My brother-in-law just showed me this absolutely amazing piece of machinery (link in Russian):
If you don’t speak Russian (mine isn’t nearly good enough to read the page), use Google Translate; it’s pretty good on this particular page, although you may wish to know that: “car” probably means “machine”, “lamp” probably means “tube”, “earn” means “work”, and maybe a few others that you can hopefully figure out as you go!
This video in particular is amazing, he shows the internals of the machine in operation as it calculates the square root of 2:
It appears that this is a rather complex purely mechanical calculator that is capable of running simple (or really, not so simple!) stored programs that are encoded as metal stops pushed into a programming board. The author of the page has written a few additional pages about programming the machine (linked near the bottom) that describe in more detail the specifics of how it works.
It looks to me like it is constructed more or less like a pinwheel calculator, simply more. I now dream of acquiring one of these!
That thing is amazing! That thing must have cost a fortune to build. I wonder how many of them are still out in the wild, stuck in old buildings gathering dust. The Russian version of the Apple 1!
Wow! What a machine! Seems like the 170 has branches but no specific loop instruction, the loop being implicit in the way the program cycles from the end back to the beginning. But that’s enough!
Some choice quotes from the (machine translated) article:
these amazing machines, perhaps the only mass-produced (more than three hundred thousand from 1955 to 1983) Turing-complete mechanical computers.
Total, the machine has ~ 50 steps of command memory and 56 integer registers of 12 decimal places.
fifty registers are organized in the form of five rotating drums with ten registers each
at the metro station of the Lenin Library there is the Lenin Library! The entire twentieth century there are flocked binding copies of books produced in our country, and books about Ascot should be there. Indeed, in the electronic catalog Leninki found more than three hundred books mentioning Ascot 170 . Moreover, it turned out that over the past decades the mode of operation of the library has greatly softened, and today anyone can come and photograph the books on a non-professional camera.
Reading these books was not easy: not only did they describe the programming of a machine developed long before the appearance of the first high-level language, Fortran. So they were also written in accounting terms.
And, via the comments on @elb’s link, the Robotron page on the various multipliers (and dividers?) which were huge machines built into the desk that the 170 sits on. Initially mechanical, then thermionic, then (perhaps?) transistorised: Zubehör für Buchungsmaschinen
Yes, my understanding is that they were made in post-war East Germany. Many East German products were transported and/or reproduced across the Soviet Union. And, of course, many European and American products were cloned by the Soviets. Perhaps most famously in the computing world, the Elektronika BK (an LSI-11 clone) and the Mikro-80 (an 8080 clone):
The people from Museo del computer (northern Italy) have several such machines of different types and sizes, but all Ascota. They brought two slightly different machines to Vintage Computer Festival Europe (Munich) in 2015, where I made several pictures of it. One of the machines even has a math coprocessor for multiplication. Signaling between calculator and coprocessor was electrical, calculation in the coprocessor itself was mechanical IIRC. The Italians struggled with programming the machine back then, due to lacking documentation. I don’t know a current status.
From the Robotron website, you already posted above, section “Programmierung” (machine translation done by DeepL):
The machine was programmed using exchangeable plug-in boards, into which differently shaped metal plugs were inserted at different points and read out by metal sensing fingers. In this way, more than 90 different commands could be programmed. Theoretically, a program could consist of up to 53 (sequentially processed) steps, whereby each step could contain several commands. In order to keep the width of the plug-in board small, several commands were possible for some sensing levers, selectable by different heights of the programming plugs and dependent on other commands. Some commands could also be executed manually with the keyboard, others could only be started via the program plug-in board. In order to be able to administer different form types or multi-line calculations, it was possible to switch between two programs (manually or also software controlled).
Behind the tabs (which also defined the table columns on paper) were the command connectors of the commands to be executed for the column in question. So the booking cart stopped at these tab stops, waited until the input was complete if necessary, then executed the program step and went to the next tab stop. For machines with an electronic multiplier or with perforated tape technology, there was another, narrower programming board on which the control of these peripheral devices was coded (triggering via electrical switching contacts).
The creation of the programs according to the customer’s requirements was usually not done by the user, but by the specialists of the booking machine factory or the Robotron sales staff. For programming, the plug-in board was pulled out of the machine, a special template was placed as a legend and then the metal plugs were hammered in according to the command sequence designed on a form.
There were also developed plug-in boards where the height of the program plugs could be changed by lever to increase the program capacity. Whether this development ever came out via the prototype status is questionable.
Most interesting fact for me: programming was usually done by the manufacturer, not by the user. And well … programming was done using a hammer. There is a picture in this section
I’m reminded, a long time ago, of a time when I said my job was somewhere in the middle of hardware and software, to the disbelief of the banking representative. I didn’t mean that I programmed with a hammer…
@EdS: On a side note: you really should use DeepL for machine translation. Their English-German and German-English is pretty good, compared to the planet-sized robot brain. They offer other languages too, but I can’t say anything about the quality. Probably very descent as well.
This fascinating computer just turned up in a rather good answer on stack exchange:
From a relatively vague second-hand recollection, the two machines identified are this thread’s “Ascot 170” and the much more sophisticated “daro (Ascota) 1750 Magnetkontencomputer”
Introduced in 1974 it soon became a major seller to replace Ascota 170 and gaining new customers (*6). The base unit featured 5 KiB of wire ROM holding the system software plus whole 8 KiB of core based RAM for data and loadable programs.
A magnet card reader “Magnetkontenaufsatz daro 1294” was available as add on. It was to be put instead and replacing the paper feeder
CONFIRMATION! My mother confirms that these are the ones! Although the specific models and variants are, of course, lost to time, but that’s not really that important. Wish I could upvote this again, such fascinating machines!