SOX lamps: Diagnostics when the lamp goes wrong

Dayburning Eleco HW747 lanterns at Buckley, Flintshire, U.K., 2004Left: Dayburning Eleco HW747 lanterns at Buckley, Flintshire, U.K., 2004.

This page contains details of how a SOX lantern might cease to work correctly and how to fix the problem.

Fewer than 1% of streetlights are not working at any time across a typical county in the UK. Usually when a lamp is not working the council or contractor does not know immediately, as there is no control room; either they have to discover the fault by their own observation, or rely on a member of the public to report it. I have heard mention of a man in Bristol who used to ride around on his bicycle at night and report any faulty streetlights to the local council, who were grateful for his service! (ref: Mike Reid) 

SOX lamps have a rated life typically of 12,000 to 16,000 hours, or 3 to 4 years of full night-time operation. When they are faulty, it can be manifest in a number of different ways:

1. Not coming on at all

Sometimes a lamp can fail to strike simply because it is ‘lazy’, i.e. it has failed to reach the required striking voltage. Next day it will be OK. However it can also indicate a more serious problem. It could be caused by a failed fuse or failed control gear. Alternatively it could be a failure of the lamp (i.e. discharge tube) itself, caused by one of the following:

End of electrode life. This is the reason why most SOX lamps reach the end of their life. The electrodes have an 'emissive coating' applied during manufacture which contains potassium. Early in the life of the lamp, the abundance of potassium on the electrodes causes a split-second purple colour when the lamp switches on before the colour settles to neon red for two or three minutes. Every time the lamp is switched on, this coating slowly gets sputtered off, and over time there will be less of the furry white coating on the electrodes. When all of the emissive coating over the surface of the electrodes has gone, the voltage required by the discharge will exceed what the control gear is capable of supplying, and the lamp will fail to ignite or will only flicker red. (ref: Mike Docherty)

End of gas life: One drawback of borate glass, which must be used in the manufacture of SOX discharge tubes to withstand attack from sodium, is that they tend to attract ionised argon (i.e. argon with an electrical charge).  Most SOX lamps have 1 to 1.5% argon in the neon gas filling, in order to reduce the striking voltage required to start the discharge when the lamp is switched on.  However, all the way throughout lamp life the glass will be absorbing small amounts of the argon.  In some lamps, especially the high-wattage types and in particular the 180W lamp, a situation can arise where the argon gas filling will all be consumed before the electrodes reach their end of life.

Reaction of the getter with the glass. This is a type of ‘premature lamp failure’. This is perhaps the most troublesome kind of failure because it cannot be predicted, and occurs relatively early in lamp life. A getter is a substance that is used to remove residual gas from a vacuum in a tube, therefore maintaining a better vacuum in the outer jacket, preventing heat losses and ensuring that lamp efficacy remains high. In SOX lamps, barium has traditionally been used as a getter. However the barium film can come into contact with the lead wires emerging from the stem, and a high electric field is then set up across the glass, which can then cause the glass stem to decompose. Alternatively since the sodium within the discharge tube is ionised, it can easily be drawn into the glass and crack it if a strong enough electric field is present. (ref: James Hooker)

2. Staying on all day (dayburning)

This indicates a failed photocell — the streetlight stays on all day and night as a failsafe mechanism. 

3. Flickering red

The lamp is trying to strike a light every few seconds but fails. This condition can be brought about in several ways:

Leaking discharge tube.  The igniter is working fine, but is unable to strike a discharge in an arc tube that has leaked. (ref: James Hooker)

The igniter is faulty. A way of testing this is to remove the lamp and turn the lantern on. If you hear a hi-pitch squealing noise the igniter is OK; the noise will go quieter after a few seconds as it cuts out. If however it is silent, the igniter has failed and a new one is required. (ref: Colin Grimes)

The electrodes have run out of emissive coating (see section 1 above). (ref: Mike Docherty)

4. Failing to warm up fully to final colour — i.e. more orange or red than normal

If the lamp does not reach its optimum operating temperature of 260°C, the sodium inside may not vaporise and the red colour of the neon in the lamp tube would be the result. This is normally attributable to a leak in the outer jacket: when the outer vacuum fails, the thermal insulation of the arc tube deteriorates, and the lamp can only partially warm up. This condition can also be caused by the lamp being under-run so that the correct electrical current is not reached, for example if a 35W or 55W ballast is being used to run a 90W lamp. This under-running would also account for a lamp only warming up to a shade of red-orange, orange, or yellow-orange, or to a lamp that shines with a steady red colour without ever changing. (ref: Colin Grimes, James Hooker, Mike Docherty)

5. Lumpy yellow appearance

This can also be caused in several ways:

Uneven sodium distribution in tube. This is the most common cause. However it has no significant effect. After a lamp is switched on, the sodium vapour will always condense back into liquid form on the coldest parts of the lamp tube. Over time, this can cause a build-up of sodium in certain areas in the tube, and a lack of sodium in other areas (the hottest areas when the lamp is running). This can also be caused by gravity, if the lamp tube is not horizontal and the sodium is allowed to run to one end of the tube. This effect of uneven sodium distribution is called “sodium migration”. In the areas where sodium is lacking, the neon will instead be involved in the electrical discharge, and so some areas of the tube may appear redder. Because the ionisation energy required for neon is higher than for sodium, the lamp will run less efficiently. Some SOX tubes are designed with dimples along their length in order to help prevent this effect. (ref: J. W. Denneman)

An uneven distribution of sodium can happen with SOX lamps in post-top lanterns e.g. Thorn Gamma 6 (although post-top lanterns of 18W or 35W normally operate correctly). 

This uneven sodium distribution can also affect the argon in the neon-argon gas mixture (see section 1 — paragraph “End of gas life” above). In the sodium-depleted regions, the neon-argon gas filling becomes ionised to a greater extent and the rate of argon absorption by the glass is accelerated.  (ref: Colin Grimes/James Hooker)

Lamp not hot enough. The effect of uneven colour can also take place if the lamp fails to reach its optimum operating temperature of 260°C. In situations where the temperature is not quite high enough, it may be that only some of the required sodium vapour is produced, and hence there is a mixture of yellow (from sodium) and red (from neon) patches — even if the sodium is evenly distributed. This is a slightly different effect from the ‘lumpy sodium distribution’ mentioned above, where the temperature is high enough but the sodium is poorly distributed. (ref: Mike Docherty)

6. Normal operation for 10-15 minutes only (for example) then switching off and staying off

It is most likely that the components are overheating and the lantern’s thermal cutout switch has come into operation. In more detail, this usually occurs after one of the following:

One of the electrodes has run out of emitter before the other electrode (see section 1 — “End of electrode life”)

Sodium metal comes into contact with one of the electrodes or its lead wires — this is another reason why the lead wires are glass-sleeved, to minimise the risk of contact with liquid sodium.

In either case, the lamp now (incorrectly) begins to convert the normal AC current into a high DC current (a process called “rectification”).  Once rectification begins, much higher currents than normal will flow in both the lamp and the windings of the ballast.  In earlier designs of lamps the high current rectification could continue for many hours.  During this process the temperature of the ballast increases owing to the high current it passes, and this can lead to the breakdown of insulation between the windings, allowing a even higher current to flow.  This causes overheating of the ballast, such that its thermal overload will trip, causing the lamp to shut off.  All SOX lamps today are fused, and if the rectification current becomes too high, the internal fuse in the lamp cap will fail, thus protecting the ballast from further overheating and destruction. (ref: James Hooker)

However, not all lanterns have a thermal cutout, so the problem may be explained as per section 7 below.

7. Normal operation for 30-60 minutes only (for example) then switching off and staying off

I have experienced this problem with my own lanterns (both on the same column), and found that the problem was a break in the wiring in the column which was overcome at first with the initial increased voltage, but would then cause the lamp to switch off as the voltage gradually settled down. Alternatively, it could be that one or more electrical components has failed - it may be necessary to try carious combinations of old/new components.

8. Normal operation for 10-15 minutes only (for example) but periodically going off and coming on again

This is caused by stray light re-entering the lamp via the detector head in the photocell. The lamp is caught in a cycle of firstly switching on because it is dark, and then its own light re-enters the photocell, causing the lamp to ‘believe’ it is light again and therefore switch off. When it is off, it is not producing any light to re-enter the photocell — and therefore the lamp switches on again because it is dark, re-starting the cycle.(ref: Colin Grimes)

Example of blackening in SOX tube9. Dark or black areas along the lamp tube

Blackening of SOX tubes is a recent phenomenon. It can happen after as little as 6 months, and has been caused by changes in manufacturing methods. The problem was almost unheard of until 1994 when GE (General Electric) attempted to cut costs in the redesign of its lamps, in order to respond to an increasingly competitive market, and eliminated the small ceramic insulator that used to support the U-bend of the discharge tube.  A new metal top clip was introduced which made direct electrical contact between the discharge tube and the indium film (which prevents heat loss from the lamp).

The problem that transpired, however, was that in a small percentage of lamps, when the barium getter (which, as you may recall, is there to maintain a good vacuum) was fired it would form another electrically conductive coating running all the way between the indium film and the one of the lead wires.  This set up a strong electric field around the weak tip-off area of the discharge tube which was sufficiently powerful to pull sodium into the outer bulb, where it proceeded to severely attack the indium coating and turn it black.

The outer jackets of sodium lamps will also blacken if the discharge tube leaks and sodium vapour is released, because the glass of the outer bulb has no resistance to sodium corrosion. Even the tiniest of leaks will cause this to happen as sodium is very chemically aggressive towards the indium coating, and it's not uncommon to see lamps which are totally black along the whole length as a result of the sodium-indium reaction. 

Manufacturers have now solved the problem by improving the getter. The PSG lamp eliminates the traditional barium getter and employs the Philips Solid-state Getter instead.  It consists of a zirconium-cobalt pellet mounted on a stalk beside the arc tube. As a result the vast majority of premature failures are avoided. (ref: James Hooker, Mike Barford)

10. The lamp warms up fully to yellow, but subsequently cools again to orange or red

This is most likely caused by a faulty lamp which should be replaced.

Gear failure:

A good indicator that the control gear has failed, if you open up a streetlight, is if the ballast is covered with “sticky golden brown gunk, possibly resin”, accompanied by an unpleasant smell. (ref: Colin Grimes, Mike Barford)

When a SOX lamp is properly manufactured and working properly, it is a impressive blend of materials technology, physics, chemistry and engineering. But like any finely-honed instrument, any slight deviation can mean that it is unusable.

© Matthew Eagles 2005-2018