Updated 28-XI-2011
Sodium Vapour
Spectral Properties
Lamp Technology
Vapour Pressure
Current Density
Gas Filling
Sodium Migration
Failure Mechanisms
Lamp Designs
Low Voltage Style
     Compton's Lamp
     Philora DC
     GE NA-9
High Voltage Style
     Philora AC
     SO/H U-Tube
     SOI/H Integral
     SOX/H Coated
     SLI/H Linear
Self-Starting Style
     Double Ended
     Single Ended
Control Gear
Series Operation
Autoleak Reactance
Ballast-Ignitor System
High Frequency Electronic

SO/H Positive Column AC Lamps

The first U-shaped positive column lamps were put on the market in 1932, a very busy year in which three generations of sodium lamp were introduced and promptly superseded! They were known as type SO/H, SO denoting SOdium vapour and the /H implying that the lamp had to be burned horizontally. For the first time in this year, the solid borate glass was abandoned because it was expensive and so difficult to work with, and it was succeeded by an ordinary soda-lime glass tube having a thin layer of borate blown onto its inner surface. The tube was bent into a U-shape, and an oxide coated electrode was sealed into each end of the tube using the reverse-pinch kind of seal that Philips had invented for its DC lamps. The lead wires were soldered to the base contacts of a porcelain bayonet cap, and an enamelled iron fork gave mechanical support to the U-tube and located the bend inside a dimple in the inner side of the dewar jacket. The fork was electrically connected to one of the electrodes and served as a third auxiliary external electrode to facilitate ignition of the discharge.

If the discharge tube was run in free air, thermal losses would prevent it from getting hot enough to vaporise the sodium with the effect that the lamp would never run up. To achieve the optimum temperature of 260°C without providing an external heater, some form of thermal insulation was required. The lamp with the best insulation would be the most efficient since less power would be needed to keep it warm. Heat is lost from the discharge tube by conduction, convection and radiation and the ideal lamp should minimise each of these three loss mechanisms.

Conduction and convection losses were greatly reduced by a relatively simple approach - the discharge tube was mounted inside a separate, detachable dewar vacuum jacket. Heat is conserved as it cannot be conducted or convected across the vacuum space in the jacket. The principle is just the same as when we use a Thermos vacuum flask, say, to keep a drink of coffee warm. However SO/H did little to combat the loss of radiated heat, and conduction and convection losses from the discharge tube to the air space around it prevented efficacy from increasing further. An SO/H lamp is shown diagramatically in Figure S21 and some examples have been photographed and can be viewed with individual details from the main page of this website.

Figure S21 - Diagram of the SO/H Dewar Style Lamp

The first SO/H lamps were offered by Philips in four wattages, details of which can be found in Table S1 below. All lamps were rated for 2,500 hours service (Dorgelo and Bouma, 1937). It is not stated whether these figures apply to 100-hour values or average through-life performance, but the former would seem to be the more common figure for lamps of this era.
Type Lamp Current Lamp Voltage Initial Lumens Efficacy
50 W 0.6 A 80 V 2,550 lm 51 lm/W
65 W 0.6 A 110 V 3,780 lm 58 lm/W
100 W 0.6 A 165 V 6,100 lm 61 lm/W
150 W 0.9 A 165 V 9,600 lm 64 lm/W
Table S1 - Specifications of the First SO/H Style Lamps

Soon after 1937, a new glass type was made which offered a reduced rate of argon adsorption and thereafter lamps could be produced with a smaller dose of that efficacy-sapping gas. The new gas filling increased lamp efficacy by a small amount, and the lamps were re-rated to lower wattages to keep the luminous flux roughly the same for each size. The data for new re-rated SO/H lamps is quoted in Table S2, these being 100 hour values (Philips UK Catalogue, 1943).
Type Lamp Current Lamp Voltage Initial Lumens Efficacy
45 W 0.6 A 75 V 2,700 lm 60 lm/W
60 W 0.6 A 100 V 4,200 lm 70 lm/W
85 W 0.6 A 140 V 6,460 lm 76 lm/W
140 W 0.9 A 155 V 10,640 lm 76 lm/W
Table S2 - Specifications of the Re-Rated SO/H Style Lamps

One of the principal problems with all dewar-jacketed lamps was that as they cooled after switching-off, the air around the discharge tube contracted, thus drawing in cold air from outside. This often carried dust into the lamp, which gradually built up and absorbed light. In addition if moisture was drawn in, this would form films of condensation and being an electrical conductor, wet lamps were often very difficult to strike up in the evenings of the colder seasons. In the early days public lighting engineers would walk the streets on humid evenings and manually switch on and off the lantern in question in the hope that it would strike up after a few attempts.

BTH-Mazda was the only firm to attack these issues and their lamps were the favoured brand for many years. A pair of sachets of silica gel were attached to the U-bend of the lamp to absorb moisture, and in the 85W rating which was the most difficult size to start up owing to its long tube of slim diameter, the glass was also coated with a water-repellent silicone film. At one point, the company also included a xenon component in the gasfilling of the 85W lamp because the this rating has the highest electrical loading on the glass, and argon absorption by the glass was causing premature lamp failure. It was first discovered by Osram-GEC that xenon is not absorbed by the glass and the BTH 85W lamps gave considerably longer life, although the high atomic weight of xenon did result in the loss of some 12% lumens.

In 1955 a new kind of glass was introduced by Philips which offered considerably enhanced resistance to sodium corrosion. A disadvantage however, was that liquid sodium exhibited rather poor adhesion to its surface and if not kept absolutely level, it would flow around the lamp forming large light-blocking mirrors. Although it improved lumen maintenance figures due to reduced browning of the glass, this was largely offset as a result of the formation of sodium mirrors. To combat that problem, the so-called 'Bamboo' lamp was invented by Philips and also manufactured very briefly by Osram-GEC a few years later. The glass was rilled in at several points giving it the appearance of a bamboo cane, and the ridges were successful in preventing the liquid sodium from flowing around the lamp (Figure S22).

Figure S22 - Bamboo type construction employed in post-1955 Philips lamps

The new glass had another drawback though, in that it absorbed argon at a much faster rate than the previous composition which would become stained brown rather rapidly. To attain a useful lamp life, the rare gas filling was changed at this time to a neon-xenon-helium mixture. The elimination of argon increased lamp life, but the new gas reduced lamp efficacy due to increased eleastic and inelastic collisions. However this was a penalty that affected only new lamps. During life it was more than offset by the greater light transmission of the new glass which did not stain so rapidly. Lamp efficacy for the 140W model rose from 76 to 84 lm/W as a result.

Meanwhile BTH Mazda set up its own production of a sodium resistant hard glass which marked an important step forward for that company. The borosilicate 2-ply tubing it created was chemically stable while also having good adhesion to liquid sodium and a fairly low argon cleanup rate. Thus their lamps, except the 85W as noted above, could operate at somewhat higher efficacy - up to 87lm/W for the 140W rating. A further advantage, and in fact the main reason why that hard glass was developed, was that it had a lower coefficient of thermal expansion than the Philips/Osram soft glasses and the occurrence of cracking in production and service was almost totally eliminated. There was a severe price penalty for the use of this glass though, and once an improved soft glass had been developed, production reverted to that style and the usual greater production scrap rates associated with soft glass were accepted.

The argon cleanup problem of the new soft glass was solved in 1958 by treating the glass with excess argon during manufacture, and in that year the efficacy of Philips and Osram sodium lamps increased from 84 and joined the 87lm/W figure of BTH because they also could re-adopt the former neon/argon gas filling.

Around the same time, Philips pioneered the introduction of small dimples in the sides of the discharge tube, which formed cool spots and served as sodium reservoirs to hold the sodium in place. This offered a marked improvement over the bamboo and earlier lamps because the dimples, being cold spots, slowed the rate of sodium being distilled towards the U-bend. A better distribution of sodium vapour along the tube length was maintained during life, and as a result lumen maintenance was greatly improved. Whereas the bamboo lamps tended to drop to about 60-65% lumen maintenance at the end of their 4000-hour rated life, the dimpled lamps attained about 75-80%. Another advantage of the dimples is that the sodium is concentrated in globules of greater thickness and smaller diameter. This has a positive effect in reducing the light-blocking effect caused by sodium mirrors in earlier lamps, and contributes about 3.5% improvement in luminous flux.

Due to the fragility of the exposed dimples, they were only employed in SO/H detatchable lamps for a brief period. It was very easy to scratch the glass surface during insertion into the dewar jacket. Additionally the diameter of the inner wall of the dewar jacket was set so as to accommodate the ordinary discharge tubes - but the dimples approached the wall of the dewar much more closely, and if not perfectly aligned, vibration may cause the inner to impact the dewar and lead to breakage of the dimples. Some references state that they were introduced as early as 1956, but quickly withdrawn after their introduction on the SO/H type lamps and production reverted to the earlier bamboo design, because the rate of breakage was too high. The dimples made a renewed appearance following the 1958 introduction of Philips' first Integral-style lamp, which is covered in the next section. Other references however state that the dimples were only first introduced on the dewar-type lamps in 1958, and that production of the sodium lamps was upgraded directly in that same year to the Integral design.

Figure S23 - Dimpled type construction employed in post-1958 Philips lamps