Cable theft delays to reduce?

[harz99]

Currently travelling on a GC service northbound which left Kings Cross 27 late due to the inbound train being 63 late following ECML delays this morning. Signal regulation in the York area pushed the delay to currently 37 minutes.

I got to wondering if, when the ECML signalling is fully digital, cable theft will cease to have the effects it currently does?

 

[swt_passenger]

Currently travelling on a GC service northbound which left Kings Cross 27 late due to the inbound train being 63 late following ECML delays this morning. Signal regulation in the York area pushed the delay to currently 37 minutes.

I got to wondering if, when the RCML signalling is fully digital, cable theft will cease to have the effects it currently does?

Electrical power supplies will still be required for operating points, so cable theft can still be a factor. Shouldn‘t be so much though.

 

[Annetts key]

Local 'tail' cables from the lineside equipment cabinet (location case) to the track mounted equipment (point operating equipment, track circuits axle counters, AWS, TPWS, etc.) or to conventional colour light signals are still required. And these all require power. The power being fed via long lengths of lineside power cable.

Also data link cables are still needed for communication between the lineside equipment cabinet (location case) and the central interlocking.

Alas, many of the cable vandals cut the cable to see if they think its worth anything before deciding to try to steal any. By which time, the damage is done.

 

[harz99]

Thank you both, that's very depressing, hopefully the eventual lack of very much profitable cable to thieve will improve matters though.

 

[Annetts key]

The BTP and Network Rail were working together to ensure that scrap metal dealers were aware of what typical railway cables look like and the markings on them. And all new lineside signalling (including signalling power) and telecommunications (including fibre optic) cables are clearly and permanently marked with "Network Rail" and the railway cable specification number.

This makes it rather more difficult to make money from stealing railway cables.

 

[duffield]

I've said this before, but it's my opinion that stealing bits of live national infrastructure (railway or non-railway) should be treated much more harshly than stealing (say) a reel of currently unused cable. Due to the massive expense, disruption and economic damage caused, there should be a specific offence of "sabotage of critical infrastructure", maybe even treated in a similar manner to a lower end terrorist offence.

This wouldn't suddenly make such criminals honest but might at least steer them away from actually cutting live cables and towards other less damaging targets.

 

[Elecman]

The BTP and Network Rail were working together to ensure that scrap metal dealers were aware of what typical railway cables look like and the markings on them. And all new lineside signalling (including signalling power) and telecommunications (including fibre optic) cables are clearly and permanently marked with "Network Rail" and the railway cable specification number.

This makes it rather more difficult to make money from stealing railway cables.

but the scrotes will just strip the cable to bare cores ( more valuable than cables with insulation still on but time consuming)

 

[ricoblade]

Happened again between Doncaster and Sheffield, which has scuppered my travel plans for today, anyone got any more details?

 

[LMS 4F]

Happened again between Doncaster and Sheffield, which has scuppered my travel plans for today, anyone got any more details?

An XC train severed the cable that had been laid across the track at about 11pm Saturday night leading to long delay for that train and a lot passengers for Doncaster being taken onward from a Wakefield Westgate by taxi.This after a reversal and then diversion via Moorethorpe.
Apparently where it occurred is a hot spot for cable theft and detection measures are in place.

 

[YorkshireBear]

It's been a bad couple of weeks in South/West Yorkshire.

 

[Deepgreen]

The BTP and Network Rail were working together to ensure that scrap metal dealers were aware of what typical railway cables look like and the markings on them. And all new lineside signalling (including signalling power) and telecommunications (including fibre optic) cables are clearly and permanently marked with "Network Rail" and the railway cable specification number.

This makes it rather more difficult to make money from stealing railway cables.

I don't think the unscrupulous dealers/receivers will care how their cable is marked. The coating is removed, and probably burned, anyway. The actual copper remains untraceable.

An XC train severed the cable that had been laid across the track at about 11pm Saturday night leading to long delay for that train and a lot passengers for Doncaster being taken onward from a Wakefield Westgate by taxi.This after a reversal and then diversion via Moorethorpe.
Apparently where it occurred is a hot spot for cable theft and detection measures are in place.

Was the cable laid across the line to sever it deliberately (i.e. by the thieves) or by accident/left behind?

 

[LMS 4F]

I don't think the unscrupulous dealers/receivers will care how their cable is marked. The coating is removed, and probably burned, anyway. The actual copper remains untraceable.


Was the cable laid across the line to sever it deliberately (i.e. by the thieves) or by accident/left behind?

I don’t think anyone except the thieves can answer that for sure but it seems more than likely.
There was another case on Monday 24 February west of Castleford that caused cancellations of the TPE service from York to Piccadilly via Castleford for part of the morning. I don’t have any idea of the MO in that incident.

 

[Annetts key]

Cable is often dragged over the rails so that the train wheels cut it. Saves the idiots having to cut the cable themselves. These cables are heavy. The thieves only want to drag off the bits they can actually pull.

Yes, if you remove the copper cores from the cable, the identification on and in the outer cable sheath will no longer help. But it takes a lot of effort to drag away the cables and a lot of effort to strip the cables. Burning it off is easier, but that generates lots of black smoke. And an unscrupulous dealer is not going to give the thieves the going price, they will pay out less than the normal rate.

 

[ricoblade]

Done again at Swinton earlier. No trains between Doncaster and Sheffield this morning.

 

[Pigeon]

Local 'tail' cables from the lineside equipment cabinet (location case) to the track mounted equipment (point operating equipment, track circuits axle counters, AWS, TPWS, etc.) or to conventional colour light signals are still required. And these all require power. The power being fed via long lengths of lineside power cable.

Most of it only requires kitten power, though - indeed a good deal of it was originally designed to run off big dry cells that someone would come along and change every so often, and the less often that needed to be the better. This makes it well suited to be powered by those solar/wind/battery units that seem to crop up every couple of hundred metres these days powering something or other (though the unit might need to be of what would be considered disproportionately large capacity in non-signalling terms). Much the same ought to apply to LED colour light signals, though I fear we may already have got locked into a design pattern of extravagance with those.

Point motors I'm not sure about; the power consumption is I would guess quite high but the duty cycle is very low, so it's possible that the average would still work out to a usefully low figure. But in any case, there are a lot fewer of them.

Also data link cables are still needed for communication between the lineside equipment cabinet (location case) and the central interlocking.

Fibre optics.

Alas, many of the cable vandals cut the cable to see if they think its worth anything before deciding to try to steal any. By which time, the damage is done.

They do that because at present some are and some aren't so they don't know in advance. If the cables were all fibre optic the word would soon get round not to bother.

 

[Annetts key]

Most of it only requires kitten power, though - indeed a good deal of it was originally designed to run off big dry cells that someone would come along and change every so often, and the less often that needed to be the better. This makes it well suited to be powered by those solar/wind/battery units that seem to crop up every couple of hundred metres these days powering something or other (though the unit might need to be of what would be considered disproportionately large capacity in non-signalling terms). Much the same ought to apply to LED colour light signals, though I fear we may already have got locked into a design pattern of extravagance with those.

Only the type of signalling equipment typically used in mechanical signal box areas ran from "dry" cells (some equipment used "wet" cells, including lead acid or Ni-Cad that were trickle charged from a local 240V AC supply). This was specificity designed to be as low power as possible. There often being no mains power available to use.

All the equipment designed for use in power box / panel box areas or bigger schemes is designed to be powered from a 650V AC single phase power distribution system. In the equipment room, relay room, REB (relocatable equipment building) or location case/cabinet/cupboard, there are one or more transformers to step this down to either 110V AC or to 50V DC.

For AWS equipment, it's stepped down to 24V DC. A standard strength AWS electro-magnet consumes between 9.6W to 12W depending on the type. A standard strength suppressor magnet consumes between 37W and 48W depending on the type (although these are normally only active when there is a train in the relevant track circuit section).

Track circuit equipment has it's own power supply units (PSU) that run off the 110V AC supply. I don't know the power figures for all types. But the older design of Aster U type / SF15 and the older design of TI21 both used PSUs with an output rated at 24V 4A. That's the maximum rated output. So that's up to 96W. But one PSU could feed two track circuits (although in practice, it's more complex than this).

Some designs of track circuit use AC for the rail circuit. My favourite was the Western Region Quick Release type. The 20V 2A AC output could rather upset welders as it would give them shocks as they used their metal ruler and tools to check the alignment as they prepared to weld up the two rail ends

It would also create an pretty display as it heated up swarf in an IRJ (insulated rail joint) where the T-piece was broken or missing.

Relay interlocking or signal control uses 50V DC relays. A neutral line relay typically has a coil of 1000 ohms. So at 50V, that's 2.5W per relay. Some relay types had a higher power rating.

The type of LED signals that are installed in place of conventional filament lamp signals have to consume the same amount of power as the conventional filament lamp signals in order for the lamp lit proving system to work as designed. This includes when the signal is controlled by computerised interlockings that use SSI TFM (solid state interlocking trackside functional modules). The conventional filament lamps used in signals are/were type SL35 (12V 24W/24W) plus add in the power lost to the transformers in the signal head.

SSI TFM run off the 110V AC supply. I don't know the power rating.

Now, of course, there are lower power LED signals available and other lower powered equipment is likely now available.

Point motors I'm not sure about; the power consumption is I would guess quite high but the duty cycle is very low, so it's possible that the average would still work out to a usefully low figure. But in any case, there are a lot fewer of them.

A HW type point machine takes around 110V DC or 120V DC at a current of around 7A to 10A (starting current/point fouled current). That's about 1.1kW. The running current is typically 4A to 5A approximately (550W). But, with conventional interlocking design (at least, the types I have worked on), ALL the points that are required to move, are all commanded to move ALL at the same time. The time of operation is however limited to a nominal 7.5 seconds per point).

There are other types of point machine, each with it's own power requirements. But they all run off either a 110V DC, 120V or 140V DC supply ( not including older types as used in otherwise mechanical signal box areas).

Older practice in some areas was to provide a centralised lead acid battery to power points at a junction. But modern practice varies. Some may be powered from local Cyclon cell banks (a type of sealed gel type lead acid) or powered via a 650V AC to 140V DC transformer/rectifier unit.

Fibre optics.

Maybe, but all the datalinks (for both computer based interlocking and for axle counter systems) I have worked on have used copper cable. CCTV systems these days use fibre optic cable and the railway telecom "trunk" networks use fibre optic (and have done in some areas for over forty years).

They do that because at present some are and some aren't so they don't know in advance. If the cables were all fibre optic the word would soon get round not to bother.

In my experience, most of these cable thieves are not exactly the sharpest tools in the toolbox. So it may take rather longer for them to work this out than you may think. After all, some have tried cutting live 11kV cables with a hacksaw before.

 

[HSTEd]

Fibre optics.

I remember seeing a slide deck from some Italian engineers about attempts to use GSM-R to control trackside equipment to eliminate data cabling requirements entirely.

I don't know if it ever went anywhere though.

As for power, about all that can be done is switching to aluminium cables which should at least reduce the value of metal that can be obtained - hopefully reducing incentives.

ETCS will help by removing the need to supply signals and the like, but the train detection equipment will still need power.

I guess if we eventually get ETCS Hybrid or similar systems (that do not require large train detection systems trackside) we can reduce the amount of powered equipment to the point that continuous power cables could be dispensed with.

Given the relatively high population density of England, it might be possible to simply take supplies as and when required over significant distances.

 

[Annetts key]

As for power, about all that can be done is switching to aluminium cables which should at least reduce the value of metal that can be obtained - hopefully reducing incentives.

Or the railway can go back to burying cables like they used to. In all my years attending cut, vandalised or stolen cable reports, no cables had been dug up by thieves. But cables running on the surface in the cess or in SCT (surface concrete troughing) routes were attacked.

 

[LMS 4F]

If the thieves aren’t able to get a price for the metal they won’t bother. It must be going to scrap metal dealers and that is where the attention needs to be, there are a lot less dealers than there are vulnerable places on the network.
40 years ago in the Police Force where I worked all the dealers were visited regularly by our Task Force officers. This stopped many years ago but perhaps this is where BTP should consider making an effort.
There are various regulations that Dealers are supposed to follow and enforcement of these would surely have a positive impact.

 

[HSTEd]

If the thieves aren’t able to get a price for the metal they won’t bother. It must be going to scrap metal dealers and that is where the attention needs to be, there are a lot less dealers than there are vulnerable places on the network.
40 years ago in the Police Force where I worked all the dealers were visited regularly by our Task Force officers. This stopped many years ago but perhaps this is where BTP should consider making an effort.
There are various regulations that Dealers are supposed to follow and enforcement of these would surely have a positive impact.

I think a lot of the low hanging fruit there has been picked.

We no longer have people trying to sell scrap dealers lengths of intact cable.
You are more likely to see people trying to sell copper ingots - and they are sufficiently small that it is probably worth moving them out of the country somewhat clandestinely.

 

[LMS 4F]

I saw an item on a news bulletin a few days ago, can’t remember which, that theft of the cables on electric vehicle points were being stolen in I believe the West Midlands.
Perhaps the increase in the number of these will take the pressure off the railway network

 

[Annetts key]

I saw an item on a news bulletin a few days ago, can’t remember which, that theft of the cables on electric vehicle points were being stolen in I believe the West Midlands.
Perhaps the increase in the number of these will take the pressure off the railway network

I saw that, which means it was either on the BBC or Channel 4 News. I think the former.

 

[Pigeon]

Only the type of signalling equipment typically used in mechanical signal box areas ran from "dry" cells (some equipment used "wet" cells, including lead acid or Ni-Cad that were trickle charged from a local 240V AC supply). This was specificity designed to be as low power as possible. There often being no mains power available to use.

All the equipment designed for use in power box / panel box areas or bigger schemes is designed to be powered from a 650V AC single phase power distribution system. In the equipment room, relay room, REB (relocatable equipment building) or location case/cabinet/cupboard, there are one or more transformers to step this down to either 110V AC or to 50V DC.

For AWS equipment, it's stepped down to 24V DC. A standard strength AWS electro-magnet consumes between 9.6W to 12W depending on the type. A standard strength suppressor magnet consumes between 37W and 48W depending on the type (although these are normally only active when there is a train in the relevant track circuit section).

Track circuit equipment has it's own power supply units (PSU) that run off the 110V AC supply. I don't know the power figures for all types. But the older design of Aster U type / SF15 and the older design of TI21 both used PSUs with an output rated at 24V 4A. That's the maximum rated output. So that's up to 96W. But one PSU could feed two track circuits (although in practice, it's more complex than this).

Some designs of track circuit use AC for the rail circuit. My favourite was the Western Region Quick Release type. The 20V 2A AC output could rather upset welders as it would give them shocks as they used their metal ruler and tools to check the alignment as they prepared to weld up the two rail ends

It would also create an pretty display as it heated up swarf in an IRJ (insulated rail joint) where the T-piece was broken or missing.

Relay interlocking or signal control uses 50V DC relays. A neutral line relay typically has a coil of 1000 ohms. So at 50V, that's 2.5W per relay. Some relay types had a higher power rating.

The type of LED signals that are installed in place of conventional filament lamp signals have to consume the same amount of power as the conventional filament lamp signals in order for the lamp lit proving system to work as designed. This includes when the signal is controlled by computerised interlockings that use SSI TFM (solid state interlocking trackside functional modules). The conventional filament lamps used in signals are/were type SL35 (12V 24W/24W) plus add in the power lost to the transformers in the signal head.

SSI TFM run off the 110V AC supply. I don't know the power rating.

Now, of course, there are lower power LED signals available and other lower powered equipment is likely now available.

Thanks for the comprehensive reply - all very interesting stuff.

It does more or less seem that once mains power feeds became commonplace, the signalling equipment designers became more inclined to take a "use plenty, there's lots of it" attitude to power consumption, for minor locations out in the sticks as well as for denser centres where long cable runs are not needed. (Some of the old battery equipment was marvellous stuff, all electromechanical and achieving wonders with currents down to the milliamp range in some cases.) Also that the "use plenty" attitude has now become more or less unavoidable due to new kit being designed to work the same way as the existing kit.

Now to me phrases like "all new digital signalling" suggest (a) a complete renewal of everything so there is no existing kit to work the same way as, and (b) that the new kit would be fundamentally incompatible with any existing kit anyway, so you're starting with a totally clean slate and are free to design the new kit how you want without any reference to any existing designs. This would mean that if you wanted to design it as a collection of minimum-power-consumption nodes each with their own local power source and interconnected only by fibre optics, there would be nothing to get in the way of you doing that.

The type of LED signals that are installed in place of conventional filament lamp signals have to consume the same amount of power as the conventional filament lamp signals in order for the lamp lit proving system to work as designed.

Ow, ow, ow, ow, my brain hurts... People buying overpriced power resistors on ebay to hook across their car lights after their new LED bulbs made their silly car complain about them is one thing, but I would have hoped that the railway would actually alter their dud lamp detector to take account of the difference in lamps, instead of bodging the lamp to fool it.

Are we still talking about the gadget with nothing but a rather neat arrangement of transformers and relays to perform the detection that operates in a manner that only makes sense with loads resembling an incandescent bulb, or have those been superseded now?

A HW type point machine takes around 110V DC or 120V DC at a current of around 7A to 10A (starting current/point fouled current). That's about 1.1kW. The running current is typically 4A to 5A approximately (550W). But, with conventional interlocking design (at least, the types I have worked on), ALL the points that are required to move, are all commanded to move ALL at the same time. The time of operation is however limited to a nominal 7.5 seconds per point).

Funny, that's pretty much what I had guessed as a maximum consumption.

It doesn't much matter if all the points are commanded to move at the same time (I guess this makes the interlocking design more general) as long as the total current isn't too much for the battery. It's the integral of consumption compared with the integral of charge, over a period of at least some days, that matters.

There are other types of point machine, each with it's own power requirements. But they all run off either a 110V DC, 120V or 140V DC supply ( not including older types as used in otherwise mechanical signal box areas).

Older practice in some areas was to provide a centralised lead acid battery to power points at a junction. But modern practice varies. Some may be powered from local Cyclon cell banks (a type of sealed gel type lead acid) or powered via a 650V AC to 140V DC transformer/rectifier unit.

Oh well, if you could do it with an accumulator without a solar/wind charging unit, you definitely ought to be able to do it with one

Maybe, but all the datalinks (for both computer based interlocking and for axle counter systems) I have worked on have used copper cable. CCTV systems these days use fibre optic cable and the railway telecom "trunk" networks use fibre optic (and have done in some areas for over forty years).

See above re. starting from scratch; they don't have to use copper, it's just that it's simpler to stick with the way everything else does it. And the distances on railways are definitely into the range where fibre optics are much easier to transmit data over than copper.

In my experience, most of these cable thieves are not exactly the sharpest tools in the toolbox. So it may take rather longer for them to work this out than you may think. After all, some have tried cutting live 11kV cables with a hacksaw before.

I know the type well Thick as two short planks except in the sole area of knowing what is and what isn't worth nicking. I think it would spread reasonably well, although of course you'll never get to 100%.

 

[MarkyT]

It does more or less seem that once mains power feeds became commonplace, the signalling equipment designers became more inclined to take a "use plenty, there's lots of it" attitude to power consumption,

Definitely. In the 1960s and 70s, modern colour light signalling was often referred to as power signalling. The prospect of endless nuclear power 'too cheap to meter', but still subject to the laws of physics as far as line losses and voltage drop are concerned.

Also that the "use plenty" attitude has now become more or less unavoidable due to new kit being designed to work the same way as the existing kit.

The railway wanted LED units that would be a like for like replacement for incandescent lamps. That could allow fast widespread rollout without design changes to circuits. When more comprehensive changes are being made, alternative products with lower consumption and a separate voltage free contact for lamp proving can be specified.

It doesn't much matter if all the points are commanded to move at the same time (I guess this makes the interlocking design more general) as long as the total current isn't too much for the battery. It's the integral of consumption compared with the integral of charge, over a period of at least some days, that matters.

Some large interlockings do have special controls to prevent too many points moving simultaneously. I've not seen a real example myself but I've seen reference to it in standard circuit books and descriptions.

Oh well, if you could do it with an accumulator without a solar/wind charging unit, you definitely ought to be able to do it with one.

Point batteries were fairly standard in relay days but fell out of favour in early SSI days, late 80s when the favoured concept was to design the 650VAC feeder with sufficient capacity for all possible instantaneous demands. That was when new architectures for power were also pioneered with duplicated diverse feeds usually on opposite sides of the line, eventually with remote or automatic switchover, the idea being no one cable cut should disable any equipment. A similar approach was applied to the data link cables.

See above re. starting from scratch; they don't have to use copper, it's just that it's simpler to stick with the way everything else does it. And the distances on railways are definitely into the range where fibre optics are much easier to transmit data over than copper.

Belgium adopted SSI in the 90s. They used a different object controller design that had fibre optics for the data link rather than the twisted pairs of the UK implementation.

 

[Annetts key]

It does more or less seem that once mains power feeds became commonplace, the signalling equipment designers became more inclined to take a "use plenty, there's lots of it" attitude to power consumption

It’s a combination of reasons.

The earlier relays were either the type used in telegraph, later, telephone systems (known within the railway as PO - for Post Office) types. Or large, heavy “shelf type” relays. Both are more sensitive than the later standardised “miniature” BR930 series (also known as Q type after the designation used by Westinghouse) or the larger P type. Both of these were specifically designed to fit to a base which in turn was mounted on a metal frame/rack. Hence they have a standard case, standard “plugboard” and relays to the same specification and type but from different manufacturers are interchangeable (or are supposed to be).

The design of most BR930 series / Q type / P type are to make them as failsafe as possible (including return springs in most types to ensure the contacts move to the de-energised position when the coil circuit is powered down disconnected). But you can’t get something for nothing, so their efficiency is less.

The problem with the PO type, was that they had many failure modes, so the railway discontinued their use for safety critical functions. Which in practice was any relay used for interlocking, track circuit, point or signal control. Some were used in some designs for use on non-vital functions. Such as repeating switch or button contacts (by this I mean switches and buttons operated by the signaller).

The shelf type were unsurprisingly intended to be placed on shelves. The biggest disadvantage with these was no overall standard (each manufacturer did their own thing) and each wire terminating on a binding post (a threaded stud with oversized 0BA nuts and washers). When they failed, generally they failed “safe side”. That is, no contacts welded and the contact assembly rested in the de-energised position. Unless someone lifted them up and turned them upside down.

As the limitations were found out over time, the railway amended the specifications.

This applies to all signalling equipment that is classed as vital (safety critical).

Once you have robust and reliable equipment that meets the specifications and requirements, the relative cost of designing and proving that a new design of equipment both meets the specifications and requirements, is reliable and is cost effective becomes more difficult. Especially if the demand for this new design is not likely to be high. Also a lack of money for complete resignalling schemes doesn’t help. Not all resignalling schemes are actually complete resignalling schemes. To keep costs down, some will reuse or retain existing equipment, so a mix of old and new.

And new designs of equipment often have to work with older designs. That’s the reason why “lamp proved” LED signals have to consume the same power as filament lamp signals. Both for compatibility and to meet the standard that requires that all main aspects have lamp proving provided.

This is one of the reasons why it takes a long time for new designs to be adopted for vital signalling systems.

Another reason is that signalling designers for schemes (not equipment manufacturers) work to standard design rules. So there is a set design for producing a standard point control circuit. The original BR Western Region free wired point control circuit uses one less relay compared to the later “standardised” point control circuit for example.

In comparison, the railway moves much quicker for non-vital (non-safety critical) signalling equipment. Modern TDM (time division multiplex) remote control (communication) equipment is faster, quicker and uses a fraction of the power of older generation systems. Plus takes up far less space. Similarly, it’s the same with TD (train describer) systems.

 

[MarkyT]

It’s a combination of reasons.

The earlier relays were either the type used in telegraph, later, telephone systems (known within the railway as PO - for Post Office) types. Or large, heavy “shelf type” relays. Both are more sensitive than the later standardised “miniature” BR930 series (also known as Q type after the designation used by Westinghouse) or the larger P type. Both of these were specifically designed to fit to a base which in turn was mounted on a metal frame/rack. Hence they have a standard case, standard “plugboard” and relays to the same specification and type but from different manufacturers are interchangeable (or are supposed to be).

The design of most BR930 series / Q type / P type are to make them as failsafe as possible. But you can’t get something for nothing So efficiency is less.

The problem with the PO type, was that they had many failure modes, so the railway discontinued their use for safety critical functions. Which in practice was any relay used for interlocking, track circuit, point or signal control.

They were prone to contact welding in certain scenarios. Still used in non-vital circuits for indications, telemetry followers etc. They're typically faster acting than safety relays.

Safety relays including shelf types and later plug in models have special silver to silver impregnated carbon contacts. These are specially formulated to resist welding. Shelf types have a heavy metal armature that will always return by gravity to normal position, while the more lightly constructed plug ins have a strong spring return action.

The shelf type were unsurprisingly intended to be placed on shelves. The biggest disadvantage with these was no overall standard (each manufacturer did their own thing) and each wire terminating on a binding post (a threaded stud with oversized 0BA nuts and washers). When they failed, generally they failed “safe side”. That is, no contacts welded and the contact assembly rested in the de-energised position. Unless someone lifted them up and turned them upside down.

The number of turns on the twin coils of a typical traditional1000 ohm relay is very large. That's how so much magnetic force can be generated by a small current.

 

[Pigeon]

Once you have robust and reliable equipment that meets the specifications and requirements, the relative cost of designing and proving that a new design of equipment both meets the specifications and requirements, is reliable and is cost effective becomes more difficult. Especially if the demand for this new design is not likely to be high. Also a lack of money for complete resignalling schemes doesn't help. Not all resignalling schemes are actually complete resignalling schemes. To keep costs down, some will reuse or retain existing equipment, so a mix of old and new.

And new designs of equipment often have to work with older designs. That's the reason why "lamp proved" LED signals have to consume the same power as filament lamp signals. Both for compatibility and to meet the standard that requires that all main aspects have lamp proving provided.

I think you've slightly missed my point here, which is that making the lamp proving work by making the lamp pretend to be something it isn't is a bodge. I've seen the circuit diagram of what was said to be "the standard" apparatus for lamp proving and switching to the backup filament, and as I remember it's a very neat piece of design which depends on the lamp having the inherent characteristics of a filament lamp for it to work - not just raw power consumption, but also being a bidirectional resistive load which can be relied upon to fail open circuit when it fails.

LED lamps have none of these characteristics inherently, and adding extra bits and pieces to make them simulate them actually does not evade the requirement for "proving dependability of new designs" which would also, and obviously, arise from replacing the lamp proving apparatus itself with a new design specifically designed to work with LED lamps. What it's doing logically is moving most of the lamp proving functionality into the lamp itself - in effect the lamp proving is now being done by the extra bits and pieces, with the output signal being encoded in the form of the simulated characteristics of a filament lamp, which the existing lamp proving apparatus then translates into a form that the rest of the signalling system understands. Therefore the dependability of the extra bits and pieces in the lamp needs to be established just as well as the dependability of a new LED-oriented design of lamp proving apparatus, in the form of a separate unit to replace the existing apparatus, would - after all in logical terms they are doing the same thing.

It's easy to assume that the method does avoid the need to check the dependability of a new lamp-proving design because in the superficial view you don't appear to have one, and perhaps this is what has been assumed, but in logical terms you do still have one embedded in the correct operation of the filament-lamp-simulating gubbins. Also, you have messed up the separation of functionality into independent logical units - instead of the lamp just lighting up and the proving apparatus just proving it, you now have the lamp itself sharing the function of proving; so the function is kind of smeared over two different units, which is widely recognised as a common cause of subtle and obscure bugs.

 

[Annetts key]

I think you've slightly missed my point here, which is that making the lamp proving work by making the lamp pretend to be something it isn't is a bodge. I've seen the circuit diagram of what was said to be "the standard" apparatus for lamp proving and switching to the backup filament, and as I remember it's a very neat piece of design which depends on the lamp having the inherent characteristics of a filament lamp for it to work - not just raw power consumption, but also being a bidirectional resistive load which can be relied upon to fail open circuit when it fails.

LED lamps have none of these characteristics inherently, and adding extra bits and pieces to make them simulate them actually does not evade the requirement for "proving dependability of new designs" which would also, and obviously, arise from replacing the lamp proving apparatus itself with a new design specifically designed to work with LED lamps. What it's doing logically is moving most of the lamp proving functionality into the lamp itself - in effect the lamp proving is now being done by the extra bits and pieces, with the output signal being encoded in the form of the simulated characteristics of a filament lamp, which the existing lamp proving apparatus then translates into a form that the rest of the signalling system understands. Therefore the dependability of the extra bits and pieces in the lamp needs to be established just as well as the dependability of a new LED-oriented design of lamp proving apparatus, in the form of a separate unit to replace the existing apparatus, would - after all in logical terms they are doing the same thing.

It's easy to assume that the method does avoid the need to check the dependability of a new lamp-proving design because in the superficial view you don't appear to have one, and perhaps this is what has been assumed, but in logical terms you do still have one embedded in the correct operation of the filament-lamp-simulating gubbins. Also, you have messed up the separation of functionality into independent logical units - instead of the lamp just lighting up and the proving apparatus just proving it, you now have the lamp itself sharing the function of proving; so the function is kind of smeared over two different units, which is widely recognised as a common cause of subtle and obscure bugs.

Um, for filament lamps, as in SL35 lamps used in main aspects, there are a number of considerations.

A SL35 lamp has a 12V 24W main filament rated at 1000 hours. The long life version has a 12V 24W main filament rated at 8000 hours.
Both the standard version and the long life version have a 12V 24W auxiliary filament rated at 1000 hours.

In the signal head there is a transformer. The primary is a 110V, but with various taps so that it can be adjusted. The secondary is a 14V and 12V winding, but again with various taps so that it can be adjusted.

Also in the signal head is a BR903 relay. On Western we call this relay the G(M)ECR, elsewhere it's known either just as ECR or by other designations. The coil is wired in series with the main filament of the lamp and normally supplied from the 14V tap on the transformer. The normally closed (or "back") contact is wired to the auxiliary filament and the 12V tap on the transformer. The normally open (or "front") contact is wired in a daisy chain (relay interlocking areas) to the first filament failed (FFF) circuit (called G(M)ESR on Western) or to an input on a SSI TFM. This is a non-safety circuit that ONLY provides an indication to either the signaller or the local S&T technicians (not both).

Hence if the main filament is okay, the auxiliary filament is not switched in. And no FFF is shown in the signal box or indicated to the S&T.

If the main filament blows, the auxiliary filament is automatically switched in. A FFF is shown in the signal box or indicated to the S&T along with an an audible alarm.

For LED signals, if there is was a G(M)ESR or equivalent circuit, it's either bypassed so not including the LED signal or removed completely.

A completely separate system is used for lamp proving. This is the "vital" safety lamp proving system.

For relay interlocking, in the location case/cupboard/relay room/etc. a BR941A specification relay (called the GECR on Western, I think it's the same elsewhere, but I can't remember) has it's coil wired in series from the 110V AC supply and the relay controlling contacts, and this is used to supply the signal aspect that should be illuminated. Hence if the lamp fails completely (both filaments) or the cable is cut, the GECR will de-energise. A normally open (or "front") contact is wired to control the signal control relay of the signal in rear, hence if the GECR is not energised, that signal will be held at red. A second normally open (or "front") contact is wired to the indication circuit to the signaller (controlled signals and certain other signals only).

For a SSI TFM, this has a designated "proved current return" terminal for signals to enable the module to detect the current that the signal is drawing. The computer based interlocking performs the same functions as a relay interlocking in controlling the signal rear and providing data for the signallers indication.

And yes, LEDs operated rather differently to filament lamps.

Just to be clear, I'm not giving my opinion, I'm just derscribing the practice used by the railway on systems that I worked on.

If you want to know more about the BR Spec 941A lamp proving relay, here's a link to the current main relay manufacturers page on them.

 

[MarkyT]

When LEDs became available. RT/NR wanted to change as many filament lamps as possible to the new tech to save on the huge maintenance effort of incandescents.

The vast majority were retrofitted to existing systems. Making the new products compatible with the old lamp supply and proving circuit was the easiest way to avoid circuit design changes at each installation, or at least minimise those changes to a minor tweak to the filament changeover alarm functionality which was no longer required.

Typically, filament changeover alarms are grouped for a logical set of signals at a particular junction, so an individual change of head where filaments remained in other signals on the same circuit would simply bridge out the affected contact that no longer existed.

Conceptually, I think the internal lamp proving of older approved LED units switches in the integral shunt resistor to simulate the filament load, so there's little danger of the resistor current falsely reporting illumination.

Low wattage versions of LED heads are also available now where the lamp proving is reported via a voltage free contact instead for use in new systems that don't rely on current proving.

 

[Pigeon]

Um, for filament lamps, as in SL35 lamps used in main aspects, there are a number of considerations.

A SL35 lamp has a 12V 24W main filament rated at 1000 hours. The long life version has a 12V 24W main filament rated at 8000 hours.
Both the standard version and the long life version have a 12V 24W auxiliary filament rated at 1000 hours.

In the signal head there is a transformer. The primary is a 110V, but with various taps so that it can be adjusted. The secondary is a 14V and 12V winding, but again with various taps so that it can be adjusted.

Also in the signal head is a BR903 relay. On Western we call this relay the G(M)ECR, elsewhere it's known either just as ECR or by other designations. The coil is wired in series with the main filament of the lamp and normally supplied from the 14V tap on the transformer. The normally closed (or "back") contact is wired to the auxiliary filament and the 12V tap on the transformer. The normally open (or "front") contact is wired in a daisy chain (relay interlocking areas) to the first filament failed (FFF) circuit (called G(M)ESR on Western) or to an input on a SSI TFM. This is a non-safety circuit that ONLY provides an indication to either the signaller or the local S&T technicians (not both).

Hence if the main filament is okay, the auxiliary filament is not switched in. And no FFF is shown in the signal box or indicated to the S&T.

If the main filament blows, the auxiliary filament is automatically switched in. A FFF is shown in the signal box or indicated to the S&T along with an an audible alarm.

For LED signals, if there is was a G(M)ESR or equivalent circuit, it's either bypassed so not including the LED signal or removed completely.

A completely separate system is used for lamp proving. This is the "vital" safety lamp proving system.

For relay interlocking, in the location case/cupboard/relay room/etc. a BR941A specification relay (called the GECR on Western, I think it's the same elsewhere, but I can't remember) has it's coil wired in series from the 110V AC supply and the relay controlling contacts, and this is used to supply the signal aspect that should be illuminated. Hence if the lamp fails completely (both filaments) or the cable is cut, the GECR will de-energise. A normally open (or "front") contact is wired to control the signal control relay of the signal in rear, hence if the GECR is not energised, that signal will be held at red. A second normally open (or "front") contact is wired to the indication circuit to the signaller (controlled signals and certain other signals only).

For a SSI TFM, this has a designated "proved current return" terminal for signals to enable the module to detect the current that the signal is drawing. The computer based interlocking performs the same functions as a relay interlocking in controlling the signal rear and providing data for the signallers indication.

And yes, LEDs operated rather differently to filament lamps.

Just to be clear, I'm not giving my opinion, I'm just derscribing the practice used by the railway on systems that I worked on.

If you want to know more about the BR Spec 941A lamp proving relay, here's a link to the current main relay manufacturers page on them.

Again, thanks for the interesting and detailed reply. I don't think the thing I saw a circuit diagram of was quite the same, but it was similar. However, the precise details don't matter - the important part is that these devices are designed to detect a filament lamp, which either takes a given current when it's working, or takes no current when it fails.

A fake-it-so-the-LED-looks-like-a-filament-lamp device not only has to draw the same large current when it's working, which is easy enough to arrange, but it also has to stop drawing current under failure conditions, which is more complicated. A failed LED does not reliably fail open circuit: quite often they fail short circuit, or turn themselves into non-light-emitting resistors. Moreover, since an LED needs a constant-current drive (or at least a reasonable approximation thereto), having one fail short-circuit barely affects the surrounding circuitry, quite unlike the potentially serious effects of a voltage-driven device going short-circuit. Then, if the individual LEDs in the array are connected in a series/parallel string arrangement - which for an array the size of a signal head is very likely - any failure, either way, of one LED in one string is probably going to be partially masked by the presence of the intact strings and so will cause even less electrical disturbance to act as a detectable indication of failure. Ideally what you need to do is monitor the actual light output of each LED rather than its consumption, for example with an LED that has an integrated monitor photodiode in the same way that many laser diodes do.

Also, how do you define "failure" anyway? A single lamp can be either on or off, but an LED signal head has lots of lamps. One LED out in a signal head, or a few LEDs out, and it's unlikely a driver will even notice. Several LEDs out and they may well notice but they still won't have any difficulty seeing the signal at the normal distance. You'd have to lose quite a large fraction of the total before sighting was seriously affected. So you have to decide how to set the threshold number of failed LEDs that gives a failure indication.

All this is to say that the failure indication is not something that happens automatically and naturally as with a filament bulb blowing, but something that needs to be derived and synthesised by additional circuitry monitoring the LEDs. Therefore the additional circuitry has become part of the lamp proving apparatus: it has to detect a failure condition and interrupt the current as reliably as a filament lamp interrupts the current when it blows, otherwise the failure condition may go unreported. And therefore it needs this reliability to be proved; it does not dodge the need for validation of a new device.