[0001] This invention is concerned with high efficacy high pressure sodium (HPS) arc discharge
lamps as disclosed e.g. in US-A-39 06 272. Such lamps have a non-vitreous, for example
alumina, art tube having electrodes at its ends and containing sodium, mercury, and
a starting gas. The invention is particularly concerned with improving the efficacy
of such lamps by design changes which reduce the wall loading, and reduce the average
arc current density while simultaneously maintaining the wall temperature above about
1100°C.
Background of the invention
[0002] It is well known in the prior art that the useful visible radiation from an arc discharge
in a mixture of sodium and mercury vapors is only one of several modes of energy dissipation
by such arcs. In order to optimize the efficacy of a high pressure sodium lamp incorporating
such an arc, it is necessary to minimize all the non-useful modes of energy dissipation
as a result of the collective effects of such variables as arc temperature, sodium
and mercury pressures, power input per unit length, tube diameter and tube wall temperatures.
As a result of such determinations, we have found that the present designs of HPS
lamps, optimized for diameter, wall loading, sodium and mercury pressures by empirical
techniques known to the prior art, suffer from a number of intrinsic compromises that
have hitherto been unsuspected by the most knowledgeable workers in the field. For
instance, we have found that, at constant power input and sodium pressure for a given
size tube, efficacy increases with increasing wall temperature by 6 to 10% per 100K.
The reason for this increase is that self-absorption of sodium D line radiation in
the self-reversed portion of the line is decreased at constant sodium pressure as
wall temperature T
w increases, because the density of neutral sodium atoms in the cooler gas near the
walls decreases as T
w increases, according to PNalkT w, where P
Nl is the sodium vapor pressure and k is Boltzmann's constant. In Figure 1 is shown
as a shaded area in a spectral power distribution the additional radiation which is
emitted (at a constant art temperature and sodium vapor pressure) at 1500K wall temperature
in comparison to 1300K. Simultaneously, the loss of energy per unit area from the
arc by conduction of heat to the wall decreases as T
w increases, since the temperature gradient between the arc and the wall decreases.
Figure 2 illustrates the measured dependence of efficacy as a function of arc tube
wall temperature determined from an experiment in which the wall temperature of a
lightly-loaded arc tube was varied by operating it inside an independently controllable
furnace.
[0003] Accordingly, if all other factors were held constant, this factor would cause the
efficacy to increase as wall loading (power/unit area of external wall surface) is
increased, because wall temperature increases as wall loading increases. High wall
loadings are best achieved by operating at high power input per unit of arc length
in tubes of small wall diameter. This has tended to dictate empirically developed
designs of HPS lamps operating at or above about 14 watts/cm
2 of wall loading, requiring power input per unit of arc length of about 30 watts/cm
or greater and tube inside diameters typically less than 1 cm.
[0004] The arc temperatures which result from such conditions of operation are typically
of the order of 4000K, and increase with increasing power per unit length. As a result
of our researches, we have determined that the dependencies on arc temperature of
two of the major useless radiative energy-loss mechanisms of the arc (infrared line
emission and infrared continuum emission) are substantially greater than that of the
useful visible emission in the sodium D lines. Accordingly, as arc temperature increases,
these two useless energy loss mechanisms increase faster than the desired sodium D
emission, decreasing the ratio of useful visible to non-useful infrared, and with
it the efficacy. Accordingly, at constant wall temperature, constant sodium pressure
and constant tube diameter, efficacy would decrease with increasing power per unit
length, and therefore wall loading. Correspondingly, from this factor, efficacy would
increase as the power per unit length and the arc temperature decrease.
[0005] Immediately, therefore, we now recognize an intrinsic compromise inherent in lamps
of the prior art. One factor increases efficacy with increasing power per unit length
and wall loading; another decreases efficacy with increasing power per unit length
and wall loading. It has never been possible to take advantage of the separate effects
of increased efficacy at reduced power per unit length, and increased efficacy at
higher wall temperature, since in prior art lamps power per unit length and wall temperature
have been inexorably tied together. In fact, since the wall temperature effect is
somewhat larger than the power/unit length effect, the net result in any practical
prior art lamp has been an efficacy which slowly increases with power per unit length
up to the maximum permitted by the temperature capability of the arc tube material,
when measurements are made at optimum sodium pressure.
[0006] The empirical dependence of efficacy on sodium pressure at constant tube diameter
and power per unit length is well known to the prior art, and results in a maximum
efficacy at that sodium pressure for which the separation between the red wing and
blue wing maxima of the self-reversed sodium D line is 8 to 10 nm. This in turn results
from the competition of two effects, to wit: as sodium pressure decreases toward very
low levels, the lumens per radiated watt of sodium D radiation approaches a constant
525 lumens/watt; however, the total sodium D radiation decreases with decreasing sodium
pressure, and hence overall efficacy decreases. On the other hand, at sodium pressures
above the optimum, the concomitant broadening of the sodium D line results in increasing
of this radiation in the far red and near infrared, to which the eye is insensitive.
Accordingly, the average lumens per radiated watt of sodium D radiation decreases
toward 300 lumens/watt. The total fraction of input energy radiated in the sodium
D line tends to approach a saturation value with increasing sodium pressure, however;
consequently the overall lamp efficacy must decrease with increasing sodium pressure
in this domain. The maximum of lamp efficacy then is found at an optimum pressure
intermediate to the "low" and "high" pressure domains.
[0007] As a consequence of our researches, we have found that the optimum sodium pressure
for maximum efficacy depends on tube diameter (d) in the following way. Maximum efficacy
is found at a D line separation of 8 to 10 nm, independent of diameter, but the sodium
pressure P
Na required to yield this D line separation decreases with increasing diameter according
to the expression, P
Na is proportional to 1/Vd. We further find that the various modes of energy loss from
the arc depend on sodium pressure and tube diameter at constant arc and wall temperatures
in the following way:
sodium D radiation per unit length of arc is proportional to P2Nad2;
infrared lines per unit length of arc is proportional to PNad3/2;
infrared continuum per unit length of arc is proportional to P2Nad2; and
heat conduction loss per unit length of arc is approximately independent of PNa and d.
[0008] When P
Na is restricted to its optimum value, varying as 1/Vd, the diameter dependencies of
the varying modes of energy dissipation at constant arc and wall temperature are:
sodium D radiation per unit length of arc is proportional to d;
infrared lines per unit length of arc is proportional to d;
infrared continuum per unit length of arc is proportional to d; and
heat conduction loss per unit length of arc is approximately independent of d.
[0009] We see, therefore, that the fraction of input energy dissipated by heat conduction
to the arc tube wall, which amounts in a typical 400 watt HPS lamp of the prior art
to approximately one-third of the input power, may be effectively reduced by the use
of larger diameter arc tubes; all radiation losses increase with diameter, while heat
conduction loss remains constant, and thereby becomes a smallerfraction of the total.
Since it is a major non-luminous energy loss, when the heat conduction fraction is
decreased, luminous efficacy must increase, i.e., luminous efficacy increases with
increasing tube diameter (provided sodium pressure is adjusted to the optimum value
at each diameter).
[0010] Immediately, of course, we again see an intrinsic compromise forced on the lamp designer
that has hitherto gone unrecognized by specialists in the field. As tube diameter
is increased, the heat input to the wall required to maintain a constant temperature
should increase in proportion to diameter; but as we have seen, the heat conduction
from the arc, a major component of that heat input, remains constant. Consequently,
without any special measures to improve heat insulation of the wall, the wall temperature
will decrease as the tube diameter increases. Because of the already-described large
dependence of efficacy on wall temperature, the decrease in wall temperature with
increasing tube diameter wipes out and reverses the gain which would have been observed
at constant wall temperature.
[0011] Moreover, we note that it is of no value to attempt to maintain the wall temperature
constant by simultaneously increasing the power input/unit length as diameter is increased.
This results in a greater increase in the useless infrared lines and continuum than
in the visible sodium D line, because of the increase in arc temperature required
and the higher temperature coefficients of the former.
[0012] As a consequence, the effects of power per unit length and tube diameter on efficacy
uncovered by our researches have in practical lamps been negated by the inverse effects
of wall temperature and have remained undiscovered by the many specialists throughout
the world attacking the problem of design of HPS lamps by the usual empirical techniques.
[0013] The results of our investigations can be summarized as follows.
1. Luminous efficacy increases with increasing wall temperature (all other factors
held constant) because of reduced self-absorption of radiation in the center of the
sodium D line. Each additional watt of radiation permitted to escape in this region
of the spectrum contributes about 500 lumens to the total luminous output.
2. Luminous efficacy increases as power input per unit length decreases below that
of prior art lamps (all other factors held constant) because useless infrared radiation
is decreased thereby to a greater degree than the useful sodium D radiation. It is
to be noted that this increase in efficacy with decrease in power per unit length
does not continue indefinitely to vanishing power per unit length. The continuing
increase in efficacy is limited and eventually reversed by the fact that the heat
conduction loss itself has a lower coefficient of dependence on arc temperature than
any radiation loss. At some low power per unit length the energy loss due to heat
conduction becomes too large in comparison to the desired D line radiation, thus limiting
and reversing the increase in efficacy. There is therefore an optimum power per unit
length which is in the vicinity of 20 to 25 watts/cm, substantially lower than the
operating values of many prior art high pressure sodium lamps.
3. Luminous efficacy increases as tube diameter increases (sodium pressure adjusted
for optimum, all other factors held constant) because useless heat conduction loss
is reduced relative to the useful radiation loss.
[0014] The several energy losses, their functional dependencies and appropriate magnitude
coefficient have been incorporated in a simple energy balance to yield the result
shown in Figure 3, which is a plot of efficacy (normalized to that of the prior art
400 watt lamp, 0.7 cm in inside diameter) vs power input per unit length, with tube
diameter as a parameter; constant wall temperature and optimum sodium pressure for
each diameter is assumed. In this simplified energy balance picture, the change in
the shape of radial temperature profile of the arc with diameter is neglected; when
this factor is included in a more detailed calculation, the increase of efficacy with
diameter is not quite as large, but the trend is identical. The existence of a maximum
in efficacy at an optimum power per unit length is clearly visible in these calculations;
the optimum power per unit length appears to be in the vicinity of 20 to 25 watts/cm,
substantially below the values of many prior art lamps.
[0015] The concepts and principles stated herein are at variance with the prior art understanding
of the means of optimizing high pressure sodium lamps for maximum efficacy. For example,
above-mentioned US-A-39 06 272 discloses, in figure 1, an optimum arc tube inside
diameter for each wattage lamp and design center arc drop; the patent does not recognize
that said optimum diameter results from two competing mechanisms which we have discovered
and disclose herein. We have discovered that with suitable thermal insulation to maintain
wall temperatures sufficiently high, efficacy continues to increase with increasing
diameter up to at least double the diameters disclosed in said patent to be optimum.
[0016] Accordingly, it is an object of this invention to provide a high pressure sodium
vapor lamp of the type disclosed e.g. in US-A-39 06 272 and comprising a non-vitreous
arc tube having electrodes at its ends and containing sodium, mercury, and a starting
gas, which has a higher efficacy as heretofore attainable. This object is achieved,
in accordance with the invention, by modifying the lamp such that the lamp includes
means to maintain the arc tube wall temperature greater than about 1100°C at the central
section, and to maintain the temperature of the sodium-mercury amalgam reservoir at
the value yielding optimum sodium vapor pressure, and that during normal operation
the wall loading is less than about 13 W/cm
2 of arc tube external wall surface, and the current density is less than about 8 Alcm
2 of arc tube internal cross sectional area.
[0017] With the low wall loading and current density taught by the invention in contrast
to the prior art, it may become difficult to maintain the arc wall temperature above
the indicated level (which is well in agreement with the prior art), and thus it may
be necessary to provide means for increasing the operating wall temperature of any
HPS lamp which is less than the maximum permitted by the arc tube material (about
1500K for polycrystalline alumina), thereby permitting an increase in efficacy of
about 6 to 10% per 100K increase in wall temperature. The operating wall temperature
may be increased by improved thermal insulation of the arc tube or by a reduction
in primary thermal radiation and/or heat conduction of the arc tube material. Means
should be provided to maintain the sodium-mercury amalgam reservoir temperature at
the value yield optimum sodium vapor pressure.
[0018] In a lamp including an outer jacket surrounding the arc tube, therefore, preferably
means thermally insulating the arc tube are disposed within or upon the inner surface
of said outer jacket. In one embodiment said thermally insulating means is an infrared
reflective coating of indium oxide and tin oxide, and preferably such a reflective
coating is disposed on the inner surface of a sleeve surrounding the arc tube within
said outer jacket.
[0019] In accordance with the invention, the means for increasing the operating wall temperature
are utilized to make feasible the employment of arc tubes of substantially larger
diameter than prior art arc tubes, in order to achieve the efficacy gain associated
with said larger diameter by keeping the wall temperature at or near the maximum permitted
by the material (about 1500K for polycrystalline alumina) in spite of the reduced
wall loading. Prior art arc tubes had arc tube outer diameters of about 0.6 to 1.0
cm and operated (when optimally designed) at wall loadings of about 14 to 20 watts/cm2.
Prior art arc tubes also generally operated at about 25 to 50 watts per cm of arc
length; in this invention, the power consumption per cm of arc length is generally
less.
[0020] To demonstrate the changes in lamp design which result from the teachings of this
invention, consider a 400 watt HPS lamp, such as has been an article of commerce since
the late 1960's and has not changed substantially in physical dimensions, materials
of manufacture or performance ratings since about 1973. Such lamps are typically rated
at 50,000 lumens, 125 lumens per watt, and do not, on the average, exceed that rating
in performance. Arc tubes used by all manufacturers are substantially similar in dimensions.
Thus, such lamps can be considered to have been thoroughly optimized according to
the teachings of the prior art.
Brief description of the drawings.
[0021]
Fig. 1 is a plot of sodium resonance radiation in terms of spectral radiant flux versus
wavelength, at wall temperatures of 1300K and 1500K.
Fig. 2 shows relative efficacy as a function of arc tube wall temperature, at optimum
sodium vapor pressure.
Fig. 3 is a plot of relative efficacy of HPS lamps versus input power (watts) per
centimeter of arc length, at optimum sodium pressure and constant wall temperature
(about 1500K), for arc tubes having inside diameters of 2.0, 1.5, 1.1 and 0.7 cm.
Fig. 4 shows an HPS lamp in accordance with this invention.
[0022] Example 1, below, illustrates the comparison between the performance of a prior art
lamp and that of a lamp constructed in accordance with the teachings of this invention,
employing translucent polycrystalline yttrium oxide (yttria) as the arc tube material
instead of alumina.
[0023] Both translucent ceramics have the property of becoming opaque in the infrared spectral
region. Alumina becomes absorbent between about 4 microns and about 7 microns wavelength,
whereas yttria becomes absorbent between about 7 microns and about 9 microns; thus
yttria will intrinsically thermally radiate less than alumina at temperatures about
1200°C.
[0024] The thermal radiant emittances of translucent polycrystalline yttria arc tubes, such
as disclosed in U.S. Patents 4,147,744 and 4,115,134, have been measured to be about
0.11, while those of polycrystalline alumina are typically 0.20. This permits the
yttria arc tube to reach a higher wall temperature for a given power per unit area
dissipation or, more importantly for our purposes, to achieve equal temperature to
an alumina arc tube wall at a lower power per unit area. Thus we can provide a higher
efficacy lamp by means of a larger diameter, lower-wall-loaded yttria arc tube maintained
at equal or nearly equal temperature as an arc tube designed according to the prior
art.
[0025] Note the substantial reduction in both current density and wall loading of this lamp
in comparison to the prior art lamp, and the substantial increase in efficacy despite
a somewhat lower wall temperature. It is noted that US-A-3 906 272 does not disclose
an optimum diameter for a prior art 400 watt lamp. However, an extrapolation of the
curves therein to the 400 watt level confirms that 0.732 cm can be considered very
nearly optimum according to the prior art.
[0026] The wall temperatures cited above and elsewhere in this specification are measured
by a radiometric method described by deGroot, J. J., "Comparison Between the Calculated
and the Measured Radiance at the center of the D-lines in a High Pressure Sodium Vapor
Discharge", Proc. 2nd IEE Conference on Gas Discharges, London, p. 124 (1972). This
method is believed to have an accuracy of plus or minus 20 to 30°.
[0027] Example 2 shows the results for a 150 watt 55 volt HPS lamp made in accordance with
this invention as compared to a 150 watt 55 volt HPS prior art lamp. The lamp as per
this invention had an 8 mm inside diameter yttria arc tube while the prior art lamp
had a 5.87 mm inside diameter alumina arc tube, which is very close to the diameter
of 5.75 mm disclosed in US-A-3 906 272 to be optimum for this lamp.
[0028] There is a substantial reduction in both current density and wall loading of this
lamp in comparison to the prior art lamp, and it has higher efficacy as well, even
though the diameter is 39% greater than the diameter disclosed in US-A-3 906 272 to
be optimum. The efficacy gain for the lamp of Example 2 is greater than that for Example
1 because the wall temperature of the new lamp in Example 2 is closer to that of the
prior art lamp.
[0029] Example 3 shows the comparison in efficacy between a 50 watt lamp according to our
invention employing an yttria arc tube for reduced thermal radiative losses, and two
different versions, A and B of 50 watt prior art lamps. Prior art lamp A has been
manufactured for only about a year and has been known to not have been optimized according
to the known prior art, by virtue of its very low wall loading and low arc tube wall
temperature. Experimental lamps manufactured according to our invention with yttria
arc tubes of identical dimension have substantially increased arc tube wall temperatures
and correspondingly increased efficacy. Recently announced prior art lamp B represents
an attempt to further optimize the 50 watt lamp according to the known prior art principles,
viz., by decreasing the arc tube diameter, shortening the arc length, increasing the
wall loading.
[0030] Optimum diameter for this lamp according to US-A-3 906 272 is 0.335 cm. It should
be noted that despite a deviation of more than 40% from said optimum diameter, the
lamp according to our invention has equivalent efficacy. Moreover, prior art Lamp
A was deliberately designed at less than optimum wall loading for alumina in order
to improve its lumen maintenance and ease of manufacture, advantages which are retained
by our lamp but are lost in the more recent prior art lamp B.
[0031] Thus far, the specific examples used to illustrate this invention have been employed
yttria arc tubes. However, other means to reduce thermal radiative losses may also
be used to provide the larger diameter, lower wall loading, lower arc current density
arc tubes that are the subject of this invention, and that have an arc tube surface
wall temperature above about 1100°C., preferably near 1200°C, in spite of reduced
heat input per unit area to the arc tube walls.
[0032] In example 4, below, we describe the use of infrared-reflecting shields to reduce
thermal radiative losses.
Example 4
[0033] A conventional 400 watt lamp was constructed with an alumina arc tube, 7.3 mm inner
diameter by 8.9 mm outer diameter, inside the usual type 7720 glass outer jacket.
However, a quartz sleeve, 29 mm inner diameter by 33 mm outer diameter, surrounded
the arc tube within the outer jacket. On the inner surface of the quartz sleeve was
an infrared reflective coating of indium oxide and tin oxide. Lamp operation is summarized
below.
[0034] At 400 watts the wall temperature is higher than 1200°C normally associated with
the conventional 7.3 mm I.D. design. Thus the quartz sleeve will permit the use of
larger diameter on tubes. However, the use of such a sleeve provides two additional
glass interferences which the light emitted by the arc tube has to pass through. A
large percentage of the reflected radiation from the glass interferences is then lost
through absorption within the lamp. If the observed efficacy of about 124 LPW is corrected
for this loss, we see that the efficacy of the arc tube has increased substantially
above that of the same arc tube mounted without heat conserving means, and is in fact,
substantially greater than the 125 LPW obtainable from prior art 400 watt lamps. This
increase in efficacy has resulted from the reduction in self-absorption of the sodium
D radiation brought about by the lower sodium atom density near the wall that is a
consequence of the higher wall temperature.
[0035] In Example 5, below, we describe the application of the radiant-reflector principle
of thermal insulation to an arc tube with a larger diameter.
Example 5
[0036] A lamp (Lamp C) was made comprising a large diameter alumina arc tube, 11.0 mm I.D.
by 12.5 mm O.D. within a cylindrical type 7720 glass outer jacket. There was an infrared
reflective coating, similar to that of Example 4, on the inner surface of the jacket.
Performance of Lamp C was compared with that of a similar lamp (Lamp D) without the
infrared reflective coating (but with niobium heat shields at the arc tube ends to
raise the end temperature, therefore the pressure, of the sodium-mercury amalgam).
Performance of the lamps is summarized below.
[0037] These results show that the infrared reflective coating raises the arc tube temperature.
A comparison of lumens at similar D lines indicates the advantage gained from the
increase in wall temperature. Conventionally designed lamps operate at 125 LPW at
400 watts and 135 LPW at 1000 watts. Comparison with Lamp C at 700 watts indicates
that higher efficacies can be obtained by this invention than by utilizing conventional
methods of HPS lamp design, Lamp C having higher efficacy at 700 watts than conventional
lamps at 1000 watts.
[0038] As a further illustration of the degree to which our invention differs from the precepts
of HPS lamp-design embodied in the prior art, we offer the data in Table I which shows
the dimensions, average arc current density, wall loading, and arc loading for a number
of high pressure sodium lamps, encompassing all wattages above 70 watts presently
commercially available, designed according to the teachings of the prior art, where
wall loading=P/(nxODxAL) and arc loading=P/AL, where I=lamp current, P=lamp power,
AL=distance between electrode tips and ID, OD=inside and outside diameters respectively.
[0039] An important point to notice is the comparison between the 250 and 250Slamps, the
latter having been optimized for higher efficacy over the former according to the
teachings of the prior art. The 250 watt lamp has a wall loading of 14.6 watts/cm2,
an ID of 0.732 cm and delivers about 26500 lumens, while the 250S lamp has a wall
loading of 19.44 watts/cm2, an ID of 0.587 cm and delivers about 29000 lumens. According
to US-A-3 906 272, the optimum diameter for this lamp is approximately 0.55 cm. Thus,
the direction of change of dimension parameters for increased efficacy according to
the teachings of the prior art is toward smaller diameter arc tubes, with a resulting
increase in wall loading. That teaching is directly opposite the disclosure of this
invention.
[0040] The lamps in Table I are typically designed for maximum efficacy according to the
teachings of the prior art. None of the lamps are designed with a diameter large enough
that the current density is as low as 8.0 amp/cm
2. Nor are any of the lamps designed with a wall loading as low as 13 watts/cm2. Moreover,
the efficacies indicated appear generally to increase with increasing wall temperature,
and all wall temperatures appear to be in excess of about 1100°C. Thus, we may conclude
that the optimum diameters cited in US-A-3 906 272 for each lamp simply represent
the largest possible diameter consistent with a minimum wall temperature of 1100°C
for conventionally constructed high pressure sodium lamps.
[0041] To repeat once more, the central concept of our invention is that still higher efficacies
can be obtained at still larger diameters when suitable steps are taken to reduce
the thermal radiative losses from the arc tube surface so that its temperature can
be maintained above 1100°C even though the heat energy input per unit area of wall
surface may be reduced.
[0042] In a preferred embodiment, a lamp in accordance with this invention comprises a non-vitreous
arc tube 1 having electrodes 2 sealed into the ends. Arc tube 1 contains sodium, mercury
and a starting gas, typically, xenon. A metal framework 3 provides support for the
arc tube and an electrical path to the upper electrode. A support wire 4 is embedded
in glass press 5 and provides electrical connection to the lower electrode. The arc
tube assembly is contained within an outer glass jacket 6. Arc tube 1 was made of
yttria and the results for a 150 watt lamp and a 400 watt lamp made in accordance
therewith are shown in Examples 2 and 1 above, respectively.