Following the Bedford incident, should TPWS - or something providing an equivalent level of protection - be fitted to all remaining unfitted signals?

[DadimF]

(Lengthy, excellent explanation snipped)

And that pretty much sums up the challenges. There’s no “one size fits all” panacea in terms of how you implement it. Each case has to be looked at on its own merits. Design, installation and testing resources are at a premium, if you can even find anyone willing to take on relay-based design: that is becoming a problem.

Me, I love designing on relay interlockings, E10k specifically. Hate black box systems as they are boring!

But to the topic in question relay interlockings are very safe, and although they can be challenging even for experienced designers in very complex areas they can interface with many different modern systems. For example I've designed a relay based GSP (Ground switch panel, located in a small case line side which controls points and signals that usually lead into a siding or depot, and can only be used if the signaller has given a release via the main signalbox panel) that interfaced with the ERTMS system in Wales.

The main issue I see as a designer is that the nature of the railways, especially since privatisation (which happened when I was still in primary school I think!) has created mini monopolies in regard to certain systems which in my opinion is why our railways cost so much to maintain or upgrade. I really think the new GBR should design their own systems that are very cheap and can be retrofitted to lines. The initial cost of R&D may be high but much cheaper than trying to implement ERTMS everywhere.

My thinking is based on the example of Ukraine. The cost of a cruise missile or even conventional unmanned drones are astronomical. Ukraine has adapted by creating super cheap but very effective drones in their thousands. The technology at the creation of AWS and TPWS is totally inferior to the technology of today, there are many ways, with a bit of thinking outside the box that could leverage modern technology to create an overlay safety system that interfaces with existing relay ones for a considerably cheaper cost if there was a will to do it. Unfortunately I don't think there is.

 

[Harpo]

Are there any potential on-train adaptations to AWS that could extend its effectiveness beyond cancellation of the horn?

 

[DadimF]

Are there any potential on-train adaptations to AWS that could extend its effectiveness beyond cancellation of the horn?

As Annette's key described above, it's just a simple system that works with magnets. I'm only familiar with the on track side of AWS so can't answer your question, but trying to think outside the box, I do wonder if it's possible to leverage the AWS somehow to act similar to TPWS OSS loops. Slightly alter the system so it's un-cancelable if it encounters a magnet placed at a certain point that hasn't energised within a certain time. No idea if that's possible though.

 

[Harpo]

As Annette's key described above, it's just a simple system that works with magnets. I'm only familiar with the on track side of AWS so can't answer your question, but trying to think outside the box, I do wonder if it's possible to leverage the AWS somehow to act similar to TPWS OSS loops. Slightly alter the system so it's un-cancelable if it encounters a magnet placed at a certain point that hasn't energised within a certain time. No idea if that's possible though.

Given that AWS is a screen indication in modern vehicles I wondered if the AWS visual could flash between horn cancellation and the next bell and even give a warning sound when power is applied in that state?

 

[Annetts key]

If it was simple and did not have such a high cost, we would likely not be having this discussion.

As I have said previously, the most effective and universal system is ETCS/ERTMS because it's an existing system where both the on-train equipment and the track/trackside already exist.

Although the only ATP system in use in this country is obsolete (no longer manufactured), if a lower cost system simply to provide information on the signal aspect in the cab is wanted with a automatic brake application if a signal is passed at danger/red, some of the ideas and some of the equipment from the GWML ATP system could be used.

The best method to get information from the existing signalling system is by using current transformers like those used by the GWML ATP system. Then a computer based module can transmit this data to an aerial mounted in the four foot of the track.

A computer on the train can then receive the data via another aerial and display it in the cab, and interface to the train braking system.

The problem is that this would be another bespoke system and it would take time and money to develop. And then there would need to be extensive testing. Followed by the approval process. And I just don't see this happening.

We still need to wait for the RAIB to report what they have found rather than jumping to conclusions too early.

 

[DadimF]

If it was simple and did not have such a high cost, we would likely not be having this discussion.

As I have said previously, the most effective and universal system is ETCS/ERTMS because it's an existing system where both the on-train equipment and the track/trackside already exist.

Although the only ATP system in use in this country is obsolete (no longer manufactured), if a lower cost system simply to provide information on the signal aspect in the cab is wanted with a automatic brake application if a signal is passed at danger/red, some of the ideas and some of the equipment from the GWML ATP system could be used.

The best method to get information from the existing signalling system is by using current transformers like those used by the GWML ATP system. Then a computer based module can transmit this data to an aerial mounted in the four foot of the track.

A computer on the train can then receive the data via another aerial and display it in the cab, and interface to the train braking system.

The problem is that this would be another bespoke system and it would take time and money to develop. And then there would need to be extensive testing. Followed by the approval process. And I just don't see this happening.

We still need to wait for the RAIB to report what they have found rather than jumping to conclusions too early.

Just to be clear my writings are intended to be more general than specifically around the consequences of the awful incident. But indeed, it is expensive and not so simple, though in my opinion that's mainly the case because of the way we think about solutions in the first place. You only have to look at the many examples of infrastructure or other major projects that balloon in costs in the UK. Yes the circumstances of why differ, but it comes down to very poor planning, imagination and a will to facilitate the projects around certain legal obstacles. Plus having to rely on external companies for products and the costs they impose.

I agree with you about the initial time and costs of R&D, but I think with a bit of clever thinking we could come up with an effective and cheap overlay system using existing equipment that is already PADS approved and configure it to work in a different way. Keep it internal to NR/GBR and roll it out on any project that alters existing layouts in legacy areas. Now the argument that would probably kill the idea is the risk benefit/cost as it always does but I think it's worth exploring at least even if it does turn out to be unfeasible.

 

[MarkyT]

If it was simple and did not have such a high cost, we would likely not be having this discussion.

As I have said previously, the most effective and universal system is ETCS/ERTMS because it's an existing system where both the on-train equipment and the track/trackside already exist.

It is the only game in town now. When the GW pilot scheme was underway I was working at Reading signalling design office. We all did a short introductory course about the tech and the design processes, but I personally never got to work on the system. The project manager admitted they knew before starting that both the Selcab and ACEC systems were already obsolete with ETCS on the horizon, but BR had to be seen as doing something, and ETCS wasn't a mature product yet to purchase off the shelf.

Although the only ATP system in use in this country is obsolete (no longer manufactured), if a lower cost system simply to provide information on the signal aspect in the cab is wanted with a automatic brake application if a signal is passed at danger/red, some of the ideas and some of the equipment from the GWML ATP system could be used.

The best method to get information from the existing signalling system is by using current transformers like those used by the GWML ATP system. Then a computer based module can transmit this data to an aerial mounted in the four foot of the track.

I agree. It can allow the system to be overlaid without any changes to the underlying interlocking and control centre systems, only the addition of the current transformers on the outgoing cable links. It's compatible with any interlocking technology, theoretically even mechanical. By contrast, a recent processor-based interlocking built 'ETCS ready' will need significant changes to interface to radio block centres etc. for level 2, and, if the signals are to be removed, that's complete reengineering of the trackside.

A computer on the train can then receive the data via another aerial and display it in the cab, and interface to the train braking system.

The problem is that this would be another bespoke system and it would take time and money to develop. And then there would need to be extensive testing. Followed by the approval process. And I just don't see this happening.

Polish Kombud Group have developed a Level 1 Limited Supervision solution for planned rollout across the Polish regional network on lines that don't justify Level 2. That's most of the network away from the busiest and fastest intercity corridors.

Signalling: Low-cost ETCS to support regional roll-out - Railway Gazette International

Kombud Group is developing a range of ETCS products to provide cost-effective interoperability on regional lines. Benjámin Zelki reports from the manufacturer’s newly opened plant in Radom.Polish signalling equipment supplier Kombud Group has been developing a range of products that meet ETCS...

www.railwaygazette.com

Cost-effective approach
First developed in Switzerland, ETCS Level 1 LS provides an intermittent ATP overlay for conventional lineside signalling, using interoperable ETCS components that would be upgradeable to support Level 1 Full Supervision or Level 2 if necessary. Although it sits in the background and reacts to overspeed events, Level 1 LS still supports full interoperability using ETCS-compliant onboard units. According to Kombud, regional lines do not need the full onboard functionality nor GSM-R data connectivity, let alone digital interlocking systems but thanks to Level 1 LS will become fully interoperable.

In terms of the lineside equipment, Level 1 LS is expected to be significantly cheaper to install, not least because of some simplified data engineering to avoid the need for any changes to the interlockings or the operating rules. In particular, Kombud favours the use of a standardised library of balise telegrams, which would only have to be approved once before being rolled out across multiple applications. Another aim is to bring together multiple lineside train detection devices and allow up to eight signals to be supervised by a single lineside equipment unit. The company has also proposed using short sections of Level 0 to simplify the interfaces between the Limited and Full Supervision applications.

Kombud Group believes that Level 1 LS would have a significantly lower life-cycle cost and be much simpler to commission than either Level 1 FS or Level 2. The company’s ETCS Specialist Karol Gruszka estimates that Level 1 FS would cost twice as much as Level 1 LS. To equip the same tracks with Level 2 it would cost another 30 to 40%, but on top of that it would require full GSM-R coverage and digital interlocking systems, whereas Level 1 LS can work with any, even mechanical signals, thus reducing the cost of interoperability even further.

 

[MadMac]

Given that AWS is a screen indication in modern vehicles I wondered if the AWS visual could flash between horn cancellation and the next bell and even give a warning sound when power is applied in that state?

The excellent BBC Horizon Rail Crash has a section on BRAWS which was being researched at the time (1972): the concept was that instead of merely cancelling, a warning was acknowledged by confirming the signal aspect shown.

 

[MarkyT]

The excellent BBC Horizon Rail Crash has a section on BRAWS which was being researched at the time (1972): the concept was that instead of merely cancelling, a warning was acknowledged by confirming the signal aspect shown.

It was SRAWS wasn't it, standing for 'Southern Region' or 'Signal Repeating'? BR(SR) had long been resistant to deploying standard AWS due to the repeated yellow risks. There was a pilot installation of SRAWS on the LSWR main line, but the concept was abandoned soon after and the SR was forced to start installing AWS instead, the last region to launch a major fitment programme.

 

[HSTEd]

I cant see any system not based on ETCS equipment getting installed now, even if the implementation is not a 'full' ETCS Level 1 implementation.

 

[zwk500]

I cant see any system not based on ETCS equipment getting installed now, even if the implementation is not a 'full' ETCS Level 1 implementation.

Well, no new system. It will be advantageous to the rollout of ETCS to avoid having lots of 'islands' because that means lots of transitions between ETCS and non-ETCS.
So a line that doesnt justify L2 won't be converted to L1 (even LS) unless its one of the later lines in an area to be converted.

So TPWS/AWS will likely see some new installations for at least the next 2 control periods

 

[MadMac]

It was SRAWS wasn't it, standing for 'Southern Region' or 'Signal Repeating'? BR(SR) had long been resistant to deploying standard AWS due to the repeated yellow risks. There was a pilot installation of SRAWS on the LSWR main line, but the concept was abandoned soon after and the SR was forced to start installing AWS instead, the last region to launch a major fitment programme.

On reflection, I think you’re right! Also known as the “Wiggly Wire”.

 

[HSTEd]

Well, no new system. It will be advantageous to the rollout of ETCS to avoid having lots of 'islands' because that means lots of transitions between ETCS and non-ETCS.
So a line that doesnt justify L2 won't be converted to L1 (even LS) unless its one of the later lines in an area to be converted.

We could probably build a system that appears like existing train protection systems from an operational and training perspective (except it will trip in more situations) whilst also being a subset of ETCS. It could be possible to develop a drop in that only uses compliant ETCS hardware and that can thus be deployed quite quickly.

That would probably ease the transition to ETCS2 rather than hinder it.

Continuing forever without modern train protection in the hope that we might get universal ETCS Level 2 could end up being a misjudgement on the scale of continuing with steam in the hope of getting universal electrification.

TPWS and AWS will only get harder to support with time.

EDIT: As mentioned above in the Polish example, we don't really care how a datagram was produced so long as the train does not interpret it as allowing it to do anything dangerous. After all, ballises are available that interpet volt-free contacts to select datagrams from a small selection of available ones.

 

[zwk500]

We could probably build a system that appears like existing train protection systems from an operational and training perspective (except it will trip in more situations) whilst also being a subset of ETCS. It could be possible to develop a drop in that only uses compliant ETCS hardware and that can thus be deployed quite quickly.

There was a proposal for 'TPWS-E' replacing the grids with eurobalises. That would likely be the option chosen.

That would probably ease the transition to ETCS2 rather than hinder it.

I expect it'd not make any difference given the neednfor the Interlocking to be compatible with an RBC for ETCS L2.

Continuing forever without modern train protection in the hope that we might get universal ETCS Level 2 could end up being a misjudgement on the scale of continuing with steam in the hope of getting universal electrification.

I agree. I wish NR were less pessimistic about Level 1, although they do appear to have lessened their stance from 'no L1' to 'L1 at depot entry' so maybe there's room for them to accept L1 only on regional lines that don't need the benefits of L2.

Interestingly, Ireland have chosen L1 only and no L2 on cost grounds.

TPWS and AWS will only get harder to support with time.

Yes, although there's quite a considerable amount of existing installations that continuing the supply chain is going to be cost-effective for some time.

If/when ETCS becomes cheaper to install than conventional signalling, then we qill start to see serious talk about adaption/replacement of AWS/TPWS.

 

[MarkyT]

There was a proposal for 'TPWS-E' replacing the grids with eurobalises. That would likely be the option chosen.

When TPWS+ was being developed, Siemens proposed a rival Level 1 LS implementation that used balises instead of the additional OSS grids. Like WCML TASS, this had to be registered with the European rail agency as a permitted custom packet 44 system. A test was carried out on the GWML at Slough with equipment fitted to an HST power car and some balises on the trackside. The additional equipment required on every train killed the idea pretty quickly.

I expect it'd not make any difference given the neednfor the Interlocking to be compatible with an RBC for ETCS L2.

Likely correct. Even for an existing processor-based system, there'll be many hardware and configuration data changes to make to support RBC connection, and reengineering of the radio infrastructure to support reliable continuous data transmission.

I agree. I wish NR were less pessimistic about Level 1, although they do appear to have lessened their stance from 'no L1' to 'L1 at depot entry' so maybe there's room for them to accept L1 only on regional lines that don't need the benefits of L2.

It's another technique to have in the toolkit. Most European railways have used level 1 to an extent, whether FS or LS. Being totally against it is an unusual stance.

Interestingly, Ireland have chosen L1 only and no L2 on cost grounds.

The approach can decouple line of route fitment from signalling renewal schemes. That is a benefit as dependency on the renewal going ahead is weakened, with L1 able to be wired up to older systems if necessary.

Yes, although there's quite a considerable amount of existing installations that continuing the supply chain is going to be cost-effective for some time.

And there are smaller, lighter-weight magnets available now, which can be carried by one person. Similar to a balise mount, they attach to a frame clamped to the rail foot rather than being screwed down to a sleeper.

 

[HSTEd]

Forgive me for being foolish, but if you were to use "transparent"/switchable balises, which monitor trackside signals to select the correct datagram, does that not also mean you could conceptually convert any existing interlocking to ETCS compatible cab signalling?

After all, the balises can send whatever datagram you want. You could prepare the relevant datagram for all signal aspects in advance and programme then.

Any signal that could show a red aspect would need at least a short length of euroloop so a train doesn't get stuck because it doesn't stop over the balise.
But transparent euroloop capable balises also exist in the Siemens catalogue so I assume they also exist elsewhere. Then you could just remove the signals themselves, perhaps replacing with dummy resistors for proving reasons.

Advantages could include elimination of sighting of signals as a consideration and a standardised driving interface - driver drives the Movement authority on their Driver interface, just as they would on an ETCS level 2 line.

I don't think, from first principles, it would be much more expensive than going for an ATP function on all signals. Obviously more equipment than simple limited supervision though.

Would be very strange if a mechanical lever frame ended up controlling cab signals though.

 

[Harpo]

Would be very strange if a mechanical lever frame ended up controlling cab signals though.

Not strange, just great British fudge. Like a fully ETCS L2 route where its only fringe is a Victorian mechanical box at a junction controlled by semaphores.

 

[andy33gmail]

It looks horribly like AWS led to sequence of events happening, had it not broke on the 810, wouldn't have been sitting there on the fast line.

Isn’t that a bit of a red herring? The leading train could have been caused to stop by a break hose coming off, misuse of a passenger alarm, the driver dropping his coffee on his lap, the pantographs getting tangled up in the overhead cables … it’s meant to always be safe to stop (minor injuries under emergency braking excepted)

If this line had been protected by TPWS what sort of speeds would the grids for the the junction have demanded, given that the signal was at yellow. I know this is a high speed mainline, but if you see a yellow you should be expecting a red at the next, so would TPWS slow you right down & keep you there until the next set of grids.

How do you pick the speed?

If you go too low, you affect performance / headrooms

Go too high, drivers may be complacent, use it as a target and come unstuck when a single leaf or snow crystal lands on the rails

Automatic Level crossings are an excellent example of how crudely designed systems work in practice; 19 seconds to wait for an express passenger train equates to a several minute delay for a stopping train or freight train

 

[zwk500]

The approach can decouple line of route fitment from signalling renewal schemes. That is a benefit as dependency on the renewal going ahead is weakened, with L1 able to be wired up to older systems if necessary.

It has another advantage for IÉ as their CAWS system uses coded track circuits and so to fulfill their electrification programme it would need modification or replacement. ETCS L1 is in the process of replacing their legacy ATP system on the DART area so upgrading from CAWS to L1 makes safety, operational and economic sense.

 

[MarkyT]

Forgive me for being foolish, but if you were to use "transparent"/switchable balises, which monitor trackside signals to select the correct datagram, does that not also mean you could conceptually convert any existing interlocking to ETCS compatible cab signalling?

After all, the balises can send whatever datagram you want. You could prepare the relevant datagram for all signal aspects in advance and programme then.

Yes.

Any signal that could show a red aspect would need at least a short length of euroloop so a train doesn't get stuck because it doesn't stop over the balise.
But transparent euroloop capable balises also exist in the Siemens catalogue so I assume they also exist elsewhere. Then you could just remove the signals themselves, perhaps replacing with dummy resistors for proving reasons.

The release speed concept covers this without loop infill. The approach movement authority supervises down to a low final approach speed; as long as the train doesn't exceed this again, the driver then has full control. The release speed is calculated such that a trainstop intervention on passing the signal/block marker at danger is guaranteed to stop the train within the overlap length. Release speed is an old concept incorporated in many older ATP systems, including the GWML pilot scheme. Just like GW-ATP, loop infill may still be worth it in selected locations for earlier better aspect updates. Radio infill for this purpose has also been used for some L1 FS/LS implementations on Alpine trunk routes in Italy.

Advantages could include elimination of sighting of signals as a consideration and a standardised driving interface - driver drives the Movement authority on their Driver interface, just as they would on an ETCS level 2 line.

I don't think, from first principles, it would be much more expensive than going for an ATP function on all signals. Obviously more equipment than simple limited supervision though.

Would be very strange if a mechanical lever frame ended up controlling cab signals though.

German PZB or Indusi, now a fairly sophisticated legacy LS system still widely used, was first used in a simpler form in the 1930s, and is able to be overlaid through colour light and semaphore areas, regardless of underlying interlocking technology. TPWS/AWS is similarly agnostic about interlocking and is deployed in many semaphore areas. I don't think it's strange at all. It's a positive feature that L1 systems can be interfaced to older installations where necessary, even if that isn't the norm.

It has another advantage for IÉ as their CAWS system uses coded track circuits and so to fulfill their electrification programme it would need modification or replacement. ETCS L1 is in the process of replacing their legacy ATP system on the DART area so upgrading from CAWS to L1 makes safety, operational and economic sense.

And it will be more compatible with long single lines, of which there are many in Ireland. The quantity of equipment required for track circuiting throughout these so trains can always receive the go code is hopelessly uneconomic. With L1 conveying the movement authority via balise and loop instead, they can use axle counters for train detection. The coded track circuits can be a maximum of 1200m long so for a notional 12km single line that would require 10 separate track circuits all with housings, power, cabling etc spread throughout the section. An axle counter by contrast would have a sensor at each end of the logical block section and an evaluator unit connected to both.

 

[Bread & cheese]

The Telegraph has reported the following:

A formal letter issued to Network Rail in January 2013 by the Rail Safety Standards Board (RSSB) said that to comply with modern safety standards, an "additional signal section" - meaning an extra signal light - needed to be installed on the "Up Fast approach to Bedford South Junction".
But Network Rail told the RSSB it opposed this, leading the safety body to conclude: "Such changes would be substantial and disproportionate to the benefits offered."

(Apologies if this is the wrong thread.)

https://www.telegraph.co.uk/news/2026/06/28/rail-chiefs-blocked-new-signals-near-fatal-crash-site/

 

[MarkyT]

The Telegraph has reported the following:


(Apologies if this is the wrong thread.)

https://www.telegraph.co.uk/news/2026/06/28/rail-chiefs-blocked-new-signals-near-fatal-crash-site/

It's a red herring, and the mention of an extra signal light is nonsense. The deviation was for a non-compliant approach locking lookback at several MML signals on the fasts. RSSB initially wanted it extended, but NR argued it would be difficult, the risks are low and that the conditional double red CDR arrangements being implemented control any new risk. RSSB accepted that and awarded the certificate of derogation.

Log in to keep reading

Register to the RSSB website and access more content.

www.rssb.co.uk

Note: and I gave my card number to them for a free trial to read that, and the inane comments of their readers... Must remember to cancel!

 

[dmncf]

I think I understand how AWS uses magnets to provide its functionality, but I don't understand why the developers of AWS (back in the 1950s?) decided this level of functionality was sufficient to meet the railway's needs.

Why were they satisfied with a system that doesn't differentiate between yellow and red signal aspects? Why wasn't it a priority for the system to automatically apply the brakes on a train that passed a red signal?

Happy to be pointed to an earlier post, as I imagine this may have been asked previously. Thanks.

 

[HSTEd]

Yes.

The release speed concept covers this without loop infill. The approach movement authority supervises down to a low final approach speed; as long as the train doesn't exceed this again, the driver then has full control. The release speed is calculated such that a trainstop intervention on passing the signal/block marker at danger is guaranteed to stop the train within the overlap length. Release speed is an old concept incorporated in many older ATP systems, including the GWML pilot scheme. Just like GW-ATP, loop infill may still be worth it in selected locations for earlier better aspect updates. Radio infill for this purpose has also been used for some L1 FS/LS implementations on Alpine trunk routes in Italy.

Using the overlap release speed would work in an ATP system where a colour light signal aspect is available to the driver.

In a full blown cab signalling arrangement I don't think there would be a way for the driver to know if they are authorised to proceed into the overlap. If they pass over the next balise before it clears they will get an emergency brake application and then be stranded with no connection to the signalling system as no balise will be under the trains antenna.

I think you'd need either a radio infill system or a euroloop to communicate with a stopped train. The latter seems much simpler to integrate into an existing interlocking.

I think you could have a transparent euroloop for every home signal, and a simple transparent eurobalise for every distant and banner repeater.

 

[45Fox]

I think I understand how AWS uses magnets to provide its functionality, but I don't understand why the developers of AWS (back in the 1950s?) decided this level of functionality was sufficient to meet the railway's needs.

Why were they satisfied with a system that doesn't differentiate between yellow and red signal aspects? Why wasn't it a priority for the system to automatically apply the brakes on a train that passed a red signal?

Happy to be pointed to an earlier post, as I imagine this may have been asked previously. Thanks.

At the time when AWS was being developed large parts of the British Railways network (I would say the majority) were controlled by mechanical signalling, the emphasis was on AWS being provided at distant signals with no provision at stop signals; this followed on largely the practice from previous systems, such as GWR ATC and the trial non-contact system (similar to BR AWS) of the LMS. The spread of multiple aspect signalling and replacement of semaphore stop signals with colour lights did lead to some compromises and inconsistencies which were never really addressed satisfactorily.

The inability to distinguish between caution and danger aspects led to the Southern Region of BR being reluctant to install AWS, particularly on the suburban lines around London which saw (still do, to be truthful) a lot of prolonged running on cautionary aspects with repetitive cancelling of the AWS horn, with the resultant increased chances of drivers constantly cancelling and it losing it's effectiveness as a warning.

A system to apply the brakes if a train passed a signal at danger was looked at by BR and it's predecessor companies at the same times as AWS/ATC were being developed; the only system in Britain which was comparable was the train-stops and tripcocks as used by the London Underground, it wasn't felt really practical for main line use and probably the cost of installing it at stop signals and on locomotives/units would have been seen as prohibitive.

 

[MarkyT]

Using the overlap release speed would work in an ATP system where a colour light signal aspect is available to the driver.

In a full blown cab signalling arrangement I don't think there would be a way for the driver to know if they are authorised to proceed into the overlap. If they pass over the next balise before it clears they will get an emergency brake application and then be stranded with no connection to the signalling system as no balise will be under the trains antenna.

The movement authority ends at a signal or block marker. The train is not ever authorised into the overlap beyond. That's just maintained as a locked contingency in case of overrun until the approaching train has stopped successfully where it is supposed to. If an overrun occurs, once it's safely stopped within the overlap, the recovery is to a restricted speed mode until the next balise update, after a conversation with the signaller. Adding a loop on approach is mainly done where justified to allow a MA extension (or aspect step up) to be signalled to an approach train earlier.

I think you'd need either a radio infill system or a euroloop to communicate with a stopped train. The latter seems much simpler to integrate into an existing interlocking.

I think you could have a transparent euroloop for every home signal, and a simple transparent eurobalise for every distant and banner repeater.

Italy uses distributed short-range radio infill on busy Alpine routes with L1 FS and LS. Such systems don't need to be so resilient, wide covering, continuous, and high capacity as the full data radio used by L2 for continuous MA updates to multiple trains, and if one local transmitter fails it just adds a little delay in a particular locality rather than perhaps a wide-ranging removal of movement authorities.

I'd bolt a small radio antenna on every signal post facing approaching traffic and make each signal its own micro RBC. Parallel signals on multitrack might share a transmitter Every train that could see the signal could communicate with it and it's likely the range would be better than visual aspect observation in many cases. Eventually, optical signal heads might be removed and the replacement block markers would become radio signals alone. All the MA updates might come through this local distributed radio system, encoded by LEUs from local signal states. The positioning balises might be passive in such a system.

 

[Annetts key]

I think I understand how AWS uses magnets to provide its functionality, but I don't understand why the developers of AWS (back in the 1950s?) decided this level of functionality was sufficient to meet the railway's needs.

Why were they satisfied with a system that doesn't differentiate between yellow and red signal aspects? Why wasn't it a priority for the system to automatically apply the brakes on a train that passed a red signal?

Happy to be pointed to an earlier post, as I imagine this may have been asked previously. Thanks.

When AWS was developed, train braking systems were not as good as modern braking systems (for passenger trains) and stopping distances were long. To stop at a signal at danger (a stop signal), it was absolutely vital that the distant signal showing caution was not missed (either a semaphore distant at caution or a colour light distant showing a yellow aspect). There had been accidents caused by drivers not braking after seeing or passing the distant signal with the distant at caution. By the time that they saw the stop signal at danger, they could not stop in time.

The original purpose of AWS was to therefore to (a) alert the driver that they were approaching a distant signal (very useful during bad visibility), give different sounds depending on if it was showing clear or caution and if the driver did not acknowledge a horn, apply the brakes.

Stop signals were not fitted with AWS.

A three or four aspect colour light combines the functions of a distant signal and a stop signal, hence were fitted with AWS. But you can't provide a bell at a red signal, so it has to be the horn.

There were various proposals and designs of an improved system to replace AWS but nothing came of any of them.

Using the overlap release speed would work in an ATP system where a colour light signal aspect is available to the driver.

In a full blown cab signalling arrangement I don't think there would be a way for the driver to know if they are authorised to proceed into the overlap. If they pass over the next balise before it clears they will get an emergency brake application and then be stranded with no connection to the signalling system as no balise will be under the trains antenna.

I think you'd need either a radio infill system or a euroloop to communicate with a stopped train. The latter seems much simpler to integrate into an existing interlocking.

I think you could have a transparent euroloop for every home signal, and a simple transparent eurobalise for every distant and banner repeater.

For the GWML ATP, the track mounted transmitting aerial is fitted AT the signal (as in, within one metre of it). If the train has not already received data from a previous aerial or infill loop with authority to pass this signal, the train should stop before the signal and hence before the track mounted transmitting aerial for that signal.

The train should never move onto the overlap beyond the signal unless it has received authority. For ATP, that means a proceed aspect.

For an ATP fitted train, it can move forward at the release speed (if the signal was at red, but has now cleared to a proceed aspect, but the ATP is not yet aware of this) so that it can move over the track mounted transmitting aerial AT the signal for the ATP to receive an update of the current signal aspect.

For a cab signalling system, that is, you have replaced the signal heads with block marker boards, you just need either an infill loop, other infill device (one or more extra track mounted transmitting aerials / eurobalises) or as @MarkyT says, a short range radio link. So that the train can receive updates so that the driver knows when a new movement authority has been received.

 

[Signal Head]

When AWS was developed, train braking systems were not as good as modern braking systems (for passenger trains) and stopping distances were long. To stop at a signal at danger (a stop signal), it was absolutely vital that the distant signal showing caution was not missed (either a semaphore distant at caution or a colour light distant showing a yellow aspect). There had been accidents caused by drivers not braking after seeing or passing the distant signal with the distant at caution. By the time that they saw the stop signal at danger, they could not stop in time.

Harrow & Wealdstone being probably the most (in)famous example.

There were various proposals and designs of an improved system to replace AWS but nothing came of any of them.

I vaguely recall some idea being bandied about which utilised a third inductor (this was well before TPWS).

I cannot remember what it would have achieved or how it did it, seeing as you've only got two magnetic poles to play with, South being the 'marker' (permanent) and North (electro) within a defined time limit after detecting South gives a bell.

South followed by nothing (within the time limit) gives a horn.

Once the detector (at least the original type) has 'seen' a South pole, it needs to be reset to North ready to detect the next South pole, and this was done by an internal coil energised when the Caution warning is acknowledged.

== Doublepost prevention - post automatically merged: 29 Jun 2026 ==

When AWS was developed, train braking systems were not as good as modern braking systems (for passenger trains) and stopping distances were long. To stop at a signal at danger (a stop signal), it was absolutely vital that the distant signal showing caution was not missed (either a semaphore distant at caution or a colour light distant showing a yellow aspect). There had been accidents caused by drivers not braking after seeing or passing the distant signal with the distant at caution. By the time that they saw the stop signal at danger, they could not stop in time.

Harrow & Wealdstone being probably the most (in)famous example.

There were various proposals and designs of an improved system to replace AWS but nothing came of any of them.

I vaguely recall some idea being bandied about which utilised a third inductor (this was well before TPWS).

I cannot remember what it would have achieved or how it did it, seeing as you've only got two magnetic poles to play with, South being the 'marker' (permanent) and North (electro) within a defined time limit after detecting South gives a bell.

South followed by nothing (within the time limit) gives a horn.

Once the detector (at least the original type) has 'seen' a South pole, it needs to be reset to North ready to detect the next South pole, and this was done by an internal coil energised when the Caution warning is acknowledged.

 

[MarkyT]

When AWS was developed, train braking systems were not as good as modern braking systems (for passenger trains) and stopping distances were long. To stop at a signal at danger (a stop signal), it was absolutely vital that the distant signal showing caution was not missed (either a semaphore distant at caution or a colour light distant showing a yellow aspect). There had been accidents caused by drivers not braking after seeing or passing the distant signal with the distant at caution. By the time that they saw the stop signal at danger, they could not stop in time.

Very much a welcome 'wake up call' at the critical point for any crew barreling along through the featureless countryside in poor visibility. Colour light distants and ATC/AWS systems also avoided fogging personnel being sent out to distant signals to place detonators for each train.

The original purpose of AWS was to therefore to (a) alert the driver that they were approaching a distant signal (very useful during bad visibility), give different sounds depending on if it was showing clear or caution and if the driver did not acknowledge a horn, apply the brakes.

This is a common theme for early warning systems on mainline networks. The Crocodile was invented in 1872 by engineers Lartigue and Forest and first employed by the French Nord railway company the same year. Conceptually, it is almost exactly the same as modern UK AWS, though Crocodile is not fail-safe. A switchable voltage is applied to the wiggly metal contact surface of a short 'ramp' in the 4ft ~200m before a distant signal. A + voltage relative to an adjacent rail indicates a restrictive aspect, while a - voltage signifies a proceed aspect. A wire brush under the vehicle makes contact with this and the onboard warnings and interventions are triggered by onboard circuits. Early implementations were warning only, activating the steam whistle, but connections to brake pipes were soon devised for automatic intervention if warnings weren't acknowledged.

The GWR ATC system of 1906 added a failsafe mechanical component to the same basic concept. The vehicle contact shoe is physically lifted by passing over the ramp. That movement will trigger a warning unless accompanied by trackside voltage detected on the ramp contact surface.

While not used, it's interesting that the GWR system potentially might have conveyed extra info by switching polarity like Crocodile, but it is hard for a system to process a red response where the intermittent transponder is before the signal, as you might get a red indication but the then signal steps up before the train physically reaches it. You'd need to add something else, delaying intervention for a measured distance or time perhaps, having a speed limit for a time but to be fully protected you'd then also need a trainstop ramp at the signal itself

The BR AWS precursor, The Hudd non-contact magnetic system, was in use with the LMS on the LT&S before WW2.

Other railways and railroads used very similar ideas. There was a widespread magnetic ATC system in the USA used mostly for distant warning. Some instances survive on the west coast I think, but most was allowed to be ripped out in the 1960s and 70s by the FRA when passenger services were being abandoned wholesale.

 

[HSTEd]

Italy uses distributed short-range radio infill on busy Alpine routes with L1 FS and LS. Such systems don't need to be so resilient, wide covering, continuous, and high capacity as the full data radio used by L2 for continuous MA updates to multiple trains, and if one local transmitter fails it just adds a little delay in a particular locality rather than perhaps a wide-ranging removal of movement authorities.

Do you have access to a info on the size of a radio infill unit? I can't seem to find a picture of one and they don't appear to be a prominent part of signal company catalogues.

I'm trying to work out of it fills a cabinet or it's the size of a mobile phone!