Throwaway signals

[MisterSheeps]

I went a walk today, that goes near the local line which has had some work done recently, and found a piece of paper in a field I was going to put in a nearby skip ... it was a 'signal information sheet' (the line has 4 aspect signals). It said the signal has no user serviceable parts, a 10 year life, the LEDs will last at least this period of time, if failure gives an alarm it should be replaced and returned for repair, and replaced after 10 years ...
(1) I thought electronics were rated as 'mean time between failures', so how can anyone specify a minimum life?
(2) Whatever happened to the idea that individual components should be replaceable? Binning the entire head (am assuming a signal in this context is the 4 aspect head) seems a nonsense.
(3) 10 years seems a short period of design life ...

 

[Annetts key]

The light output from LEDs gradually decreases over time. How long it takes depends on the LED colour, the intensity (power) that the LED is operated at and the technology used to make the LED.

Ever since LED signals were introduced, the manufacturers have specified a ten year life. As the manufacturer has no control over which aspect in a LED main aspect signal will be lit for most of its life, it’s not possible to derate it on duty factor grounds.

For example, some signals spend over 95% of their time showing red. While automatic signals on lightly used lines in the middle of nowhere could spend 95% of their time showing green. An automatic (or controlled signal in automatic mode) on the approach to a junction signal in contrast, could spend most of its time showing a yellow aspect.

Some designs of LED signals do have replaceable modules. But with other types, yes, the whole signal head would have to be replaced. Also keep in mind that most LED signal designs combine the red, (single/lower) yellow and green in one module/physical aspect/lens assembly.

Network Rail is working on the assumption that replacing a complete LED signal head every ten years is cheaper overall than using a conventional filament lamp head (that may last 30 years or more) and having to replace the filament lamps every year (or when the system indicates a failed filament). Note that these filament lamps are special types and cost a fortune compared to normal domestic lamps.

Edited to correct a letter missing off a word

 

[Ken H]

The light output from LEDs gradually decreases over time. How long it takes depends on the LED colour, the intensity (power) that the LED is operated at and the technology used to make the LED.

Ever since LED signals were introduced, the manufacturers have specified a ten year life. As the manufacturer has no control over which aspect in a LED main aspect signal will be lit for most of its life, it’s not possible to derate it on duty factor grounds.

For example, some signals spend over 95% of their time showing red. While automatic signals on lightly used lines in the middle of nowhere could spend 95% of their time showing green. An automatic (or controlled signal in automatic mode) on the approach to a junction signal in contrast, could spen most of its time showing a yellow aspect.

Some designs of LED signals do have replaceable modules. But with other types, yes, the whole signal head would have to be replaced. Also keep in mind that most LED signal designs combine the red, (single/lower) yellow and green in one module/physical aspect/lens assembly.

Network Rail is working on the assumption that replacing a complete LED signal head every ten years is cheaper overall than using a conventional filament lamp head (that may last 30 years or more) and having to replace the filament lamps every year (or when the system indicates a failed filament). Note that these filament lamps are special types and cost a fortune compared to normal domestic lamps.

But why buy a signal head where you have to replace a whole head rather than just the LED clusters. Thats just wrong IMHO.

 

[SargeNpton]

But why buy a signal head where you have to replace a whole head rather than just the LED clusters. Thats just wrong IMHO.

The way that technology advances, and the way that manufacturers catalogues change, it's likely that the individual components might not be in production in 10 years time.

By replacing the head as a complete unit you simply fit whatever is the current model rather than trying to source out of date components. Also allows you to swap suppliers rather than being beholden to the original manufacturer.

 

[MisterSheeps]

cheaper overall

thank you, kind of makes sense, send a team out every 10 years. the sheet does say replacement or return for assessment

just the LED clusters

yes, seems sensible, keep the physical head but replace the clusters. depends on the deterioration i guess, this area gets a lot of weather

 

[pdeaves]

The original post states that in the event of failure, the head should be returned for repair. So it isn't thrown away, it's repaired in a better environment than outdoors, hopefully meaning it lasts longer anyway. This is basic 'component exchange', the same as (say) a locomotive swapping bogies to get it back in traffic, then the bogies are repaired after to go in another loco.

 

[zwk500]

But why buy a signal head where you have to replace a whole head rather than just the LED clusters. Thats just wrong IMHO.

Worth pointing out that most LED signals nowadays can show multiple colours from the same position, so you have a single LED cluster that gives all 3 colours. Some 4-aspect signals now only have 2 aspect positions, with the Red green and single yellow from the lower cluster and double yellow using both illuminated.

 

[Snow1964]

Ever since LED signals were introduced, the manufacturers have specified a ten year life.

I cannot find any scientific reason for 10 years, (ie 10 years is more optimal than say 9 years 6 months, or 11 years), it just appears to be a round number that has been picked.

It works out at about 87,660 hours (depending on when leap years fall), could just have easily picked the slightly longer 100,000 hours if wanted a round number.

Personally I am not a fan of a future calendar date as an indicator of threshold of unacceptable degradation point, rather than a condition review date.

 

[Ken H]

Worth pointing out that most LED signals nowadays can show multiple colours from the same position, so you have a single LED cluster that gives all 3 colours. 4-aspect signals now only have 2 aspect positions, with the Red green and single yellow from the lower cluster and double yellow using both illuminated.

I assume the 2 clusters are the same. Its just the way the head is wired so the upper cluster can only show yellow.

 

[MisterSheeps]

4-aspect signals now only have 2 aspect positions

interestingly not this one, possibly as was a slot in replacement?

 

[zwk500]

interestingly not this one, possibly as was a slot in replacement?

May be different depending on manufacturer or back end interlocking control. I'm not a signal engineer so couldn't say for sure.

 

[Railsigns]

4-aspect signals now only have 2 aspect positions

Some have two apertures and some have four. It depends on the manufacturer and style of signal.

 

[zwk500]

Some have two apertures and some have four. It depends on the manufacturer and style of signal.

I did acknowledge this earlier, but have now amended my post to clarify

 

[edwin_m]

I cannot find any scientific reason for 10 years, (ie 10 years is more optimal than say 9 years 6 months, or 11 years), it just appears to be a round number that has been picked.

It works out at about 87,660 hours (depending on when leap years fall), could just have easily picked the slightly longer 100,000 hours if wanted a round number.

Personally I am not a fan of a future calendar date as an indicator of threshold of unacceptable degradation point, rather than a condition review date.

It's probably just a figure that sounded about right and was put in a specification.

 

[DelW]

(3) 10 years seems a short period of design life ...

I cannot find any scientific reason for 10 years, (ie 10 years is more optimal than say 9 years 6 months, or 11 years), it just appears to be a round number that has been picked.

Isn't this effectively the manufacturer's warranty period, similar to the way a car manufacturer will offer (say) a six year warranty, but the car would be expected to last considerably longer.

Presumably if the signal head fails within the ten years the makers will repair it, beyond that it's the customer's responsibility (i.e. Network Rail) to decide whether to repair it (if still possible) or replace it.

 

[HSTEd]

(1) I thought electronics were rated as 'mean time between failures', so how can anyone specify a minimum life?

You can design an electronic system for a reasonable degree of assurance that it will last this period of time.

(2) Whatever happened to the idea that individual components should be replaceable? Binning the entire head (am assuming a signal in this context is the 4 aspect head) seems a nonsense.

Replacing the components by the trackside in the rain is hardly conducive to doing the job properly or avoiding mistakes.
And every time you make and break the seals you wear them and increase the chance it will start leaking - a unit sealed at the factory does not have this problem.

 

[MisterSheeps]

Replacing the components by the trackside

i was thinking more of a modular light unit slide in & out (4 in this case) without the requirement to bin the whole head ... if the assembly has a design life of 10 years the frame should last much longer. So maybe you change the two usually yellow ones more often than the usually red or green ones ... but from what @Annetts key wrote, it makes sense to just do the lot and sort the rest out back in the factory.

 

[John Webb]

I assume the 2 clusters are the same. Its just the way the head is wired so the upper cluster can only show yellow.

The upper (double yellow) cluster we have on display at St Albans South appears to be yellow only - or else it's simply only wired to illuminate the yellow LEDs - there's only two wires coming out of the cluster unit. (It's a Dorman signal, by the way.)

 

[AlexNL]

Dorman sell two variants: clusters which can show red+yellow+green, and clusters which can only show yellow.

The positioning in a 2 aperture frame is such that the RYG cluster will always be closest to the driver's eye. Normally that it will be on the bottom, but for ground placed co-acting signals this will be reversed.

 

[Annetts key]

So, I’m not sure what the official policy is. But here is a photo that shows what sometimes happens.



Now does that look like removed signal heads (in this case LED position light junction indicators, often called feathers) are being carefully stored ready for transportation back to the manufacturer?

 

[Signal Head]

thank you, kind of makes sense, send a team out every 10 years. the sheet does say replacement or return for assessment


yes, seems sensible, keep the physical head but replace the clusters. depends on the deterioration i guess, this area gets a lot of weather

You can only do that with certain types. In the original Dorman (now known as 'Classic'), and the later Mk 2 (or CLS Lite), the case is basically an empty shell into which the LED modules are fitted. Moduke replacement on site is fairly simple. The Classic type is wired on to links, so requires cable removal and retermination, the Mk 2 uses the same modules as the 'iLS' range (plastic lollipops!), which are plug coupled.

The iLS type - I'm not sure if they are intended to be opened up by the end user, as far as I know they are supplied fitted to the post, the plug couplers are at the bottom.

For other manufacturers - VMS (who appear to have morphed into an outfit called Mallatite), supply a conventional-looking head, albeit with no hoods, with separate apertures for each colour and replaceable modules, and as an alternative, a 'flat panel' type, which as far as I know are sealed and plug-coupled externally, and are treated as an indivisible unit. The flat panel type (and possibly the multi- aperture type, I've never seen an installed one up close) include monitoring contacts which, when they are used to replace filament heads, are typically cut into the 'first filament' alarm circuit which advises the signalman of a problem and triggers a technician visit.

The Dorman varieties in common use do not include any monitoring connections, but my understanding is that the modules are self proving and will reduce their current draw below the lamp proving relay drop away value if a fault likely to degrade light output is detected, ie the signal may not be a complete failure, but the interlocking is told it is 'out'.

 

[AlexNL]

VMS as a company no longer exists, it was wound down and sold off in parts.

 

[Annetts key]

The Dorman type are available in lamp providing types and some types may be available in non-lamp proving types (e.g. GPL types).

The lamp proving type draws a similar current to an equivalent filament lamp signal, hence normal lamp providing relays, or SSI TFM (solid state interlocking trackside functional modules) lamp proving current return paths can be used.

We’ve had one Dorman LED signal module where the lamp proving circuitry failed, the appropriate lamp proving relay no longer had enough current flowing to energise it (contacts broken), hence the interlocking treated it as unlit, even though it looked normal brightness on site.

With the Dorman LED signals, any first filament proving circuitry is rewired to ‘bypass’ the wiring to the signal. This is included in the design modifications.

 

[Signal Head]

VMS as a company no longer exists, it was wound down and sold off in parts.

Thanks. I hadn't heard about it until a few months ago when I tried to go to their website and got redirected.

Their kit never seemed to gain the same 'traction' as Dorman's, and I am aware of signal sighting chairs who refuse to accept the flat panel type, although there have been some fairly large scale renewals with them recently, spotted loads of them in at Crewe and Manchester Piccadilly when I passed through recently.

Images of 'VMS' flat type, and conventional individual aspect module attached.

 

[Ashley Hill]

@Signal Head Is there any particular reason these aspects have silly patterns in them?

 

[MisterSheeps]

VMS (who appear to have morphed into an outfit called Mallatite

Thank you. Yes, the information sheet was for this new outfit. I hadn't realised there were still UK based manufacturers.

treated as an indivisible unit.

and this type ... I am surprised NR do not specify a more recyclable type, but the context of this is that a 10 year may see cab signalling (WCML in Cumbria).

 

[MadMac]

Thanks. I hadn't heard about it until a few months ago when I tried to go to their website and got redirected.

Their kit never seemed to gain the same 'traction' as Dorman's, and I am aware of signal sighting chairs who refuse to accept the flat panel type, although there have been some fairly large scale renewals with them recently, spotted loads of them in at Crewe and Manchester Piccadilly when I passed through recently.

Images of 'VMS' flat type, and conventional individual aspect module attached.

The lesser spotted GEC Geographical set in abundance there!