2001
Format: Notebook
Display size: small, probably 5.4 inch, 130 x 50 pixels
Languages: English

Power Zone Plus Desktop

Year: 2001
Format: Desktop
Display size: small, probably 5.4 inch, 130 x 50 pixels
Languages: English

Genius 5005 X

Year: ???
Format: Notebook
Display size: 13.8 cm = 5.4 inch, 130 x 50 pixels
Languages: German & English

Genio 6000

Year: ???
Format: Notebook
Display size: 13.8 cm = 5.4 inch, 130 x 50 pixels
Languages: Spanish & English

Genius 6600 CX

Year: 1999
Format: Notebook
Display size: small, probably 5.4 inch, 130 x 50 pixels
Languages: German

Genius 8008 CX

Year: 2000
Format: Notebook
Display size: large, probably 18.1 cm = 7.1 inch, 237 x 138 pixels
Languages: German & English

Genius Tabletop Black Magic CX

Year: 2000
Format: Desktop with separate keyboard, see-through case
Display size: 18.1 cm = 7.1 inch, 237 x 138 pixels, with background light
Usable pixels under Logo: 202 * 86
Languages: German
Keyboard: wireless (infrared) keyboard with connector for the mouse
RAM: 128 kB Static RAM (GM76v8128cllfw70)

In conclusion, VTECH produced a number of models of educational computers offering a built-in Logo interpreter. Given the typical volumes of VTECH products, this makes them the largest (and only) manufacturer of Logo computers. If you want to own one yourself, get a model with the large screen.

Genius 9009 CXL

Year: 2000
Format: Notebook
Display size: large, probably 18.1 cm = 7.1 inch, 237 x 138 pixels
Languages: German & English


EO 440

June 22, 2020

From time to time I am baffled that I own a computer that I do not have a blog entry about. Either I had at the time of purchase too much going on and forgot or I felt at the time that there was already enough information available in the Internet. However, as I did quite some research on some adjacent areas (which I reported on in this blog), I feel I should do the honors for this model. Meet the EO 440, a pen-based tablet computer with an usual CPU.

Wbvanrij at English Wikipedia / Public domain

According to some sources, the 440’s (and its bigger brother, the 880’s) claim to fame is that PC Magazine called it in 2012 the the first true phablet. But that’s, of course, not the real story here. The point of the EO 440 is that it should have been the first pen-based tablet with a dedicated operating system. It should have been (but wasn’t) because it was conceived (and nearly built) by a start-up company that was one of the first that forsaw the potential of a computer that one uses like a block of paper (with a pen): the GO Corporation.

I could now paint you the picture of this company, but I already made a presentation on that in 2017 (English slides here, recorded German presentation here). The presentation largely follows the 1995 book of Jerry Kaplan (the founder of GO) called “Startup: A Silicon Valley Adventure”. It is a very entertaining book on the specific history of GO, and a good example of how startups work and what troubles they face.

GO incorporates in 1987, and subsequently develops a hardware prototype and an operating system to go with it. It uses an Intel 80286 processor. In January 1991 finally GO presents the Developer’s Release of their operating system called PenPoint. At that stage, PenPoint is already licensed to three other companies, namely IBM, GRiD, and NCR who want to use it on their own hardware developments. You may ask how to develop for PenPoint without GO hardware? Well, after all the prototype hardware is basically only a PC, and you can use several PCs models with specific graphic tablets to simulate the system.

In March 1991, Microsoft (that battled GO on every event stage since years) presents PenWindows, an add-on for Windows 3.1. This allows Windows to be used with a pen instead of a mouse and a keyboard, but (obviously) cannot change the structure of a Windows program in order to adequately support a pen. As a result, users have to e.g. hit UI elements with a pen that are far to small, and have to use the virtual keyboard too often. All Microsoft attempts that use a pen (including Windows CE) fail, and it is maybe this problematic experience that leads to a Microsoft that tries in 2012 to come up with a Windows version that is primarily meant for pen-based tablets although the majority of computers it runs on do not have a pen. This version, of course, is Windows 8.

In June 1991 NCR presents the first tablet that can run PenPoint, the NCR 3125 (proudly designed in Germany). It weighs 1.5 Kg and costs about $4800…

In April 1992 IBM presents the original ThinkPad computer which is a pen-based tablet (and weighs 2.8 Kg). It can run PenPoint. Tablets will not be a hit product for many years (spoiler alert :-), so IBM re-uses the “ThinkPad” trademark for its line of notebook computers (renaming the ThinkPad to ThinkPad 700T so it fits in the overall naming scheme). In case you ever wondered why a Notebook computer is called “pad” without resembling a pad or having a pen…

Apple starts selling the Newton in May 1992. It weighs only 400 grams and costs only $700. Although not sold in huge numbers, the Newton is probably the bestselling pen product of its time.

Where did all this leave the EO 440? And why is it called EO, not GO?

Well, in 1991 GO finds itself choked on the mobile PC market. Microsoft controls the PC operating system market and makes it impossible for GO to find manufacturers for components like BIOS or to sell their operating system to smaller PC manufacturers. The solution seems to be to change the CPU to a RISC model, thus not competing in the PC market. They have two choices: ARM or a chip from AT&T called the Hobbit. Apple just has decided not to use the Hobbit CPU and to switch to ARM for the Newton. When in doubt, GO (as maybe also other startup companies) always decides not for the objectively best option, but for the one they have to use as they are (again) short for money. AT&T, short of a customer for their Hobbit project, offers 10M$ in capital, so they go with the Hobbit (EO will be the only company to ever use the Hobbit and AT&T closes the Hobbit project down some months later).

As another consequence of the deal, GO has to cut out the hardware development division from the company and contribute it to another company AT&T wants to invest in. This new company is called EO (which means GO in Latin…). After that, GO is a pure Operating System company. GO finally releases version 1.0 of PenPoint in April 1992.

In November 1992 finally EO presents their first two models, the EO 440 and the EO 880, but sells them only from April 1993. The design has not the clear, time-less esthetics of the NCR 3125. It is designed by Frog design and has two ear-like protrusions on the top. These protrusions do not have a technical reason, the designers wanted to have them. It has a 7.5″ monochrome display and weighs 1 Kg. This is much handier than the 1.5 Kg of the NCR 3125 or the 2.8 Kg of the ThinkPad, but much heavier that the 400 grams of the Newton even if the Newton has a much smaller display of 5.2″ and can offer only roughly a quarter of the resolution of the EO 440.

The EO 440 sports a 20 Mhz Hobbit CPU and between 4 and 12 MB of RAM. PenPoint is provided on a 8 MB PCMCIA card. The computer offers serial, parallel, PS/2 keyboard, SCSI and VGA interfaces as well as a PCMCIA card slot. The 7.5″ monochrome display has a resolution of 480*640 pixels. Harddisks are optional. The EO 440 cost $2000 initially. There was also an optional mobile phone module which added a handset and an antenna, so you could use your device as some sort of smartphone… The batteries lasted abouth 3h.

The differences of the EO 880 to the 440 are a larger display, a 50% faster clock for the CPU and a much higher weight (and price).

Handwriting Recognition

The key of using a tablet and a pen in the same way as paper was considered to be handwriting recognition. This ability largely determined also the computational needs which exceeded what mobile computers until then had to provide. The handwriting recognition algorithm-wise seems not to have been a problem given enough computational power was provided.

This need was a problem for some 1993 pen tablets, namely the Amstrad PenPad which used a Z80 variant, the Z8S180 at 14.3 MHz and the Zoomer incarnations (e.g. the Tandy Z-PDA) which uses a NEC V20 at 10 MHz. This was the reason for the PC-based pen tablets to start at Intel 80386SL CPUs and for EO and Apple to look at RISC CPUs. For the Newton, every model used a faster ARM chip and early models were known for their sluggishness.

The need for a high computational power drove also the price and the battery requirements, which in turn increased the weight. Therefore, US Robotics’ Palm series of pen-based PDAs later on decided on another design alternative and did not attempt handwriting recognition. Instead, the user had to learn an easily recognizable alphabet (“Graffiti”), so the Palm could use a simple 68000-family CPU. This lead to a cheap, light-weight device that became very popular.

PenPoint as an Operating System

PenPoint supported the following features:

  • Priority-based, preemptive multitasking with processes and threads
  • Interprocesscommunication and semaphores
  • 32-bit flat memory model
  • Ability to run on RAM-only as well as disk-based computers
  • Support of DLLs
  • Heap memory allocation with transparent relocation and compaction (no fixed-length buffers)
  • Object-oriented message passing and subclass inheritance
  • Detachable networking and deferred data transfer
  • All hardware dependencies are isolated into a “machine interface layer” to facilitate porting to a wide variety of hardware and processor architectures
  • Kernel runs on both PC and pen-based machines

User-interface wise, PenPoint programs use mainly (paper) notebook metaphors. You could use tabs to go to a certain part of the notebook, you could have different notebools in a bookshelf, you had tables of contents and so on.

On a page you had the usual means of GUIs of the day: drop-down menus, buttons, scroll margins, the lot.

Also you had a number of commands you could execute with your pen on pages: the pen gestures. You could insert text, remove and edit text, call the options or selecting things.

BTW, PenPoint requires a portrait orientation of the display. Windows wants a landscape orientation. You can typically see for what operating system a computer model is meant when you look at the orientation of the manufacturers logo and the buttons 🙂

PenPoint-based computers are very rare today, the EO 440 is even rarer, and even more so the EO 880. Typically the NiCd batteries are long gone, and sometimes you do not even have a power supply. Some years I was lucky enough to find a EO 440 and last year I could purchase even a power supply for it!

Technical Data

EO 440EO 880
CPUHobbit @ 20 MHzHobbit @ 30 MHz
Display7.5″, 480 * 6409.4″ backlit
RAM4 – 12 MBsame
Portsserial, parallel, PCMCIA, PS/2 keyboard, SCSI, VGAsame
HDD (opt.)20 MB64 MB
Weight1 Kg1.8 Kg
Initial price$1999$2499


Gepard – Part 8: The 3rd life of the Gepard

May 23, 2020

Somewhat unexpectedly another person contacted me and told the story of the 3rd life of the Gepard. This person is Hans Carlos Hofmann. I already read his name before as it appears in copyright notices of some of the later versions of GDOS. As it turns out, he can shed some interesting light on the history of the Gepard after the phases I explained earlier. So, in the following, please find his information (translated from German by me). In the following, “I” means Hans Carlos Hofmann.

I received at that time the source code of the compiler from Mr. Müller as well as license to distribute it as a binary for computers of the Gepard architecture. The reason for that was that modifications of the compiler were needed in order to support MMUs (for the Motorola 68030) and FPUs (for the 68020 and 68030).

I, together with another student, then also developed GDOS further as I had a need for computational power. This other student was Harald Hellmann who developed the data processing for the measuring system for the ion drive at the University in Gießen.

I then used the finished system for my theses in Math and Physics so the system was also represented in the “Experimental Physics II” department (Ion drives were and are a topic in the Experimental Physics I department, translator’s note). At that time I validated a program written by Martin Berz, today a professor (at MSU, translator’s note) that calculated ion optics. The validation took place by comparing the results from his program with the ones from my independant implementation in Modula 2. Thanks to Modula, my implementation was more readable, but could not be run on a CRAY. I’m still a little bit proud in that the most errors could be found in his implementation.

The improved operating system that we called “OS/Science” (you will see in a moment why) received many, partially spectacular modules.

The first module was the interrupt handler. The original Gepards did not do interrupt handling, all the PCBs had was a simple jumper to order the CPU to get an auto vector. A program had to handle interrupts itself. If a program exited, e.g. due to an exception, an active interrupt jumped into nowhere, and you had to reboot the computer. All these problems were solved by my interrupt handler by dynamically creating machine code for the current situation. Using this handler, all the drivers were updated, one by one, so no polling was necessary anymore. This ended up that later applications could stand up to 3000 interrupts per second.

Exploiting this ability a “virtual processor” module (nowadays these would be called threads) was developed that ended up in a preemptive multitasking scheduler offering cooperative multi-user management. The latter features was added due to Modula-2 because else the compiler would have needed modification.

After the Gepard Computer company had ceased to exist, not only HS Computer built Gepard clones, but also the MediSyst company of Prof. Dr. Dimpfel (which existed until September 2005, translator’s note).

With my metrological applications I helped a lot of people to get their PhD, enabling quite spectacular presentations like these ones:

The technology of representing data by dynamically generated code was then used in other modules and applications. One of these applications was a Raytracing program from the begin of the 1990s. It computed an entire day for a single picture and heavily swapping data in and out using the MMU as there was 20 times more memory needed as existed physically. Here are some examples of the generated graphics:

Info Graphics
Visualisation of Atomic Force Microscopy

The “Gepards” were then built until the beginning of the 2000s. New cards were added, e.g. better graphic cards. One Atari (ST)-compatible graphics card could be used both for text and graphics, one that could switch between 1 byte color for every pixel and having 8 different, monochrome pixel layers, and one based on an Intel GDP chip that could execute graphics command in hardware. One big strength of the system was that, already in the 1990s, I could use up to four displays with realtime graphics.

Other things added were a SCSI subsystem for hard disks, tapes, floppy disk drives, optical drives, ZIP drives, CD and DVD drives, photo typesetters, a relational DB, a video subsystem, and a NFS V2 server.

Unfortunately, there was no new CPU/memory board among the new things. HS Computer did develop a Motorola 68040 prototype card, but the design turned out to be unusable due to problems that could not be solved in software. In a system like the Gepard where the bulk of the software is compiled by the onboard compiler, there is this chance to compensate for hardware bugs by centrally providing suitable code generation. The same problems occurred when a 68060 prototype was tried out. MediSyst wanted even to develop a 3 x 68030 + memory card but failed out of economic reasons.

The successor company of MediSyst, MEWICON CATEEM-Tec had the last generation of Gepard-like computers in production until early May of 2020(!). Due to the economic consequences of the Corona virus the now 75-year-old owner of MEWICON (and MediSyst), Prof. Dimpfel has now decided to close his company. Therefore, only a few weeks ago I had the honor to switch off the probably last of the Gepards in production – after over 30 years of Gepards still being used at the one or the other location.

The RANDOC Index: Part 2: Examples

May 1, 2020

Here are some examples of RANDOC Index values for some models as I see them.


model Rarity Specialty Attractivity Impact
C64 1 3 3 4
Atari ATW 4 4 4 1
ZX Spectrum 2 2 3 4
Apple I 4 2 4 2
Mattel Aquarius 3 1 1 2
Canon Cat 4 3 2 1
MCM/70 4 4 3 1
Tatung PC-2000 4 1 1 1

The RANDOC Index: Part 1

May 1, 2020

I like to classify things, and rare and old computers are no exception. Here is my classification scheme for Rare AND Old Computer, the RANDOC Index. As you might know or not know, current prices of old computers fascinate me, and so the Index aims mainly at explaining these prices. My secret hope is that it can also be used for other purposes, but I would not bet on it.

The target of the RANDOC Index is to cluster old computer models according to a number of categories. After having explained the categories in this post, I will give a short list of examples in the next one and try to explain some price phenomena based on the RANDOC Index values of these examples in Part 3.

The RANDOC Index is based on three categories with four values each. As these three categories cannot explain everything, I added a fourth one in case it is needed. This table names all the categories and values:



The “Rarity” category tells us how rare currently a computer appears to be on the market. You can think of it as the inverse value of the supply of a model on the market. It does not say how many units were built. We can safely assume that a model that has been produced in low quantities will never be in high supply on the market, but on the other hand, a rare-but-not-too-rare model with a certain value might be sold more often than a low value model. As this category depends on the market, it can move over time.

When one tries to explain why some models are rarer than others even if the same amount of them were produced, one can have some interesting theories. One (quite probable) theory is that computers meant for industrial or clerical use are much more often scrapped than home computers. That’s why e.g. a Canon Cat (which hides its talents quite well) is much rarer than a Commodore SX64. My favorite theory is what I call “supply retention”. An owner of a computer does not feel the urge to sell it simply because he does not know the price (because it appears rarely on the market) or because the known price is so low that the owner thinks that it is not worth the effort to sell it. A connected theory of mine is the “avalance theory” that says once a model suddenly appears on the market to a price attractive to the owners, it triggers several selling decisions. As a result, a model is not seen on the market for a long time, and then suddenly, some of them pop up in a short period. I always wondered whether one can trigger this by offering such a model on the market without having it, achieving seemingly a certain price, and then be able to buy such a model afterwards 🙂 This would cost only the ebay fee on the achieved price, but you would have to ensure that a) you win the auction yourself (else you are in trouble) and b) that the price is not so high that the ebay fee makes you poor. As it is too dangerous for my taste, I never tried it.

From time to time, also the market changes radically due to technical innovation. When you watch programs like “Antiques Roadshow”, sometimes you hear how different the market pre and post Internet are. When your market is restricted to a country or a region and if you do not really know how many things of a models were made, your perception of rarity might be quite different from a world where things like ebay and a more globalized trade suddenly reveal that a model is much more common than what you thought. For some antiquity types, this drove prices down quite a lot.

The values of the Rarity category are:

The supply of this model is plentiful at all times, and this seems never to change. A Commodore C64 is e.g. certainly always available even in the next 5 years.

available from time to time
These models might not be available every day, but with some persistence you’ll find them offered once every three months. A Sinclair ZX80 might be such a model.

rarely seen
Now for such models, you need some effort. They are not impossible to find, but rarely, maybe once every year, a few of these pop up. An Enterprise 128 might be such a model.

practically unavailable
You have read about these models, you might have seen pictures, and you might even want to own one, but even when you scan ebay for a year you will not find them. Some people have them, but these are probably not the people that will sell these models to *you*. An Atari ATW 800 is currently such a machine.

Technical or Historical Specialty

This category tells us how much a certain model differs from its “run of the mill” competitors at about the same time. This difference can be due to different technical features or because it played a certain special historical role (although sometimes, both aspects coincide or are caused by each other). Once stated (probably with a little bit of hindsight), this value basically stays the same.

The values of this category are:

like any other
This model is no different from the typical model of the era. Basically the same technical features, the same original price, the same historical value. Nowadays PCs (of the same time period) are like any other PC.

a little bit different
For these models, there is at least one clear difference to the competition of the same era. Let’s take the ZX Spectrum. It is technically nothing special at the time, but it packages its abilities in only a few, cheap chips. Therefore, it can be offered for a much lower price. This is the economic difference (coming from a technical difference). As a consequence, it triggers almost single-handedly the creation of a thriving games development industry in the UK.

quite different from average
For this model, there is more than one difference to the competion, but these differences are still within the theoretical technical reach of the competing manufacturers. A Commodore C64 has a better sound than all its competitors and comparable graphics to other top end models, but it offers the entire package to a lower price.

This model has features that change the market. It has not only differences, but these differences has the potential to make the competition look old and to be needed to be replicated by the competition. However, this does not necessarily mean that this potential was fulfilled historically (this information is represented by the “Impact” category). From this point of view the Apple Lisa and the Apple Macintosh were disruptors. The Lisa was famously unsuccessful on the market, and only after the Macintosh got some traction (which took time), all new computers needed a GUI.


This category represents how attractive a certain model is for the market. Therefore, it is a sort of demand (like in supply and demand). Attractivity results from a multitude of factors. Each of these factors typically addresses a different slice of the market, and the more of these boxes are ticked by a model, the more people want to have it. Some of these factors (and their corresponding market slices) are:

  •  rarity
    there are only a few people that are drawn to rare computers. In itself this does not motivate too many people, but as an additional aspect it is important to a lot of people. One example is the guy who collects Commodore computers and wants to differentiate from other Commodore collectors by also owning e.g. a P500.
  • attractivity (and unattainability) at the time when this model was on the market
    One strong motivator to buy a model is when one always wanted to have one, but couldn’t afford it at the time. Now that prices are (maybe much) below the original price and the own income is much higher, the old desire is re-awakened.
  • specialty
    Also for this factor there are relatively few people that buy something solely out of this reason, but as an additional factor, it often motivates people.
  • having it owned at the time (of when this model was on the market)
    Another strong motivator is having once owned a certain model, often when one was young. As a consequence, often positive feelings are connected to such a model and re-owning it might connect one to these positive feelings.
  • other reasons
    Of course, there are many more reasons to find a certain model appealing. Maybe you know one and want to add it in the comments?

This value changes over time. The values of this category are:

(Almost) noone is interested in this model. It was never attractive at the time, maybe it was an industrial model that was never in young ownership at the time. It has maybe no specialty merits, and the current rarity completely satisfies the few people that wanted to have one. In my opinion many generic PCs fall into this category, many yet-another-CP/M machines.

attractive to some
This model ticks a few, not so important boxes. A Canon Cat is not a houshold name and probably not many people knew about it when it was new. However, the connection to Jef Raskin, its technical speciality and its rarity makes it attractive to some people. If you have never heard of this model, that’s exactly my point. If you have heard about it, well, you are the target audience of this blog, aren’t you?

attractive to most
This model ticks some important boxes. A C64 was (almost by definition) never rare, but it was either owned by a lot of people, or wanted by some of the others. It is a historically important computer and has a few technical merits.

attractive to all
This model ticks all boxes or ticks some important boxes strongly. A NeXT Cube is (not very, but quite) rare, it was at all times very attractive, but at the time also very expensive. It is a historically interesting computer, and is of a very attractive design.


This category represents the influence of a computer model on the further development of the industry. It is sometimes needed to explain effects outside the other categories. Also this value stays basically the same.

The values of this category are:

no impact
This value is for models that made no lasting impression on the market. Either because they appeared to early and the market was not yet ready for them or they appeared only in small numbers. Good examples for “no impact” computers are the MCM\70 and the Atari ATW 800. You’ve never heard of them? That’s what I mean 🙂

local impact
Such models had a strong impact in their local market, but not outside. The Micronique series of Hector computers were very well known in France in the 1980s, but outside France barely anyone knew them. Typical reasons for that (as in this case) is that a product is exclusively available in a certain language which is not English :-).

regional impact
Such models had a large impact, but, for some reasons, only on a certain region of the world. A good example are MSX computers. A household name in Asia, but only an exotic appearance e.g. in Europe.

global impact
Such models changed the industry. Either because they were so innovative that the advantages they offered could not be ignored by the competition (like in the IBM S/360 family case) or they came from such a strong competitor that this fact alone ensured market success and orientation of parts of the industry towards it (like the IBM PC).

Now, you might object heavily to some of my examples. First, the values in the categories are not fine-grained, and their definition gives much room for interpretation, and that’s all true, but I wanted to keep the thing manageable. Second, and more importantly, all these categories look at computers purely from a market’s perspective. This perspective changes over time, but it changes even more significantly with different markets, e.g. different countries. A Sinclair ZX80 is much more common in the UK than in Germany.

Intel 4004 μ-Computer (SIM4-01)

April 19, 2020

Some time ago, there was an Intel 4004-based singleboard computer for sale on ebay. The name of this computer was printed on the board: “Intel 4004 μ-Computer”. I did not buy it (also because the price was too high), but I never had heard about it, plus it was based on the first commercially available microprocessor, the Intel 4004. Now, the Intel 4004 is not only well known for its historical importance, but also that it was not really meant to be the heart of a small computer as it was designed to be used in a calculator. So what is this “mu-Computer” contraption?

Intel designed the 4004 as the CPU chip in its 4-chip family from 1971, the MCS-4. The other MCS-4 chips were the 4001 (masked-programmed) 256 bytes ROM, the (80 x 4 bit) 4002 RAM, and the 4003 shift register. The 4003 was meant as the basis to provide MCS-4-based systems output ports e.g. in order to control displays or other peripheral devices. The MCS-4 chips used a 4-bit data bus (although the 4004 supported 8 bit commands). The clock frequency of the 4004 was 500 – 740 kHz.

In 1971 Intel was mainly a RAM and ROM manufacturer who sold only components to their customers. (Mask-programmed) ROMs were ROMs whose content needed to be set during the actual manufacturing of the chip silicon, so a change in ROM contents needed several weeks until a new ROM with the new content was available. This was of course a problem for the software developers because they could change their software quickly, but then had to wait weeks until the new version could be used. Therefore, also in 1971 Intel announced a new product, the first EPROM. This was a ROM type whose content can be changed in a matter of minutes as it was stored in a re-writable manner on the chip (EPROMs are the chips with the glass window in the middle and often a sticker on top of that window). So, developers could program their system, create a new version of their software on their system, and when the software was stable, a traditional (and much cheaper) ROM could be produced without the need of many iterations.

Therefore, Intel felt the need to support developers using their MCS-4 products with first, a development system based on the MCS-4 that accepted either ROM or EPROM chips and second, with an EPROM programming device that allowed to load new content to an EPROM (an EPROM burner in today’s technology). The first development system is the 4004 μ-Computer. “mu” like the greek μ sign for “micro”. How did that device look like (that later on basically every CPU manufacturer would produce and sell for promoting their own design)?

The 4 large sockets on the top left corner can hold either a ROM or an EPROM chip each. The small chip on the bottom left corner is the 4004 CPU (tiny because of the 4 bit bus), and the 4 sockets to the right of it can hold a 4002 RAM chip each, giving it a whopping maximum of 160 bytes of writable memory. The other chips on the board are not really interesting (at least from my point of view). I read that the board actually has no 4003 chip on it, probably because there is no real standard design in a MCS-4 computer for it as this depends on what devices you want to connect. There are no recognizable interfaces on the board, just an edge connector of no recognisable standard at the bottom. So how do you connect something to this card? Meet the Intel MCB4 (Micro Computer Connector Board) chassis:

The chassis (the box on the bottom helds an 4004 mu-Computer on the front slot, and the second device, the EPROM burner on the back slot. It offers some switches, some lights, some interfaces and one (E?)PROM socket. You connect a TTY to the chassis, and voila: an entire microcomputer, ready to use.

Now, if you looked at the first picture and wondered where it said “mu-Computer”, well it doesn’t, it says “SIM4-01”. This is the (slightly) later name for the board. See, the problem was that Intel did feel that it was a component manufacturer, not a producer of computers, and that it should not be perceived (especially by its computer manufacturing customers) as such. Therefore, after some management thinking, this was not the 4004 micro-computer, this was the MCS-4 simulation device (SIM4-01), and you will find it much more often under that latter name. If you are interested in the history of the 4004 mu-Computer and especially the aspect of Intel shying away from being perceived as a computer manufacturer, I can wholeheartedly recommend the excellent article of Zbigniew Stachniak on exactly this topic (see [Stachniak] in the references).

The 4004 mu-Computer was probably the first computer with a 4004 CPU, the first microprocessor. Therefore, one can probably also call him the first microcomputer (if its definition requires a microprocessor). It (and its descendants like the twin brother SIM4-01 and the younger, healthier brother SIM4-02) can even be the only historic, commercially available, programmable computer with a 4004 CPU.

Technical Data

Manufacturer: Intel
Model: 4004 mu-Computer (a.k.a. SIM4-01)
CPU: Intel 4004
ROM: 4 sockets for 256 bytes Intel 4001 ROM, or 1701 or 1702 EPROM each
RAM: 4 sockets for Intel 4002 RAMs (80 x 4 bit = 40 bytes each)
Interfaces: 72-pin edge connector, allowing for 6 I/O ports (2 in and 4 out), and TTY
Released: 1971
Initial price: ???


The Value of a Computer over its Lifetime

April 13, 2020

I’m currently in the process of compiling a presentation on “Collecting Old Computers as a Hobby” for the upcoming VCFB 2020 event. I’m not sure whether it really fits in that presentation, but I made a highly speculative, subjective chart on, what I think, the value of a computer over its lifetime looks like. Here it is:


Let me explain. First, I differentiate the lifetime of a computer into 4 phases.

The “phase of usefulness” starts when the model comes onto the primary market, i.e. when it’s new. Obviously, the manufacturer thinks that it will be bought because it offers some benefit to the buyer. It is sold at the initial price (which typically will decrease over the time the model is offered by the manufacturer). I think it is not unreasonable to assume that the value of the computer, once bought, drops exponentially in this phase, especially when it it a state-of-the-art model. As it approaches the end of the usefulness phase, it typically drops to a more symbolic value that is basically determined by the minimum amount a seller is willing to accept in order for the entire transaction to make sense at all (i.e. compensating for the effort of the seller to offer, handle, and sell the thing. If the offered price would be any lower, it would not be worth the effort of the seller to do anything. If this computer model was bought originally by a company, it is most probably written off since some time anyway, to the company cannot put any value on it. The length of this phase depends probably on different factors, but maybe something like 10 years is not completely wrong for a computer. This phase transitions into

The “garbage phase“. The computer has no usefulness anymore, but it is not so old already that there are sentimental feelings connected to it. In this phase, the value is minimal and probably quite stable around the symbolic value I tried to establish in the last paragraph. Maybe in this phase it is not handled by the original owner, but by traders who specialize in surplus goods, making the price of the computer maybe even a tad lower because such traders are more efficient at low prices. I put a length of this phase also at around 10 years and try to substantiate that in the next paragraph, because the next phase is

The “sentimental phase“. This phase has a lot to do with the average age of a typical collector of a computer. Typically, they either had this computer model when they where young, or they were around when this model was fresh on the market. Since this phase probably 20 years or more has passed and they remember this time in retrospect with positive feelings (who does not remember his/her youth with joy? The music was much better than nowadays, the world was not that complicated and there was much less hatred and selfishness in the world). Also, being his his/her late 30s, 40s, and early 50s, things start to settle down for yourself. Often, there is less shortage of money, and less time needed to find a partner. For sure you don’t have a midlife crisis, but all this urge to buy a motorcyle… 🙂 Now, this is the time to feel young again with the dream machines of your youth, to fulfill unfilled dreams of owning cool stuff for a bargain. As a result, people start collecting computers. This process starts slowly, only a few people start collecting, but now the prices raise again. More and more people collect computers, maybe collecting becomes cool, and , with the raising prices, maybe even a good opportunity to invest money.

My theory is that the prices of a computer a collector wants to buy are:

  • not dependent on the initial price
  • dependent on a bunch of other factors, e.g.
    • how rare the model is
    • how many other collectors the model want
    • and, almost more importantly, what the buyer is willing to pay for a sentimental, or decorative, or conversation piece

I have also the impression (although I did not invest time into this to prove it) that these new values are clustered (and I put some estimation of how large these clusters are in the figure). Also, these price clusters increase over time as inflation and salaries increase over time. Finally, people think that the prices will go up forever and start thinking of them as investments (which will draw in also non-collectors that will act as investors).

This phase ends when the majority of the collectors starts getting so old, cool stuff from one’s youth is getting less important. This happens when no younger collectors replace the old ones, e.g. because the age of computers being cool stopped at some point. There are some areas where new collectors seem to continously “re-grow”, e.g. art and antiquities. There are other areas that flourished over many years, but suddenly came to a lack of collectors, e.g. stamps. My assumption is that, as computers became utilitarian in the 2000s instead of being fresh and not understood by parents, they aren’t as magical to millenials as they are to earlier generations (this is probably different to consoles and game-related computing, these will probably thrive much longer). If my prediction is correct, then prices at the end of this phase will drop dramatically as the market vanishes and the budget per collector gets much smaller.

It is also no clear to me what the next phase after the sentimental one is, and how this changes the prices for old computers.

In summary, here some conclusions:

  • most computers have their highest value when they are sold initially
  • the best phase to buy an old computer is the garbage phase (which is long gone for computers up to the year 2000)
  • do not buy old computers as a long-term investment

So, here you have it, my wild, highly speculative look into a crystal ball. What do you think?