Now for something different - Retrocomputing and data recovery...
Posted: Tue Feb 18, 2025 11:42 pm
Like many other people of my age who ended up with careers in computing, I first encountered computers in the early 1980s, with a series of home computers and also ones encountered at school.
The very first home computer we had was an odd one - It was a machine called the 'Video Genie' (https://en.wikipedia.org/wiki/Video_Genie) - a far eastern clone of the popular Tandy TRS-80 computer. From my recollection of the details of the machine we had, I think we would have had it in early 1981. It didn't do much on its own when turned on apart from print 'READY', but in a way that was a good thing - you had to type in basic programs to make it do something, and that was a great way to start learning to program in basic. One quirk of the machine was that it had wood efect end panels - unthinkable today, but at the time a lot of hifi equipment had the same sort of thing, so I guess the designers felt it added 'quality'...
Next was the ubiquitous (in the UK, anyway) Sinclair ZX Spectrum in about 1983 - it's difficult to know what to say about the Spectrum.... From one point of view, it was a masterpiece of design in getting the most possible functionality at the minimum price point. Looking at it another way though, it had deep flaws caused by the penny pinching design - the dreadful keyboard and the quirky basic implementation. It was great for games though and there were an awful lot of them. I did some programming on the Spectrum, but not really a great deal.
Getting into programming properly had to wait for the arrival of the next computer in our household - an Amstrad CPC6128 (these were sold under the Schneider brand in Europe) - this was purchased I think at about Christmas 1985. The CPC6128 was a much more professional machine than the Spectrum (even though both had the same Z80 processor) - it had a decent keyboard, a dedicated monitor, a floppy disk drive and something approximating a simple operating system as well as a decent basic interpreter. The reliability and speed of the floppy disk compared to tape meant that you no longer needed to spend several minutes hoping your program would load from tape before using or editing it. The CPC6128 saw lots of use until 1989 when we got our first IBM PC clone computer.
Until recently, I thought that all the programs I had written from those pre-PC days had been lost, but last month, helping my father sort through some old stuff, we found a box of Amstrad floppy disks that have been untouched since the late 80s or early 90s. There was even one disk labelled 'Astronomy'...
The Amstrad machine itself is sadly long gone (as has the Video Genie, the Spectrum, and I suspect all the tapes that contained programs saved from those), but I decided I had to have a go at recovering the information from the Amstrad floppy disks.
Anyone who remembers 'normal' 5.25 and 3.5 inch floppy disks will look at the photo above and be thinking that the disks look odd - that's because the Amstrad computers used an odd 3 inch floppy disk which wasn't used by much else. Apparently Amstrad (another manufacturer who liked to keep costs down to a minimum) got a brilliant deal from the 3 inch drive manufacturers that undercut the cost of 3.5 inch drives and guaranteed to stay below that price over time, so they went with the 3 inch drives. Actually, the disks themselves are really robust, but getting hold of the drives these days is hard as they haven't been made since the 90s. Basically you have to find second hand Amstrad hardware and get a drive that way.
Looking on Ebay, I was lucky enough to find an original Amstrad 'FD1' external drive for sale as 'untested - spares or repairs'. These were sold as an add on second drive for the CPC6128 systems or a way of adding a drive to the earlier casette only Amstrad 464 (along with a separate interface add on in that case).
This drive was missing the disk interface component that would have allowed it to be used with the cassette based computer, but I didn't need that, so the lower price was welcome. I decided to take a punt on the 'spares or repairs' status as the drives are apparently fairly reliable with exception of the rubber drive belt that perishes over time (and is fairly easy to replace).
Obviously just having a floppy drive wasn't enough - you have to have some way to control it unless you have the computer that it was supposed to plug into. Fortunately there's now a clever device called a 'GreaseWeazle' (https://github.com/keirf/greaseweazle) that you can buy for about £25 - this is a USB device that connects to a PC and can control, read from and write to pretty much any floppy drive via a standard floppy drive cable. Essentially, it's a tiny computer (ARM based) that can take the place that a floppy disk controller chip would have had in the old days, with software running on the little ARM cpu doing the job of hardware circuits in the old days.
I had to do a refurbishment on the floppy drive itself - test the PSU was putting out the right voltages, remove the old rubber drive belt, clean the capstan/rollers and replace the belt, clean the heads, etc. In this particular type of drive there are a couple of extra 'gotchas' - there is a 5V supply plug that needs to be removed internally, otherwise the drive feeds 5V back down the ribbon cable and can fry the Greaseweazle device. The designers of the drive also decided to label the pins of the standard 34 pin floppy cable in reverse, so if you connect the cable 'Pin 1 on drive to Pin 1 on Greaseweazle', nothing works (the drive spins and nothing else). Connect it in reverse though and all is fine... This information can all be found in more detail online - ie https://muckypaws.com/2020/09/03/greaseweazle/
With all of this done, I was able to start reading data from the floppy disks and getting the disks converted into a format that can be used with modern emulators. Surprisingly, over the 28 or so floppy disks there were only about 10-12 bad sectors which could not be read properly (checksum errors), and about half of those were in blank areas of the disk or duplicates, so very little data loss.
Things became more interesting when I started looking at the data on the disks. Back in the late 80s, I did a fair bit of assembly language programming using an assembler called 'Laser Genius'. I found a lot of the ASM files containing the code, but I was shocked to discover that they were binary files in a proprietary format, not just plain text. Thinking about it, given the limited memory and storage available at the time, it made sense to store these files in a more compact format (basic was saved in tokenized form in those days too), but it feels horribly wrong to my modern programmer senses...
Fortunately I was able to find an online copy of the Laser Genius assembler software and use it to load the source code files then 'print' them (fortunately printers of the day simply printed ASCII, and emulators can capture and save the printer output) - hurrah, now I have my source code back into text format. Unfortunately not quite so simple - at least a couple of the files were corrupt (even though the disks read correctly) - it looked like the end of the file had been overwritten by part of another file
One of the files affected in this way was sadly some code that I was keen on seeing again - an emulator for the 6502 CPU that I had written in Z80 assembly. You might ask why write something like that? Well, at school we had BBC Microcomputers, which used the 6502 CPU. We also had an excellent Physics teacher who also taught a bit of computing. He ran a course after school to learn 6502 assembly language programming and taught us all sorts of interesting things like how to multiply and divide in assembly when you only add add and subtract instructions. I'd wanted to carry on learning about the 6502 at home, but only had a Z80 computer, so I obviously decided to write an emulator (and in fact a 6502 assembler written in basic). I obviously had too much time on my hands back then...
With modern text editors, cross-assemblers, etc, it took about an evening to re-write the missing parts of the 6502 emulator following the original design from 1987 (probably not how I would attack the same problem now). Along with a minor bug fix in the assembler, that let me get to the happy position of running 'Hello, World' in 6502 assembly in an assembler/emulator mostly written by me back in 1987...
(Those who know 6502 will spot the 'c80' and 'r80' instructions that I added to call z80 code and return control back to the z80)
I still have a bunch of disks to explore in more detail to find out what is on them - some things I have found include fractal plotting programs, basic ray tracing programs, etc.
And the disk labelled astronomy? It contained some programs including a planetarium program named 'Galshow' which I think was a type-in program from a book - I can't find any reference to it online - and a couple of other programs that I wrote in the same style.
Here's the output of a program I wrote called 'Skyplot' for this evening's southern sky (18th Feb 2025). The stars are a little sparse and it took a long time to calculate, but Mars and Jupiter are rougly right. Obviously whatever approximation I had for the moon wasn't really designed to cope with a date nearly 40 years into the future as the crescent moon is *not* currently between Mars and Jupiter. Ah well, can't win everything...
cheers,
Robin
The very first home computer we had was an odd one - It was a machine called the 'Video Genie' (https://en.wikipedia.org/wiki/Video_Genie) - a far eastern clone of the popular Tandy TRS-80 computer. From my recollection of the details of the machine we had, I think we would have had it in early 1981. It didn't do much on its own when turned on apart from print 'READY', but in a way that was a good thing - you had to type in basic programs to make it do something, and that was a great way to start learning to program in basic. One quirk of the machine was that it had wood efect end panels - unthinkable today, but at the time a lot of hifi equipment had the same sort of thing, so I guess the designers felt it added 'quality'...
Next was the ubiquitous (in the UK, anyway) Sinclair ZX Spectrum in about 1983 - it's difficult to know what to say about the Spectrum.... From one point of view, it was a masterpiece of design in getting the most possible functionality at the minimum price point. Looking at it another way though, it had deep flaws caused by the penny pinching design - the dreadful keyboard and the quirky basic implementation. It was great for games though and there were an awful lot of them. I did some programming on the Spectrum, but not really a great deal.
Getting into programming properly had to wait for the arrival of the next computer in our household - an Amstrad CPC6128 (these were sold under the Schneider brand in Europe) - this was purchased I think at about Christmas 1985. The CPC6128 was a much more professional machine than the Spectrum (even though both had the same Z80 processor) - it had a decent keyboard, a dedicated monitor, a floppy disk drive and something approximating a simple operating system as well as a decent basic interpreter. The reliability and speed of the floppy disk compared to tape meant that you no longer needed to spend several minutes hoping your program would load from tape before using or editing it. The CPC6128 saw lots of use until 1989 when we got our first IBM PC clone computer.
Until recently, I thought that all the programs I had written from those pre-PC days had been lost, but last month, helping my father sort through some old stuff, we found a box of Amstrad floppy disks that have been untouched since the late 80s or early 90s. There was even one disk labelled 'Astronomy'...
The Amstrad machine itself is sadly long gone (as has the Video Genie, the Spectrum, and I suspect all the tapes that contained programs saved from those), but I decided I had to have a go at recovering the information from the Amstrad floppy disks.
Anyone who remembers 'normal' 5.25 and 3.5 inch floppy disks will look at the photo above and be thinking that the disks look odd - that's because the Amstrad computers used an odd 3 inch floppy disk which wasn't used by much else. Apparently Amstrad (another manufacturer who liked to keep costs down to a minimum) got a brilliant deal from the 3 inch drive manufacturers that undercut the cost of 3.5 inch drives and guaranteed to stay below that price over time, so they went with the 3 inch drives. Actually, the disks themselves are really robust, but getting hold of the drives these days is hard as they haven't been made since the 90s. Basically you have to find second hand Amstrad hardware and get a drive that way.
Looking on Ebay, I was lucky enough to find an original Amstrad 'FD1' external drive for sale as 'untested - spares or repairs'. These were sold as an add on second drive for the CPC6128 systems or a way of adding a drive to the earlier casette only Amstrad 464 (along with a separate interface add on in that case).
This drive was missing the disk interface component that would have allowed it to be used with the cassette based computer, but I didn't need that, so the lower price was welcome. I decided to take a punt on the 'spares or repairs' status as the drives are apparently fairly reliable with exception of the rubber drive belt that perishes over time (and is fairly easy to replace).
Obviously just having a floppy drive wasn't enough - you have to have some way to control it unless you have the computer that it was supposed to plug into. Fortunately there's now a clever device called a 'GreaseWeazle' (https://github.com/keirf/greaseweazle) that you can buy for about £25 - this is a USB device that connects to a PC and can control, read from and write to pretty much any floppy drive via a standard floppy drive cable. Essentially, it's a tiny computer (ARM based) that can take the place that a floppy disk controller chip would have had in the old days, with software running on the little ARM cpu doing the job of hardware circuits in the old days.
I had to do a refurbishment on the floppy drive itself - test the PSU was putting out the right voltages, remove the old rubber drive belt, clean the capstan/rollers and replace the belt, clean the heads, etc. In this particular type of drive there are a couple of extra 'gotchas' - there is a 5V supply plug that needs to be removed internally, otherwise the drive feeds 5V back down the ribbon cable and can fry the Greaseweazle device. The designers of the drive also decided to label the pins of the standard 34 pin floppy cable in reverse, so if you connect the cable 'Pin 1 on drive to Pin 1 on Greaseweazle', nothing works (the drive spins and nothing else). Connect it in reverse though and all is fine... This information can all be found in more detail online - ie https://muckypaws.com/2020/09/03/greaseweazle/
With all of this done, I was able to start reading data from the floppy disks and getting the disks converted into a format that can be used with modern emulators. Surprisingly, over the 28 or so floppy disks there were only about 10-12 bad sectors which could not be read properly (checksum errors), and about half of those were in blank areas of the disk or duplicates, so very little data loss.
Things became more interesting when I started looking at the data on the disks. Back in the late 80s, I did a fair bit of assembly language programming using an assembler called 'Laser Genius'. I found a lot of the ASM files containing the code, but I was shocked to discover that they were binary files in a proprietary format, not just plain text. Thinking about it, given the limited memory and storage available at the time, it made sense to store these files in a more compact format (basic was saved in tokenized form in those days too), but it feels horribly wrong to my modern programmer senses...
Fortunately I was able to find an online copy of the Laser Genius assembler software and use it to load the source code files then 'print' them (fortunately printers of the day simply printed ASCII, and emulators can capture and save the printer output) - hurrah, now I have my source code back into text format. Unfortunately not quite so simple - at least a couple of the files were corrupt (even though the disks read correctly) - it looked like the end of the file had been overwritten by part of another file
One of the files affected in this way was sadly some code that I was keen on seeing again - an emulator for the 6502 CPU that I had written in Z80 assembly. You might ask why write something like that? Well, at school we had BBC Microcomputers, which used the 6502 CPU. We also had an excellent Physics teacher who also taught a bit of computing. He ran a course after school to learn 6502 assembly language programming and taught us all sorts of interesting things like how to multiply and divide in assembly when you only add add and subtract instructions. I'd wanted to carry on learning about the 6502 at home, but only had a Z80 computer, so I obviously decided to write an emulator (and in fact a 6502 assembler written in basic). I obviously had too much time on my hands back then...
With modern text editors, cross-assemblers, etc, it took about an evening to re-write the missing parts of the 6502 emulator following the original design from 1987 (probably not how I would attack the same problem now). Along with a minor bug fix in the assembler, that let me get to the happy position of running 'Hello, World' in 6502 assembly in an assembler/emulator mostly written by me back in 1987...
(Those who know 6502 will spot the 'c80' and 'r80' instructions that I added to call z80 code and return control back to the z80)
I still have a bunch of disks to explore in more detail to find out what is on them - some things I have found include fractal plotting programs, basic ray tracing programs, etc.
And the disk labelled astronomy? It contained some programs including a planetarium program named 'Galshow' which I think was a type-in program from a book - I can't find any reference to it online - and a couple of other programs that I wrote in the same style.
Here's the output of a program I wrote called 'Skyplot' for this evening's southern sky (18th Feb 2025). The stars are a little sparse and it took a long time to calculate, but Mars and Jupiter are rougly right. Obviously whatever approximation I had for the moon wasn't really designed to cope with a date nearly 40 years into the future as the crescent moon is *not* currently between Mars and Jupiter. Ah well, can't win everything...
cheers,
Robin