Filament signals with LED replacement bulbs

[ewsclass]

Saw at Woking some of the conventional filament signal heads were retrofitted with LED lamps (could tell by the instantaneous snapping between aspects, ie no fade as the filament cools). The route indications remain filament however. Is this a simple drop in replacement? Surely the LEDs would trip the current sensing for lamp proving? Would the SSI have to be reprogrammed to account for the change in current?

 

[pompeyfan]

Saw at Woking some of the conventional filament signal heads were retrofitted with LED lamps (could tell by the instantaneous snapping between aspects, ie no fade as the filament cools). The route indications remain filament however. Is this a simple drop in replacement? Surely the LEDs would trip the current sensing for lamp proving? Would the SSI have to be reprogrammed to account for the change in current?

Woking has had its 4 aspect filaments replaced with Dorman heads. This is slowly being rolled out across most of the Wessex route.

As far as I can tell they’re just standard Dorman 2 lens heads.

There was a few signals in the Farncombe box area where they had 3 lens, but the actual lens was led. They were unusual.

 

[Randomer]

Prefaced by saying I'm not professionally involved with railway signalling.

Although I can say that in similar safety critical applications I am familiar with the lamp itself can be fitted with a method of drawing current (usually circuitry that creates heat and a heat sink) which fits into the same size envelope as the original lamp. Although you do not get the efficiency advantage of going to LED these systems are normally much longer lasting than filament lamps with resulting savings in maintenance and staffing costs.

As the previous poster has pointed out it is much more common to switch out the whole lamp housing which normally allows space for control circuitry to prove the LED is lit and then send current to the proving circuit to show when the "lamp" has failed (again not in the railway industry to be clear). In the use cases I'm familiar with it actually tends to be the control circuitry that is the complication that fails rather than the relatively simple circuitry of the LED itself.

 

[swt_passenger]

I think this came up a while back, and it turned out the LED “Light engines” did have additional stuff that made them appear to the rest of the signalling as a normal filament lamp.

 

[Bald Rick]

I think this came up a while back, and it turned out the LED “Light engines” did have additional stuff that made them appear to the rest of the signalling as a normal filament lamp.

That’s correct. It’s the same with LED level crossing lights.

 

[Philip Phlopp]

That’s correct. It’s the same with LED level crossing lights.

The heatsink was added or moved around so as to heat the front of the fitting, as I recall, when it became clear snow wasn't melting off them in the same way as it would with a filament bulb.

 

[ewsclass]

Woking has had its 4 aspect filaments replaced with Dorman heads. This is slowly being rolled out across most of the Wessex route.

As far as I can tell they’re just standard Dorman 2 lens heads.

There was a few signals in the Farncombe box area where they had 3 lens, but the actual lens was led. They were unusual.

The signals on the first gantry country end of Woking have had the retrofit, not replaced by new Dorman heads

== Doublepost prevention - post automatically merged: 21 Aug 2020 ==

Prefaced by saying I'm not professionally involved with railway signalling.

Although I can say that in similar safety critical applications I am familiar with the lamp itself can be fitted with a method of drawing current (usually circuitry that creates heat and a heat sink) which fits into the same size envelope as the original lamp. Although you do not get the efficiency advantage of going to LED these systems are normally much longer lasting than filament lamps with resulting savings in maintenance and staffing costs.

As the previous poster has pointed out it is much more common to switch out the whole lamp housing which normally allows space for control circuitry to prove the LED is lit and then send current to the proving circuit to show when the "lamp" has failed (again not in the railway industry to be clear). In the use cases I'm familiar with it actually tends to be the control circuitry that is the complication that fails rather than the relatively simple circuitry of the LED itself.

Surely that goes around the efficiency advantage of LEDs? Reprogramming the signal modules can't be that hard to account for the change in current

 

[John Webb]

Saw at Woking some of the conventional filament signal heads were retrofitted with LED lamps (could tell by the instantaneous snapping between aspects, ie no fade as the filament cools). The route indications remain filament however. Is this a simple drop in replacement? Surely the LEDs would trip the current sensing for lamp proving? Would the SSI have to be reprogrammed to account for the change in current?

The filament bulbs use a low-resistance relay in series with the filament; if the filament fails the relay drops out, switches on the secondary filament and sends an alert signal to the appropriate supervising box. I've not handled the replacement LED modules or seen circuitry for them but assume they operate the same relay as long as the LED is working alright.
As for the modern LED signals, we have a few Dorman heads working in demonstration at St Albans South, but they are sealed units and I've not got inside them to discover their workings!

 

[Randomer]

Surely that goes around the efficiency advantage of LEDs? Reprogramming the signal modules can't be that hard to account for the change in current

If efficiency was purely about energy consumption then yes that would be the case, as would be common in domestic or most commercial properties for example.

If you view efficiency as being the sum of; power consumption, "lamp" life, cost of manufacture (now that filament lamps have all but died out in domestic or commercial use they are becoming much more expensive to obtain for exempt usage) and mean time between failure you can come to a different view. Essentially for a critical application where failure will become a problem the higher MTBF tends to be the overriding factor rather than energy consumption especially when it allows a reduction in maintenance time used for proactive lamp changes when they are "out of hours". A lot of the traditional lamps in examples I can think of are relatively low wattage anyway.

 

[ABB125]

I can think of are relatively low wattage anyway.

How much power would a typical signal filament bulb use? 60W maybe?

 

[Randomer]

How much power would a typical signal filament bulb use? 60W maybe?

No idea, I'm not involved with railway signalling. I'm sure somebody more knowledgeable about such things will be along shortly.

The applications I've seen that are safety critical are not signalling needing to be viewable from a huge distance so are generally less than 20W.

 

[edwin_m]

Surely that goes around the efficiency advantage of LEDs? Reprogramming the signal modules can't be that hard to account for the change in current

In relay-based signaling lamp and filament proving is by a relay circuit with no software to re-program, and the small current needed to operate an unmodified LED lamp might not be enough to operate a relay. Similarly for a computer-based installation the current detection on the module might be designed for filament lamps as a hardware circuit with a fixed sensitivity, just feeding a binary yes/no input into the software.

 

[MadMac]

How much power would a typical signal filament bulb use? 60W maybe?

The common bulb of the past few decades is the SL35, with two independent filaments each rated at 24 watts. There’s a small relay in series with the “main” filament: if that filament fails, the relay drops out and changes over to the “auxiliary“ filament. That then triggers an alarm to the local technicians that one of a group of signals is working on its ‘auxiliary“ filament and needs attention.

 

[Llanigraham]

How much power would a typical signal filament bulb use? 60W maybe?

I'm sure I remember some spares being left in our Box that were 25w and looked just like some of the bulbs in my car.
With a good lens you don't need a powerful bulb.

 

[MarkyT]

The common bulb of the past few decades is the SL35, with two independent filaments each rated at 24 watts. There’s a small relay in series with the “main” filament: if that filament fails, the relay drops out and changes over to the “auxiliary“ filament. That then triggers an alarm to the local technicians that one of a group of signals is working on its ‘auxiliary“ filament and needs attention.

Techs became adept at determining which lamp was on auxiliary by watching trains move about on the panel, knowing what aspects should be being displayed on the ground, and observing when the alarm appears and disappears. In remote relay rooms, there's often a local panel showing track and signal indications (sometimes also incorporating a local emergency operations facility) on which they can watch all this activity without bothering the signallers.

 

[ABB125]

The common bulb of the past few decades is the SL35, with two independent filaments each rated at 24 watts. There’s a small relay in series with the “main” filament: if that filament fails, the relay drops out and changes over to the “auxiliary“ filament. That then triggers an alarm to the local technicians that one of a group of signals is working on its ‘auxiliary“ filament and needs attention.

Thanks.
That seems surprisingly little power for something that can be seen from miles away. Presumably it's used with some sort of reflector?

 

[MarkyT]

Thanks.
That seems surprisingly little power for something that can be seen from miles away. Presumably it's used with some sort of reflector?

NEVER a reflector, in UK at least. That is considered dangerous for possibly reflecting low sunlight, which could cause a false 'phantom aspect'. The insides of traditional signal housings are painted matt black to avoid any risk of this. Signals use multi-element Fresnel lenses to create the tightly focussed directional beam.

 

[broadgage]

AFAIK, some railway signals used twin filament bulbs with one 12 volt 16 watt filament, and one 12 volt 24 watt filament. I purchased some of these lamps from a government surplus dealer years ago.
They have a special 3 pin bayonet cap. One of the two base contacts for each filament with a common return via the shell of the lamp base.

 

[MarkyT]

AFAIK, some railway signals used twin filament bulbs with one 12 volt 16 watt filament, and one 12 volt 24 watt filament. I purchased some of these lamps from a government surplus dealer years ago.
They have a special 3 pin bayonet cap. One of the two base contacts for each filament with a common return via the shell of the lamp base.

I was always told the auxiliary in a SL35 was slightly more powerful than the main filament. This was because the main was at the ideal sweet spot in the optical system while the auxiliary was necessarily slightly offset and needed to generate a little more light to compensate.

 

[MadMac]

I was always told the auxiliary in a SL35 was slightly more powerful than the main filament. This was because the main was at the ideal sweet spot in the optical system while the auxiliary was necessarily slightly offset and needed to generate a little more light to compensate.

It’s been a while, but I think you’re right there.

 

[ABB125]

NEVER a reflector, in UK at least. That is considered dangerous for possibly reflecting low sunlight, which could cause a false 'phantom aspect'. The insides of traditional signal housings are painted matt black to avoid any risk of this. Signals use multi-element Fresnel lenses to create the tightly focussed directional beam.

That's interesting, thanks.

 

[Belperpete]

Surely that goes around the efficiency advantage of LEDs? Reprogramming the signal modules can't be that hard to account for the change in current

To make it simple to convert filament heads to LED ones, the LED signal heads are designed to mimic the current drawn by SL35 filament bulbs. There is then no need to change the associated control and indication circuitry. The changeover can, I think, be done as a simple maintenance "like-for-like" changeover.

It would be possible to modify the existing lamp-proving circuitry to prove a much lower current with LED heads. For a traditional relay interlocking, this would involve replacing the existing current-proving relay in the adjacent equipment case with a different type designed to prove a lower current. For SSI signal modules, the module is not reprogrammed, but instead the wiring to the module would need to be modified so that the current is passed through an external current-sensing resistor, rather than the one built into the module. The central interlocking's diagnostic data should also be amended to match.

However, there are two problems with this:
1) With the very low currents drawn by LEDs alone, the current-sensing circuitry in the equipment case would be very susceptible to false operation by currents induced on the trackside cables, current leakage, etc. In practice, it would still be necessary for the LED heads to draw additional current, in order to operate the current-sensing reliably.
2) The alterations to the external control circuitry would need to be designed, verified and tested in a properly controlled manner. Changing a number of heads each night/weekend would involve a series of properly staged changeovers. The design and testing costs would soon mount up, and signalling design and testing staff are a scarce resource.

 

[ewsclass]

To make it simple to convert filament heads to LED ones, the LED signal heads are designed to mimic the current drawn by SL35 filament bulbs. There is then no need to change the associated control and indication circuitry. The changeover can, I think, be done as a simple maintenance "like-for-like" changeover.

It would be possible to modify the existing lamp-proving circuitry to prove a much lower current with LED heads. For a traditional relay interlocking, this would involve replacing the existing current-proving relay in the adjacent equipment case with a different type designed to prove a lower current. For SSI signal modules, the module is not reprogrammed, but instead the wiring to the module would need to be modified so that the current is passed through an external current-sensing resistor, rather than the one built into the module. The central interlocking's diagnostic data should also be amended to match.

However, there are two problems with this:
1) With the very low currents drawn by LEDs alone, the current-sensing circuitry in the equipment case would be very susceptible to false operation by currents induced on the trackside cables, current leakage, etc. In practice, it would still be necessary for the LED heads to draw additional current, in order to operate the current-sensing reliably.
2) The alterations to the external control circuitry would need to be designed, verified and tested in a properly controlled manner. Changing a number of heads each night/weekend would involve a series of properly staged changeovers. The design and testing costs would soon mount up, and signalling design and testing staff are a scarce resource.

Thanks for the reply, interesting stuff. Completely overlooked the planning and testing phase. What's the maximum distance cables from signal heads to their associated LOCs?

 

[MarkyT]

To make it simple to convert filament heads to LED ones, the LED signal heads are designed to mimic the current drawn by SL35 filament bulbs. There is then no need to change the associated control and indication circuitry. The changeover can, I think, be done as a simple maintenance "like-for-like" changeover.

It would be possible to modify the existing lamp-proving circuitry to prove a much lower current with LED heads. For a traditional relay interlocking, this would involve replacing the existing current-proving relay in the adjacent equipment case with a different type designed to prove a lower current. For SSI signal modules, the module is not reprogrammed, but instead the wiring to the module would need to be modified so that the current is passed through an external current-sensing resistor, rather than the one built into the module. The central interlocking's diagnostic data should also be amended to match.

However, there are two problems with this:
1) With the very low currents drawn by LEDs alone, the current-sensing circuitry in the equipment case would be very susceptible to false operation by currents induced on the trackside cables, current leakage, etc. In practice, it would still be necessary for the LED heads to draw additional current, in order to operate the current-sensing reliably.
2) The alterations to the external control circuitry would need to be designed, verified and tested in a properly controlled manner. Changing a number of heads each night/weekend would involve a series of properly staged changeovers. The design and testing costs would soon mount up, and signalling design and testing staff are a scarce resource.

Some modern LED signals have a different form of lamp proving using a voltage free proving contact within the signal head. The circuit is modified or provided new to omit the current proving method altogether, and a repeat of the proving contact is used instead in functions that need to confirm the signal is alight. For example, the lightweight signal range from Unipart Dorman can be ordered with this feature, resulting in very low power consumption (5W).

Lightweight Signalling - Unipart Dorman

We have worked in close collaboration with Network Rail to develop the latest generation of LED railway signals.

www.unipartdorman.co.uk

Following the Modular Signalling challenge to the industry, we have worked in close collaboration with Network Rail to develop the latest generation of LED railway signals. This has resulted in the introduction of our innovative Lightweight Signalling Range, which is specifically designed to meet the requirement for cost effective signalling in the 21st century.

Low power signals may be available from other manufacturers.

 

[John Webb]

The whole of the sensing of the current taken by the traditional filament bulb is done within the signal head. From one of our late 1970s 4-aspect signal heads:

(This head was modified by us to use 5W car tail-lamp bulbs as we found the 24W bulbs were too bright for close viewing.)
On the left can be seen the relay that monitors the current through the main filament. Just above it is the transformer that supplies the 11 volts or so to the lamp - there are various taps on the transformer to allow precise adjustment of the voltage so that the lamp life is prolonged.
On the upper lamp holder it is just possible to see the wire/lead seal under the holder on the screws holding the holder; these protect the factory-set alignment of each aspect.
This photo also shows the inner coloured lenses - the outer lens is plain glass to minimise coloured reflections due to low-level sunlight.

The auxiliary is the same wattage as the main filament - but being slightly out of focus appears dimmer to approaching train drivers (so I am told!) as a back-up to allow being spotted by a driver.

 

[ewsclass]

Some modern LED signals have a different form of lamp proving using a voltage free proving contact within the signal head. The circuit is modified or provided new to omit the current proving method altogether, and a repeat of the proving contact is used instead in functions that need to confirm the signal is alight. For example, the lightweight signal range from Unipart Dorman can be ordered with this feature, resulting in very low power consumption (5W).

Lightweight Signalling - Unipart Dorman

We have worked in close collaboration with Network Rail to develop the latest generation of LED railway signals.

www.unipartdorman.co.uk

Low power signals may be available from other manufacturers.

So would I be correct in thinking that the contact is closed during normal operation and opened under failure?

 

[Belperpete]

Some modern LED signals have a different form of lamp proving using a voltage free proving contact within the signal head. The circuit is modified or provided new to omit the current proving method altogether, and a repeat of the proving contact is used instead in functions that need to confirm the signal is alight. For example, the lightweight signal range from Unipart Dorman can be ordered with this feature, resulting in very low power consumption (5W).

Lightweight Signalling - Unipart Dorman

We have worked in close collaboration with Network Rail to develop the latest generation of LED railway signals.

www.unipartdorman.co.uk

Low power signals may be available from other manufacturers.

Agreed, and these are being used on a number of new signalling schemes, particularly the "modular signalling" schemes. However, it would be difficult to retrospectively fit these types of heads to existing signalling schemes, such as Woking, where the lamp-proving is currently done in the trackside equipment cases.

== Doublepost prevention - post automatically merged: 21 Aug 2020 ==

So would I be correct in thinking that the contact is closed during normal operation and opened under failure?

That would be the fail-safe way of doing it.

== Doublepost prevention - post automatically merged: 21 Aug 2020 ==

The whole of the sensing of the current taken by the traditional filament bulb is done within the signal head. From one of our late 1970s 4-aspect signal heads:
View attachment 82558
(This head was modified by us to use 5W car tail-lamp bulbs as we found the 24W bulbs were too bright for close viewing.)
On the left can be seen the relay that monitors the current through the main filament. Just above it is the transformer that supplies the 11 volts or so to the lamp - there are various taps on the transformer to allow precise adjustment of the voltage so that the lamp life is prolonged.
On the upper lamp holder it is just possible to see the wire/lead seal under the holder on the screws holding the holder; these protect the factory-set alignment of each aspect.
This photo also shows the inner coloured lenses - the outer lens is plain glass to minimise coloured reflections due to low-level sunlight.

The auxiliary is the same wattage as the main filament - but being slightly out of focus appears dimmer to approaching train drivers (so I am told!) as a back-up to allow being spotted by a driver.

Sorry, but what you are showing is the first-filament proving in the signal head. The relay will de-energise when the first filament fails, in order to switch on the second, backup filament. But provided this second filament is working, the signal will still be alight.

There will normally be a separate current-sensing relay in the lineside location case to prove that the signal is alight. And it will be contacts of this relay that control the signals in rear, and the signaller's indications.

 

[John Webb]

.....Sorry, but what you are showing is the first-filament proving in the signal head. The relay will de-energise when the first filament fails, in order to switch on the second, backup filament. But provided this second filament is working, the signal will still be alight.

There will normally be a separate current-sensing relay in the lineside location case to prove that the signal is alight. And it will be contacts of this relay that control the signals in rear, and the signaller's indications.

Interesting - the circuitry we were given shows a second voltage-free contact on the lamp-proving relay going to the monitoring equipment in the supervising box. with no further current sensing. I'm unclear how you could further sense the current in a trackside Location cabinet or relay room - it would have to be done on the 100V supply at about a quarter the current - plus whatever the transformer takes on its own. Is it possible this sensing was developed after our 1970s heads were put into use?

 

[Bald Rick]

The whole of the sensing of the current taken by the traditional filament bulb is done within the signal head. From one of our late 1970s 4-aspect signal heads:
View attachment 82558
(This head was modified by us to use 5W car tail-lamp bulbs as we found the 24W bulbs were too bright for close viewing.)
On the left can be seen the relay that monitors the current through the main filament. Just above it is the transformer that supplies the 11 volts or so to the lamp - there are various taps on the transformer to allow precise adjustment of the voltage so that the lamp life is prolonged.
On the upper lamp holder it is just possible to see the wire/lead seal under the holder on the screws holding the holder; these protect the factory-set alignment of each aspect.
This photo also shows the inner coloured lenses - the outer lens is plain glass to minimise coloured reflections due to low-level sunlight.

The auxiliary is the same wattage as the main filament - but being slightly out of focus appears dimmer to approaching train drivers (so I am told!) as a back-up to allow being spotted by a driver.

Wow - a signal maintenance engineer would have a fit if he/she saw all that non-standard wiring!

 

[MarkyT]

Presumably it's used with some sort of reflector?

NEVER a reflector, in UK at least.

Further to my previous comment, parabolic reflectors WERE used in the case of the traditional searchlight signal. These have an electrical solenoid mechanism that moves a three-colour filter in front of the single lamp. In that case, if low sunlight enters the signal through the lens and gets reflected back it can only illuminate the aspect the signal is supposed to be showing, so there is no risk.