| [ Main index | BBC micro site: General 8 bit computer stuff | Durability of old computer equipment | ] |
Original commercial tapes that I tried, most of them for the BBC micro, were almost always perfectly readable. Tapes that people recorded their own programs on (including my own tapes) develop a problem with dropouts.
The magnetic layer has become rough (I have no idea what causes this, but see below with the 5.25 disks for more on this) and the tape recorder runs slower and slower until it quits (as if it reached the end of the tape). Cleaning of the heads helps very briefly suggesting material is given off to the heads (which would give friction even if the tape surface was not rough). One tape where I thought the problem might be in the uneven winding of the tape, I played on a strong tape recorder back and forth for 20 minutes or so. Result: no improvement but there was a little stack of dark particles that had come from the tape... Again the rough surface problem.
I estimate this happens in about 1 per 100 (commercially produced) cassettes.
This is almost not a problem with commercially produced cassettes. Probably because they only get recorded on once; cassettes for ones own programs are used much more intensively, recording multiple times on the same piece of the tape, and reading lots of times. This probably causes the magnetic layer to wear, and the signal to deteriorate. On some of my own tapes I have found dropouts where the volume went completely to 0 for 0.5 s... The commercial tapes do sometimes have 'almost dropouts', where the volume suddenly changes to almost 0, but the signal is still present, also this volume drop is very short, and if the tape is not readably by the computer (very rare, rough estimate is also 1 in 100) it is still decodable with my tape reading program bbctape.
This gunk is actually some of the layer of the disk... Result: a high pitched whine until you've used a few other disks (or a head cleaner disk) to rid the drive head of it. I've only experienced this with the following brand:
The German magazine C'T had an article about a similar problem a few years ago for PC disks (HD?DD? would have to look it up); In that case the disks gave off material that would stick to the drive head and would not come off, effectively killing the drive.
The page "recovering data from improperly stored floppy disks" was mentioned on the BBC micro mailing list. It mentions mould causing degeneration. I'm not sure this is really the case, in the cases I experienced this appears not to be the case: The disk surface was degenerated but there was no visible trace of mould. In fact, I can usually only see the degeneration of the surface by breathing on it, if the surface remains evenly coloured, then it's ok. If some parts get darker than others, then there's a problem. This has happened to disks that I know have been properly stored. Furthermore, other disks from the same source were ok so I just don't believe it's about mould at all. perhaps natural degradation similar to what happens with plastic cables where the plasticiser that makes the plastic pliable eeks out of the cable (this is the stuff that eats into styrofoam and even some types of plastic casing btw.).
This page further mentioned something that I completely disagree with: using alcohol on spotted disks to read and copy them. This is very very very very bad because alcohol will exacerbate the problem. Some type of grease/fat would be better. Not sure what's best yet, but I've tried for example using skin grease. This worked perfectly, no whine, the disk was read without problems.
Note that the same degeneration also occurs on tapes. You can see the uneven colouring (but no mould) directly on some, on others breathing on the tape will show the decay. Here's an example with a disc:
Early Acornsoft disks are among the worst, many are bad, and some were destroyed when I tried to read them. The surface was soft on these disks, completey unlike good disks. This stuff scraped off, stuck on the drive head and then cut a groove in that same disk. The disk was thus nuked... Trying to format/read a good new disk (onto which I wanted to copy the original) cut a groove on that too. I had to use a head cleaning disk at that point to rid the drive's heads of the gunk.
These give a scraping sound every revolution, i.e. resulting in more wear on the drive head and the disk itself; this means the disks aren't flat... I've experienced this with the following brands:
But reliability is always good if disks are kept dust free. "How good is that?" you may ask: well, I've not had any problems reading in any disk that was kept dust free (and I've read in hundreds of disks from many sources). Contrary to this, 3.5 disks from 1994 on that I used (various branded and brandless types) invariably go bad in the rate of ca. 20% (rough estimate) per year. I.e., disks that are stored dry/dust free go bad in very short time even without using them!
Interestingly, 3.5 disks are acclaimed in 1984/1985 for being reliable and rugged. Of course, long term data retention hasn't been tested at that time, so by reliable I guess they mean 'can stand dust and stuff' (things like leaving disks out of their jackets is bad for 5.25 disks and this WILL make them go bad (as dust accumulates, the disk surface can be damaged when the head touches the surface, so scraping the surface)). Cleaning the disks from dust before trying to read them should prevent that. That would leave only the impact of direct sunlight, it not only yellows your equipment, but kills bits apparently. It would make an interesting experiment to check this...
By 1994, floppies were of course quite a bit cheaper than in 1984, so I wonder if the quality of 3.5 disks is simply crap since they got cheap...
It's difficult to judge how often bits go bad, as many times, ROMs have first been copied to sideway RAM (in case of the BBC micro) and this may corrupt the images by either bad RAM or other problems like those related to the non-buffering of the address bus in Solidisk's SWR (BBC micro), or ROM copying programs that corrupt some bytes (this happened with a certain program published in "Micro user"). However, in an Acorn Atom I found evidence that EPROM 0-bits (=programmed state) do lose this state and revert to 1-bit (=erased state) after a long time (18 years in this case lie between programming and the time that I checked the EPROMs). A few bits in both the modified OS and modified FP-ROM, both 2532s, were 1 that were supposed to be 0... (and these bits must have been correct at programming time, as the machine did not work properly with these bad bits). I have no idea what the rate of this 'unprogramming' is though! Some datasheets mention that the EPROMs should last at least 15 years. I've also encountered unstable bits in an EPROM, obviously the state just before complete unprogramming. These bits read out as 1 or 0 on different times. You might be able to use them as a random number generator ;-)
There is an occasional dead CPU, a RAM chip, ROM (I've seen a bad Viewsheet ROM for the BBC), video controller chip or a disk controller chip in machines I've used and/or repaired for others (in total some 20 odd machines), but considering that many of these machines had been used extensively for many years, the number of dead/dying chips per se is not great and not much of a problem. Whereas defective floppies lead to data loss, with ICs you don't lose anything (assuming you can get replacement ICs, which is usually not a problem, I'll write more about this at a later date). The problem is mainly diagnosing which is the defective chip and having to desolder chips that aren't in sockets...
I've encountered 1 dead PS in a BBC, and a slow starting one (capacitor problem probably, not checked yet). Other things that one needs to be aware of is corrosion on connectors (for the BBC computer: the connectors on the keyboard in particular). A can of contact cleaner spray is really essential...
Example: On 17 October 2003 a capacitor in the power supply of my original BBC B exploded...
This has later happened with another 2 BBC micro's. I've also heard from other people who had this happen. So be warned when buying a second hand BBC micro, this is a common thing...
| To email me, go to this page |
Last modified: Thu Nov 1 23:47:03 CET 2012