Obsolescence of electronic components on early computer era traction

[contrex]

I particularly remember reading a comment from a senior BR engineer involved in the cl. 74 project, along the lines of 'never again' as far as doing that sort of major conversion.

mods note - split from here

I seem to remember reading that a contributing factor in the lack of reliability was the electronic systems, in particular the lack of robustness of the printed circuit boards, the way they were mounted, and the connections thereto. I gather that the rail environment is particularly hard on electronics. On a side note, don't the Class 91s suffer from shortages of 1980s components, e.g. 8080 processors?

 

[ac6000cw]

I seem to remember reading that a contributing factor in the lack of reliability was the electronic systems, in particular the lack of robustness of the printed circuit boards, the way they were mounted, and the connections thereto.

Yes, I remember those sort of comments too.

As someone who's been designing electronics since the 1970s to the present day (from hobbyist to professional), yes, in the early/mid 1960s I think the rail industry was still working out how to keep electronics reliable on moving vehicles. The shock forces and vibration can cause e.g. cracking of solder joints and cracking where wires enter the body of a component (in those days most electronic components mounted on printed circuit boards had wires which were soldered into the board). Also temperature cycling causes expansion and contraction of the PCBs and the components, plus low temperatures can cause water condensation problems.

 

[Haywain]

On a side note, don't the Class 91s suffer from shortages of 1980s components,

Well, 1980s technology is inevitably going to be more difficult to obtain now than it was it the time that is was state of the art. I would imagine that applies equally now to much of the technology of the 2000s.

 

[ac6000cw]

On a side note, don't the Class 91s suffer from shortages of 1980s components, e.g. 8080 processors?

Given when it was designed, I would have expected it to be using something more modern than the Intel 8080 (dating from 1976), even if it was only the slightly later, but faster and simpler to use, 8085. But more modern and powerful MPUs were around by the early 1980s, like the 16/32-bit Motorola 68000 (the first MPU I ever designed into anything, in the mid 80s, running at the heady heights (!) of 4MHz clock rate and whose architecture I still have a soft-spot for - it was very easy to program in assembly language).

 

[mike57]

Well, 1980s technology is inevitably going to be more difficult to obtain now than it was it the time that is was state of the art. I would imagine that applies equally now to much of the technology of the 2000s.

An interesting side effect of the march of technology:

A 100 year steam locomotive can be kept running,and new parts fabricated locally should they be needed as long as you have access to a skilled machinist and the right tools

A 50 yo engine which is incorpoating 1st gen integrated circuits, CPUs etc is going to be far more of a challenge,because local manufacture on a one off basis is not a viable option. So looking ahead one could visualise a situation where fully functional units from this time period could become much rarer than older steam powered and early diesel stuff

 

[contrex]

Given when it was designed, I would have expected it to be using something more modern than the Intel 8080 (dating from 1976), even if it was only the slightly later, but faster and simpler to use, 8085. But more modern and powerful MPUs were around by the early 1980s, like the 16/32-bit Motorola 68000 (the first MPU I ever designed into anything, in the mid 80s, running at the heady heights (!) of 4MHz clock rate and whose architecture I still have a soft-spot for - it was very easy to program in assembly language).

You are quite right. I ought to have checked... from a leaflet titled "GEC Transportation Projects Class 91 25kV BoBo locomotives for British Rail"...

The Class 91 locomotive and the Class 90 locomotives (also being equipped by GEC) have similar microprocessor control modules with obvious advantages for spares and test equipment.

The armature and field converters are under the direct digital control of a microprocessor based control system using a single Intel 8086 16 bit micro-processor. The programme itself is stored in EPROM (erasable programmable read only memory), and the data in RAM (random access memory). This provides flexible and adaptable control (being software based) whilst using standardised hardware and software. The package provides high accuracy with minimum drift and high immunity to interference. The self diagnostic (ie fault logging) facility will be particularly helpful in maintenance.

It also says 'The locomotives and their equipments have been designed for a 35 year life when averaging some 420,000 km each year.' I presume this is from 1988.

GEC Class 91 Locomotives

== Doublepost prevention - post automatically merged: 3 Jul 2025 ==

Well, 1980s technology is inevitably going to be more difficult to obtain now than it was it the time that is was state of the art. I would imagine that applies equally now to much of the technology of the 2000s.

The 8086 cpu (as I now know) used in the microprocessor control systems of the Class 91 was not exactly 'state of the art' in 1988, having been released 10 years before. So coming up to its 50th birthday.

 

[Harpo]

A 100 year steam locomotive can be kept running,and new parts fabricated locally should they be needed as long as you have access to a skilled machinist and the right tools

Off topic, but it’s a major difference between steam preservation’s ability to build everything it needs, versus diesel preservation only existing by burning its way through a diminishing stockpile of consumables.

 

[eldomtom2]

An interesting side effect of the march of technology:

A 100 year steam locomotive can be kept running,and new parts fabricated locally should they be needed as long as you have access to a skilled machinist and the right tools

A 50 yo engine which is incorpoating 1st gen integrated circuits, CPUs etc is going to be far more of a challenge,because local manufacture on a one off basis is not a viable option. So looking ahead one could visualise a situation where fully functional units from this time period could become much rarer than older steam powered and early diesel stuff

I suppose the big question is to what extent you could replace them with something more modern. The Intel 8086 mentioned as used in the Class 91 is a well-documented and easily emulated chip.

 

[Cowley]

As requested by @D365, this thread has been split from the discussion about Class 74s

 

[ThirteenArches]

A lot depends on the design techniques and adequacy of documentation.

In the early days, pure logic design may well have been used, and if the circuit diagrams of the various modules are available (and accurate), it would be a relatively simple task to implement the logic using, say, one or more FPGAs (more than one if modularity is a useful aspect of the design). Logic levels will be different, but that really comes down to fairly obvious interfacing, and perhaps these days off-the-shelf, tested and approved, interface modules are available. For systems that are not too time- critical, a PLC might be an option, and these are available in safety- hardened implementations.

Later, the processors will have been programmed using, at first, assembly language specific to the processor used. If this is well constructed (modularised, with each module documented as to required behaviour) and commented (functionally, with the intention to illuminate operation), it may be possible to translate it into a modern language.

Later systems may have used higher- level language from the start, and these should be still simpler to translate, as long as it wasn't APL or FORTH. Again assuming the design was properly partitioned and documented.

Usually, in my experience, whatever the hardware and/or language used, systems generally weren't designed or maintained in a well- documented way, until safety became more of an issue. Les Hatton's Safer C, MISRA C and so on have possibly removed the worst abuses, but the whole design process also got bogged down in a maze of obfuscation and bureaucracy, with recursively- nested consultants and glacial approval procedures used by the Gods for arse- protection rather than improving safety by using KISS and transparency.

So, with a diesel with early hardware, it all comes down to someone knowing (or being able to find out):

- what the functionality does and what its design criteria were
- the safety criticality of the function
- the environment in which it operates
- how to reproduce that functionality using available stuff
- the approval procedures

And then getting someone to build it, test it and install it. Just the sort of stuff I used to do 35 years ago, until I realised that the market had collapsed.

 

[Ashley Hill]

Wasn’t there a preserved 37 that had some of its cards stolen some years ago? I remember the owners appealing for replacements as new ones were prohibitively expensive.

 

[The exile]

So looking ahead one could visualise a situation where fully functional units from this time period could become much rarer than older steam powered and early diesel stuff

We have already reached (long passed?) this point with signalling, have we not?

 

[mike57]

So, with a diesel with early hardware, it all comes down to someone knowing (or being able to find out): ....

What you summarised is actually a huge amount of work, involving lots of different specialisations, very different from a machine shop, a chunk of metal and a skilled machinist. Probably not the sort of resources that the average heritage railway or preservation group will have on tap or be able to afford.

It also begs the question of what you actually want to present, take the 1903 NER Petrol Electric Autocar, its now got a modern diesel engine and generator, even if you wanted to run with the original engine/dynamo it is long lost as prior to restoration it was used as a holiday home. Thats before you get into issues of using petrol v diesel (thinking flammability and related issues). So what you see now looks like what you saw in 1903, but is quite different under the hood. Equally the Brighton Belle restoration has followed a similar path.

I see the problems associated with keeping the technology exactly as it was in 1960/70/80/90 as being almost impossible, so more modern technology 'under the hood' become almost a must, and even that comes at I suspect a considerably higher cost and difficulty than maintaining an older piece of kit, probably steam driven.

We have already reached (long passed?) this point with signalling, have we not?

Same issues with the added complexity that signalling has even more safety significance, so any changes are even more difficult to validate.

I was leading a project replacing outdated (and unobtainable) control systems in a safety critical industry a few years ago, and we took the easy way out, identify the safety dependency of every system (they all used the same basic technology), leave safety related systems alone, and replace the non safety systems, releasing spares to maintain the safety related systems for the remainder of the plant life. The majority of components were 'commercial off the shelf' (COTS) so by avoiding those systems with any safety significance we were able to solve the immediate problem, but once the 2nd hand stuff we 'released' is used up the task becomes huge. In this case plant remaining life was finite, with preservation you are trying to keep whatever it is going for 'ever'

In signalling terms I could see that once you move from mechanical signalling to electrical/pneumatic you could keep period 'look and feel' but it would have to be currently availalable systems internally.

On a more general note, I wonder if institutions like the NRM and the Science Museum are looking at these problems, and coming up with a road map for maintaining the most significant items, to be honest the run of the mill stuff used by the preservation movement will be a problem and it may be that stuff from the early 'electronic' era will become 'static exhibits'. The problem is compounded by the fact that the rail industy has not generally used COTS equipment

 

[eldomtom2]

There are plenty of people who are involved with vintage electronics as a hobby, it's just that they usually aren't part of the railway preservation community. As I previously mentioned, something like the Intel 8086 used on the Class 91s is a common and well-documented piece of kit. Publicly documenting the electronics used and reaching out to computer and electronics communities would go a long way in assessing how much of a problem there actually is.

I would also like to know why you think a modern diesel generator and electric motor would be more expensive and difficult to maintain than a steam engine.

 

[Paul AC]

GEC had a purchasing deal with Intel in the 1970s and 1980s which meant that it made economic sense to use Intel processors, memory chips etc. Not just processors for trains: also industrial, military and aerospace equipment via the very many sub-divisions of GEC. So, as an apprentice in the late 70s and then through industrial then railway systems designs as an electronics and software engineer, I used, and in many cases designed systems using, 8031 (8-bit embedded controller) on various jobs, 8085 (8-bit processor) on a 70s industrial PLC, 8096 (16-bit embedded controller) from the mid 80s on both industrial and railway jobs, and 8086 as used on pretty much every traction system designed by GEC at Manchester and Preston from the Docklands P86 fleet through 319, 90, 91, 465, 456, 365, Manchester phase 1 trams, and locos and units in Hong Kong, South Africa and elsewhere overseas. That 8086 MPU card was used in literally dozens of applications over a ten year design period, with one design revamp half way through that retained compatibility.

It's not just the processor cards: digital and analogue input cards of various types were produced in huge quantities for the same fleets. Hence why a class 90 owner has been buying up electronics racks from scrap units, as many of the cards are nearly identical and can be converted to suit the different fleets fairly easily. There's an electronics company in north west England that has an arrangement with Alstom (as successor to GEC) so that they can repair, convert and upgrade cards from older GEC fleets.

(for info, all the Brush cards I've seen from the GEC-8086 era use a Motorola 68000 processor; I know some strongly preferred the one over the other, but I doubt there was much difference in reality)

Roger Ford did an article in the August 2025 Modern Railways where he reported on a discussion with Stuart Broadbent, who is Obsolescence Management Director at Alstom (and I know them both very well and even got a mention in the article...). The discussion is wide-ranging, but obsolescence of components is as much a present-day problem as for older fleets such as 90s and 91s. In fact it may be worse, as modern chips may well go obsolete much quicker than older parts. Alstom monitors the availability/obsolescence of around 60,000 different components and arranges "last-time buys" of critical parts to support the current Alstom fleets worldwide.

Another point Roger makes is that he considers class 90/91 as about the most recent fleets of electronically-controlled trains that can be practically preserved. My view is that even these will rely on the availability of someone involved with their operation on fleet service, and access to the specific software maintenance tools.

 

[mike57]

I would also like to know why you think a modern diesel generator and electric motor would be more expensive and difficult to maintain than a steam engine.

Having been involved in other industries with similar obsolesence problems this is my take:

Motors and electromechanical control gear (Contactors, overloads, fuses, switchgear) Simple replacement with few problems beyond a bit of 'fitting' in cases where sizes are slightly different

Systems using well documented microprocessor architectures, 8085, Z80, 8086 and upwards (80x86) ARM, 68000, and a few others. More involved, but firmly in the 'doable' category, but with the proviso that these systems frequently have I/O cards or interfaces which use custom or obsolete chips, in one case I did a system where ALMOST everything was straightforwards, with the manufacturer offering an upgrade path, and the tools to support it, except for one 3rd party I/O card. One fairly simple part of the system actually soaked up as much time as the rest of the conversion to a modern system took. (This was late 70s tech in the late 2010's)

Custom systems, these are the ones that really cause trouble, I have seen systems with 'CPUs' made out of TTL logic chips for example. Frequently the supplier no longer exists, documentation tends to be sketchy at best. Really these systems have to be completely removed and re-engineered from scratch once the supply of spares has dried up, in our industry if they are basic control systems with no safety 'significance' then that process is time consuming but doable, as soon the system delivers any safety functions, rather than relying on other systems to maintain safety it becomes an almost impossible task, which would only be undertaken if there really was no alternative.

I have no personal experience of the rail industry, but talking to others who have the issues are the same.

The problem, if for example a motor needs an adapter plate machining thats a fairly simple task, draw it up, get it made, fit it. It takes time and money, but as a job has a clear 'path' and the likelihood is it will fit and work. The same with the majority of steam engine parts I suspect.

As soon as you get into software and firmware the testing required demand expertise which will not come cheap, and there maybe unexpected results during testing, including 'bugs' in the original system, my experience is that you frequently encounter sometimes major issues which can mean costs and timescales mushroom.

On the mechanical side diesel engines can be swapped out for a similar modern unit, again fairly straight forwards, it happens all the time across all industries.

In terms of the preservation movement, there is a good infrastructure and knowledge base for steam stuff, but dealing with for example early 80s tech I suspect would be way beyond the resources of the preservation movement in a lot of cases. A series of 'bodges' which allows an enthusiast to run an early arcade game for example would probably not be acceptable on a railway engine, even one limited to 20mph or so.

 

[edwin_m]

With software there is a risk that the source code has vanished, or is incomprehensible to anyone other than the original authors who moved on years ago.

 

[ThirteenArches]

With software there is a risk that the source code has vanished, or is incomprehensible

Vitually a certainty, especially if it is very old, written in assembler (or FORTH godhelpus) or some obscure early PLC language. One comment I heard of, from Denisovan C code running on Precambrian Linux, for a particularly incomprehensible block of code was

/* Because I can */

On the other hand, any maintenence of existing systems beyond utterly trivial needs similar awareness of the safety- criticality of its function. If it's only controlling cab heating, troll ahead. If it's engine control, you can probably get away with a lot, perhaps with reduced performance. If it's head and tail lights, get them wrong and someone could get hurt. If it's signals, DSD or brakes....

 

[birchesgreen]

With software there is a risk that the source code has vanished, or is incomprehensible to anyone other than the original authors who moved on years ago.

To be honest its incomprehensible to the original author after a few years in many cases too...

 

[Paul AC]

With software there is a risk that the source code has vanished

and even if you have the source code, do you have the compiler (etc.) tools, and the correct hardware and operating system to run them on?

The Intel 8086 (etc.) toolset (as used on 90/91 etc. as previously mentioned) was originally designed to run on Intel's own development system hardware (Series 3, with 8" floppy disks, and Series 4 with a hard disk) but later had wrapper code added to allow it to run on MS-DOS. It still runs ok "today" from a Command Prompt or Batch File on 32-bit versions of Windows. A full code rebuild for class 91 (binary file size about 50 kbytes) used to take over 20 minutes in the late 1980s; longer if using floppy disks (temp files written to disk as you didn't have much RAM available). It takes less than 10 seconds on a modern PC.


There's a similar problem with documentation (I've used IBM Displaywrite, then WordPerfect) and CAD systems for drawings, printed circuit boards, ......

 

[ThirteenArches]

8" floppies.... CP/M.... arrrgghhh. About 40 years ago, I worked for a firm that did, amongst other things, alarm annunciator systems. One guy had just come back from Khartoum where he'd spent 6 months commissioning alarms in a power station. All the configurations were on an 8" floppy, which he brought into the office and started to copy in the dual- drive system. After about 15 minutes, there was a power cut. Both floppies irretrievably corrupted...

And that's another problem... even if the source & documentation were archived, the media might have died in the interval.

 

[Helvellyn]

You are quite right. I ought to have checked... from a leaflet titled "GEC Transportation Projects Class 91 25kV BoBo locomotives for British Rail"...


It also says 'The locomotives and their equipments have been designed for a 35 year life when averaging some 420,000 km each year.' I presume this is from 1988.

GEC Class 91 Locomotives

What I like is the what was presumably the original leaflet. The blunt end drawing gives me 'Western' vibes, whilst the extra bodyside grills suggest GEC originally perceived cooling all that electrical equipment might be an issue (and of course GNER had extra grills fitted when the 91/1 refurbishment happened). I still think LNER missed a trick when they started showing the IC225 label on information screens to differentiate from Azuma operated services and should have gone for Electra instead.

 

[mike57]

Both floppies irretrievably corrupted...

With software there is a risk that the source code has vanished

It brings to mind a few events I have witnessed over the years, as soon as you get into 'legacy' software your problems multiply.

Again power cut related the supplying company were doing a firmware upgrade on some industrial control kit. Fine day, summer, last thing anyone was expecting was a power cut. 5 minutes in to a half hour download and upgrade power fails. CPU turned into a 'door stop' as there was no documented recovery method for the partly updated memory image. They were eventually fixed but it was a difficult process that took more than a few weeks.

Also the potential for loss of source code etc. In another case there had been a serious disagreement between senior managers after a take over, and the departing engineering manager removed the battery from a PLC system, which controlled a large section of the plant on his departure, making some NSFW comments, it was before the days of wide spread use of non volatile memory. All listings had been shredded, EM had obviously pre planned, they were shreded by incoming 'doc control' who had no idea what they were shredding and he didn't stop them, and the floppies were unreadable, which I suspect was as a result of his efforts. Took a long time to sort out.

These incidents just highlight additional problems that you face when dealing with legacy software systems over and above electro mechanical systems. Its another layer of complexity, cost and delay.

 

[D365]

Also the potential for loss of source code etc. In another case there had been a serious disagreement between senior managers after a take over, and the departing engineering manager removed the battery from a PLC system, which controlled a large section of the plant on his departure, making some NSFW comments, it was before the days of wide spread use of non volatile memory. All listings had been shredded, EM had obviously pre planned, they were shreded by incoming 'doc control' who had no idea what they were shredding and he didn't stop them, and the floppies were unreadable, which I suspect was as a result of his efforts. Took a long time to sort out.

Off-topic but surely, depending on what year this took place in, this would fall under the auspices of intentional corruption or deletion of electronic files.

 

[mike57]

Off-topic but surely, depending on what year this took place in, this would fall under the auspices of intentional corruption or deletion of electronic files

Not sure, it was around 1991. In any case the shenanigins surrounding the event meant it was quietly covered up, no one wanted any spotlight on it.

I was an independant contractor brought in after the event to recover the system. As a measure of my faith in the people involved I insisted on daily payment, as in at the end of each day... No money, no me the next day. I did get the system going again, by rewriting the code from scratch. It did prove lucrative for me as the day rate I quoted was an 'I dont really want the job' rate, but no one else was interested either. It took me 2 weeks to get the system working again which was pretty much what I expected, I just sat with the various production line people and said "Right what did this bit do". Write some code, test it, move on to the next bit. To be fair workface people were fine and were helpful, they could see their jobs on the line if the system couldn't be got working again. As for the management, I didn't trust them as far as I could throw them.

 

[Recessio]

Not quite electronics related, but I remember that Networker Turbos had an incredibly long waiting list for repairs/spares to the toilets due to the failure of a supplier of some parts. The more modern the train, I guess the less "off-the-shelf" the component becomes, and harder to simply just manufacture a replacement from scratch yourself.

 

[edwin_m]

Not quite electronics related, but I remember that Networker Turbos had an incredibly long waiting list for repairs/spares to the toilets due to the failure of a supplier of some parts. The more modern the train, I guess the less "off-the-shelf" the component becomes, and harder to simply just manufacture a replacement from scratch yourself.

3D printing does offer a solution for some mechanical components - to over-simplify, an existing one can be scanned and replicated. No use for electronics though...

 

[eldomtom2]

We have already reached (long passed?) this point with signalling, have we not?

With signalling I believe there's a fair bit of preservation of equipment in operating condition but not actually controlling anything. So obviously in those cases the safety concerns with signalling equipment that actually controls stuff don't apply.

 

[Belperpete]

I would also like to know why you think a modern diesel generator and electric motor would be more expensive and difficult to maintain than a steam engine.

The castings and patterns used on a diesel are an order or more complex than anything used on a steam engine. If you don't have access to the original patterns, then you are going to struggle to replicate components with sufficient accuracy. The travails of the group restoring Kerr Stuart 4415 are a case in point.

== Doublepost prevention - post automatically merged: 12 Jan 2026 ==

With signalling I believe there's a fair bit of preservation of equipment in operating condition but not actually controlling anything. So obviously in those cases the safety concerns with signalling equipment that actually controls stuff don't apply.

Most of the electronics used in pre-1990s signalling scheme was for non-vital functions, such as remote control and train describers. These have generally all been replaced by more modern systems as they became obsolete. The underlying interlocking used relays, that are still available.

From the late 1980s onwards, Solid State Interlocking became the defacto signalling system used for most mainline resignalling in the UK. It was designed using military specification components, which tend to have a long availability. Even so, as components became unavailable, replacement modules were designed, that were largely backwards compatible with the original. However, SSI development effectively ended with privatisation.

The two major UK signalling suppliers now produce their own proprietary interlocking systems, Smartlock and Westlock. However, these can both be loaded with SSI code, and used to replace old SSI interlockings, one modern interlocking taking the place of multiple existing SSIs. For example, the old SSIs that used to control the signalling out of Paddington and Liverpool Street were replaced with Smartlocks as part of the Elizabeth line work.

 

[The exile]

With signalling I believe there's a fair bit of preservation of equipment in operating condition but not actually controlling anything. So obviously in those cases the safety concerns with signalling equipment that actually controls stuff don't apply.

I was thinking more of the mainline - where most electronically controlled areas have either had or are due a second “resignalling” due to obsolescence while Wircester and Shrewsbury soldier on.