INSTITUTE OF PHYSICS PUBLISHING
`
`J. Phys. D: Appl. Phys. 38 (2005) 2995–3010
`
`JOURNAL OF PHYSICS D: APPLIED PHYSICS
`
`doi:10.1088/0022-3727/38/17/R01
`
`REVIEW ARTICLE
`UHP lamp systems for projection
`applications
`
`Guenther Derra, Holger Moench, Ernst Fischer, Hermann Giese,
`Ulrich Hechtfischer, Gero Heusler, Achim Koerber,
`Ulrich Niemann, Folke-Charlotte Noertemann, Pavel Pekarski,
`Jens Pollmann-Retsch, Arnd Ritz and Ulrich Weichmann
`
`Philips Research Laboratories, Weisshausstrasse 2, D-52066 Aachen, Germany
`
`Received 15 February 2005, in final form 30 March 2005
`Published 19 August 2005
`Online at stacks.iop.org/JPhysD/38/2995
`
`Abstract
`Projection systems have found widespread use in conference rooms and other
`professional applications during the last decade and are now entering the home
`TV market at a considerable pace. Projectors as small as about one litre are
`able to deliver several thousand screen lumens and are, with a system efficacy
`−1, the most efficient display systems realized today. Short arc
`of over 10 lm W
`lamps are a key component for projection systems of the highest efficiency for
`small-size projection displays.
`The introduction of the ultra high performance (UHP) lamp system by
`Philips in 1995 can be identified as one of the key enablers of the commercial
`success of projection systems. The UHP lamp concept features outstanding arc
`luminance, a well suited spectrum, long life and excellent lumen maintenance.
`For the first time it combines a very high pressure mercury discharge lamp with
`extremely short and stable arc gap with a regenerative chemical cycle keeping
`the discharge walls free from blackening, leading to lifetimes of over 10 000 h.
`Since the introduction of the UHP lamp system, many important new
`technology improvements have been realized: burner designs for higher lamp
`power, advanced ignition systems, miniaturized electronic drivers and
`innovative reflector concepts. These achievements enabled the impressive
`increase of projector light output, a remarkable reduction in projector size and
`even higher optical efficiency in projection systems during the last years.
`In this paper the concept of the UHP lamp system is described, followed by
`a discussion of the technological evolution the UHP lamp has undergone so far.
`Last, but not least, the important improvements of the UHP lamp system
`including the electronic driver and the reflector are discussed.
`(Some figures in this article are in colour only in the electronic version)
`
`1. Introduction
`
`Large screen projection systems became increasingly popular
`during the last decade. The business developed rapidly and
`growth is expected to continue to be high, as shown in figure 1.
`Looking back, the introduction of the ultra high performance
`(UHP) lamp concept by Philips in 1995 [1–4] was a significant
`technological breakthrough for the projection market and can
`be identified as one of the key enablers of the commercial
`
`the
`success of projection systems. Following its launch,
`UHP lamp system has been improved with a fast innovation
`rate [5, 8–10, 12, 14, 19, 20, 22–24]. UHP lamps today are
`the standard for most commercially available front and rear
`projectors and have replaced the previously used metal halide
`lamps.
`About ten years ago, the performance of liquid crystal
`display (LCD) projectors was rather poor: a 46 litre size,
`21 kg projector delivered just 400 screen lumens with VGA
`
`0022-3727/05/172995+16$30.00 © 2005 IOP Publishing Ltd Printed in the UK
`
`2995
`
`Energetiq Ex. 2087, page 1 - IPR2015-01377
`
`

`

`Figure 4. Luminance of light sources compared with the black body
`radiator luminance.
`
`Short arc lamps are a key component for projection
`systems for achieving the highest efficiency for small
`projection display sizes. Projection is a very demanding
`application for the lamp. The light source should be point-like,
`provide extremely high brightness, high total light flux and a
`white spectrum. Besides, high demands on lamp efficiency
`and lifetime have to be fulfilled.
`Further progress in both displays and optics increases
`the optical demands to be fulfilled by the light source. The
`innovation speed therefore depends on the availability of
`improved light sources with smaller size and even shorter arc
`length.
`
`2. The concept of the UHP lamp
`
`2.1. A pure mercury discharge for highest luminance
`
`(cid:1)
`
`·
`
`dλ.
`
`(1)
`
`For highly efficient projection systems the arc luminance
`should be as high as possible. With today’s small display
`−2 is needed.
`sizes, an average luminance of above 1 Gcd m
`The maximum luminance L(T ) that can be reached in thermal
`equilibrium is physically linked to the discharge temperature
`by the well-known Planck’s law:
`L(T ) = k ·
`V (λ)· 2hc2
`1
`ehc/ kT λ − 1
`λ5
`Here, k is the Boltzmann constant, λ is the wavelength of
`emitted light, V (λ) is the wavelength sensitivity curve of the
`human eye, h is the Planck constant, c the speed of light and
`T the radiator temperature.
`Metal halide additives that are used in many lamp types
`for improving the colour properties of the lamp spectrum
`mostly reduce the arc temperature owing to their comparably
`low ionization potential. This leads to a lower luminance of
`the arc and hence does not make metal halide lamps ideal
`for projection. A high luminance (see figure 4) can only be
`reached by rare gas and pure mercury discharges, as used in
`the UHP lamp. A pure mercury discharge, however, is superior
`to rare gas discharges in luminous efficiency while reaching the
`same luminance.
`The UHP lamp follows this approach and contains only
`mercury as radiating species. UHP lamps typically reach an
`−2 and are, in that
`average arc luminance well above 1 Gcd m
`respect, an ideal light source for projection applications.
`
`World Projection Market
`
`Microdisplay PTV
`Front Business
`Crossover
`Front Home
`
`G Derra et al
`
`16000
`
`12000
`
`8000
`
`4000
`
`0
`
`k units
`
`2003
`
`2004
`
`2005
`
`2006
`
`2007
`
`2008
`
`Figure 1. World projection market according to Techno Systems
`Research.
`
`Figure 2. Ultra-portable projector with UHP lamp (InFocus).
`
`Figure 3. Rear projection TV with UHP lamp (RCA).
`
`resolution. Today, projectors of one-tenth that size can create
`high-quality XGA pictures with more than 3000 screen lumens
`brightness with a single UHP 200 W lamp. On the other end of
`the product spectrum, ultra-portable projectors (see figure 2) of
`1 kg weight and less than 1 litre volume enable a bright XGA
`presentation with more than 1000 screen lumens.
`Front projectors—now commonly called ‘beamers’—have
`found their place in almost each meeting room during the last
`ten years. In the last few years also, rear projection TV sets
`(see figure 3) have become important in the market and sales
`are growing rapidly.
`
`2996
`
`Energetiq Ex. 2087, page 2 - IPR2015-01377
`
`

`

`Review: UHP lamp systems for projection applications
`
`WO2Br2
`
`Sp(W)
`
`WO2Br
`
`WO2Br2
`
`W
`WO
`
`WBr
`
`WO2
`
`1 E-04
`
`1 E-05
`
`1 E-06
`
`1 E-07
`
`1 E-08
`
`Pressure / bar
`
`1100
`
`1600
`
`2100
`
`2600
`
`3100
`
`3600
`
`wall
`
`Temperature / K
`
`electrode tip
`
`Figure 6. Principle of regenerative cycle: typical gas phase
`composition of a UHP lamp as a function of temperature. Sp(W) is
`the sum of saturation pressures of all tungsten containing species.
`
`c
`
`b
`
`a
`
`1600
`
`2100
`
`2600
`
`3100
`
`3600
`
`Temperature / K
`
`1.E-03
`
`1.E-04
`
`1.E-05
`
`1.E-06
`
`1.E-07
`
`1.E-08
`
`1.E-09
`
`Sp(W) / bar
`
`1.E-10
`1100
`
`Figure 7. Schematic view of the summed tungsten pressure Sp(W)
`as a function of temperature for different bromine levels (curves a,
`b, c are explained in the text).
`
`tungsten transport
`The direction and magnitude of
`strongly depend on the effective vapour pressure of
`tungsten,
`i.e. on the sum of saturation pressures of all
`tungsten-containing species for a given halogen and oxygen
`concentration. Figure 7 shows the summed tungsten pressure
`Sp(W) as a function of temperature for three bromine levels:
`(a) Without or with insufficient bromine, Sp(W) at the wall
`−20 bar); thus
`will remain at an extremely low level (<10
`only material transports from hot to cold sites will occur
`causing massive blackening with no chance of tungsten
`recovery from the wall.
`−4 bar), Sp(W)
`(b) With sufficient bromine (in the order of 10
`at the wall will exceed Sp(W) at hotter sites of the electrode
`enabling chemical transport of tungsten from cold to hot
`sites (i.e. from wall towards electrodes).
`(c) With excess of bromine, the chemical cycle will still work,
`but at enhanced transport rates. Especially at the electrode
`feed-through with temperatures around 1800 K, chemical
`attack of the electrode rod can occur (see figure 8) which
`will lead to early lamp failure. This effect is strongly
`enhanced if larger amounts of gaseous impurities (H2,
`H2O, CO, CO2) are present (for details see discussion
`in [6]).
`The fruit of all these efforts is lamps with exceptionally
`long lifetimes and good maintenance, as shown in figure 9. The
`integral lumens remain nearly constant for more than 10 000 h
`
`2997
`
`Figure 5. UHP Spectra for different mercury pressure, an arc gap of
`1 mm measured at 120 W and through an aperture representing an
`optical system etendue of E = 10 mm2 ster. This figure is best
`viewed in colour in the online edition. The 120 bar curve is the
`highest in the peaks and the lowest between peaks; the 290 bar curve
`is the lowest in the peaks and the highest between the peaks.
`
`2.2. High pressure for continuous spectrum
`
`For an efficient projection system the spectral properties of the
`light source are at least of the same importance as the lamp
`luminance. Filters are typically used for colour matching and
`for rebalancing white. With an UHP lamp, the efficiency of
`this colour matching and rebalancing is typically 25% for a
`single-panel, sequential colour device and 70% for a three-
`panel projector. In any case, light is lost by this filtering. To
`reduce these losses the lamp should exhibit an even distribution
`of the spectral contributions between red, green and blue.
`Figure 5 shows spectra for various lamp pressures
`measured through an aperture (for details see section 2.7)
`representing the usable light in a typical optical system of
`a modern projector.
`It can be clearly seen that for mercury
`pressures above 200 bar more light is emitted in the continuum
`radiation than in the atomic spectral
`lines.
`Especially,
`the important red light contribution above 600 nm strongly
`depends on the lamp pressure. The more even distribution of
`the emitted light over the wavelength range is directly related
`to a higher colour balancing and therefore to the projector
`efficiency. For good colour balancing in projection systems
`it is essential to realize ultra high lamp pressures.
`
`2.3. Regenerative cycle for long life
`
`The lifetime of UHP lamps can exceed 10 000 h by far. This
`is realized for the first time in a commercial HID lamp by a
`so-called regenerative chemical transport cycle using a halogen
`filling [1].
`As principally known from halogen incandescent lamps,
`in the presence of halogen and oxygen (O2 level fixed by
`the tungsten oxide temperature) evaporated tungsten material
`can be transformed near the relatively cold discharge vessel
`wall into stable ternary tungsten compounds, e.g. WO2X2
`(X = halogen), see figure 6.
`In the hotter regions close to the lamp electrodes the
`oxy-halide molecules are decomposed. By these chemical
`transport processes, tungsten atoms are brought back to the
`lamp electrodes in a regenerative manner. This cycle prevents,
`or at least considerably reduces, wall blackening caused by the
`evaporation of tungsten from the electrodes.
`
`Energetiq Ex. 2087, page 3 - IPR2015-01377
`
`

`

`G Derra et al
`
`Figure 10. Temperature of plasma and quartz bulb, calculated with
`(right) an without (left) inclusion of radiative transfer. Please note
`that the same colour coding is used for the space inside the lamp and
`for the bulb, but the real temperature scale is largely different.
`
`The performance of UHP lamps is determined largely by
`the temperatures on the inside of the burner. The Hg pressure
`inside the lamp has to be higher than 200 bar to allow for
`good colour quality and high efficiency. This requires bulb
`temperatures above 1190 K at the coldest spot inside the
`lamp. At the same time the hottest parts of the quartz
`envelope have to stay cold enough (<1400 K) to resist the high
`pressure without deformation and to stay clear without any
`recrystallization. A sophisticated burner design is necessary
`to keep the temperature differences within these limits.
`Especially for a long life product,
`the temperature
`differences should be as small as possible. Owing to the
`strong convective energy transport in the lamp plasma, the
`temperature at the upper wall is considerably higher than that
`at the lower wall.
`Because the crucial temperatures on the inside cannot be
`measured directly, a thermal model is required for lamp design.
`A model of the lamp plasma is needed to predict the distribution
`of the thermal flux that heats the wall. The plasma model has to
`include heat transport owing to thermal conduction, convection
`and radiative transfer.
`The calculation of radiative transfer requires an enormous
`numerical effort, but is inevitable, because its neglect leads to
`wrong temperature profiles of the plasma and the [21, 27], as
`is shown in figure 10. The left-hand side shows the result of
`a model calculation without radiative transfer, while the right-
`hand side is based on a model including radiative transfer. As
`can be clearly seen, radiative transfer leads to a reduction of
`the top–bottom asymmetry introduced by convection. Both
`the plasma temperature distribution and the bulb temperature
`distribution show much smaller deviations between top and
`bottom when radiative transfer is included in the plasma model.
`In figure 11 the comparison with a spectroscopic measurement
`of the plasma temperature demonstrates the quality of the
`model.
`
`2.5. Electrical properties
`
`The burning voltage of a UHP lamp can be described by
`Ulamp = Uelec + Uarc = Uelec + a · ed · pHg,
`
`(2)
`
`where Uelec is the voltage drop at the electrodes, i.e. the sum
`of cathode and anode fall voltages, ed the electrode distance
`or arc gap, pHg the pressure and a is a constant. The higher
`
`Figure 8. X ray of the electrode rod with chemical attack by too
`much bromine. The inner bulb contour is indicated by the white
`dashed lines.
`
`Figure 9. Light output versus burning time of an UHP 100 W lamp,
`1 mm arc gap, operated in a hot environment of 250˚C. ((cid:1))
`integrated luminous flux, ( ) luminous flux collected in a typical
`projector system.
`
`owing to the halogen transport cycle. A breakdown of the
`chemical cycle, however, would lead to strong blackening,
`resulting in an overheating of the bulb and a very fast end
`of life. The collected lumens drop over life under the present
`experimental conditions. This is a consequence of electrode
`burn back and whitening of the bulb.
`In the past years, the continuous demand for ever higher
`brightness was satisfied by increasing lamp powers. The higher
`power loads lead to a decrease in the lifetime of these lamps
`to about 2000 h, which is a good compromise for professional
`applications. For the consumer market (cf figure 1), a longer
`lamp life becomes a key requirement again. We also see a trend
`where the size of micro-display projection TV (MD-PTV) is
`shifting towards screen diagonals >50 inches. This asks for a
`combination of longest life and high brightness and challenges
`the design of new UHP lamp generations.
`
`2.4. Burner models
`
`The design of higher power UHP lamps—while keeping the
`arc short—is guided by advanced lamp models [7, 21], which
`include the plasma, the plasma–electrode interaction and the
`thermal balance of the lamp envelope. This allows for a proper
`design of lamps operating close to the limits of the bulb and
`electrode materials.
`
`2998
`
`Energetiq Ex. 2087, page 4 - IPR2015-01377
`
`

`

`Review: UHP lamp systems for projection applications
`
`88 / (18 / Uarc + 1)
`efficacy 120W
`efficacy 150W
`efficacy 200W
`efficacy 250W
`
`80
`
`60
`
`40
`
`20
`
`luminousefficacy[lm/W]
`
` spectroscopy
` plasma calculation
`
`8000
`
`6000
`
`4000
`
`2000
`
`temperature T [K]
`
`-2
`
`-1
`
`0
`
`1
`
`2
`
`vertical position [mm]
`
`0
`
`0
`
`20
`
`40
`
`60
`arc voltage Uarc [V]
`
`80
`
`100
`
`120
`
`Figure 11. Plasma temperature as a function of vertical coordinates
`in the mid-plane as measured by plasma spectroscopy and
`calculated for a 120 W UHP burner.
`
`Figure 13. Luminous efficacy of UHP lamps as a function of arc
`voltage for research lamps operated at lamp powers between 120
`and 250 W. The solid line is calculated from equation (3).
`
`Figure 12. Burning voltage of UHP lamps as a function of the
`product of mercury pressure and electrode distance.
`
`Figure 14. Measured temperature along a UHP electrode
`(250 W, 1.3 mm), at full operation (green) and dimmed operation
`(200 W, blue).
`
`the burning voltage, the smaller is the relative contribution of
`the electrode losses, and a higher lamp efficacy will result.
`However, with decreasing arc length, the voltage drop over
`the arc is reduced and, relative to the total input power, the
`electrode losses become more important.
`UHP lamps show a clear linear dependence of the burning
`voltage on the mercury pressure pHg [26]. This is shown
`in figure 12 for three different types of burners designed for
`different lamp powers, the larger burners being used for higher
`power. All the lamps used here were research samples, where a
`large variation in electrode distance and pressure was possible.
`From figure 12 the electrode fall voltage Uc = 18 V and
`the slope a = 0.26 V mm
`−1 can be determined. With
`−1 bar
`these two parameters, the electrical efficiency of UHP lamps
`(cid:3)−1
`(cid:2)
`can be written as:
`εel = Parc
`
`flux we could then determine the radiation efficacy of the arc
`plasma, using equation (3). The result is shown in figure 13.
`The straight line is fitted to the experimental results with
`the plasma efficacy ηplasma = 88 lm W
`−1 as a fit parameter. The
`plasma efficacy is slightly dependent on the power on which
`the lamp was operated: for higher powers, it is slightly higher
`than for the same burners operated at lower powers.
`
`2.6. Electrode design
`
`A long-life lamp requires electrodes with a very stable shape:
`the electrode tip should not recede (burn-back) or move
`laterally, owing to evaporation and transport of tungsten.
`These phenomena depend mainly on the temperature of the
`electrode [24,27]. Therefore, long-life electrode design aims at
`controlling electrode temperature. The optimum temperature
`distribution features a moderately hot electrode body (for little
`burn-back but sufficient cooling by radiation) and a hot tip (for
`a stable arc attachment).
`More detailed criteria can be deduced from systematic
`life tests of
`lamps with known electrode temperatures,
`flanked by model simulations [24]. Figure 14 shows the
`
`2999
`
`,
`
`(3)
`
`= Uarc
`Uarc + Uc
`
`=
`
`1 +
`
`Plamp
`
`69.2
`pHg · ed
`where pHg is in bar and ed in millimetres. As an example, for a
`UHP lamp with pHg = 200 bar and ed = 1 mm, the electrical
`efficiency is εel = 0.74.
`The burners from the lamps shown in figure 12 were
`measured in an integrating sphere. From the total emitted light
`
`Energetiq Ex. 2087, page 5 - IPR2015-01377
`
`

`

`2*30°
`
`2*15°
`
`integrating
`sphere
`
`aper-
`ture
`
`Figure 16. Measurement set-up to measure collected light flux
`versus etendue with a variable aperture in front of an integrating
`sphere. With decreasing aperture diameter (smaller etendue), the
`30˚ rays would be cut off, because they were then falling outside the
`collection etendue.
`
`0.5mm
`0.75mm
`1.0mm
`1.25mm
`1.5mm
`1.75mm
`2mm
`meas 1.3mm
`
`10
`
`20
`
`30
`
`40
`
`50
`
`Etendue
`
`6000
`
`5000
`
`4000
`
`3000
`
`2000
`
`1000
`
`collected lumens / 100W
`
`0
`
`0
`
`Figure 17. Simulation of collected luminous flux (in lumen) as a
`function of optical system etendue (measured in mm2 ster) for
`several arc lengths and a measurement for a 1.3 mm arc.
`
`of the display area and the solid angle defined by the possible
`transmission angles through the display.
`Because the etendue is at best conserved throughout any
`optical system, only light from the lamp within the etendue
`of the display system can be used. The light collection for
`a given source inside a certain etendue can be measured,
`e.g. through apertures of variable diameter [11], or simulated
`[18]. A simple measurement set-up is shown in figure 16.
`To predict the performance of future UHP lamps, figure 17
`shows the results of ray tracing simulations, with the simplified
`assumption that the light-technical data scale with the electrode
`gap. Nevertheless, the simulated numbers agree quite well
`with measurements for the electrode distance of 1.3 mm also
`shown in the figure.
`Clearly, a short arc is very important especially for low
`etendue projection systems, i.e. systems with small display
`size. Current projector designs use displays as small as
`0.5 inch, the system etendue is typically about 5–25 mm2 ster.
`
`2.8. Stable arc
`
`It is well known that the lamp electrodes of all short arc lamps
`(UHP, Xenon, Metal Halide) change their shape after some
`ten or hundred hours of operation. The electrodes have to
`
`G Derra et al
`
`Figure 15. Two different front geometries (false coloured intensity
`image, dimensions in micrometres), leading to different heat loads
`of 15.9 W (left), 13.2 W (right).
`
`measured electrode temperature in a 250 W UHP lamp. The
`cylindrical tip is partly molten (>3685 K), and the radiator
`body temperature is about 3200 K at the massive part down to
`2500 K at the coil.
`Given a target electrode temperature, one must also be
`able to predict the temperature expected for various design
`alternatives. Here,
`the most difficult part
`is to predict
`accurately how much power Pin the electrode will receive from
`the plasma and where the heat load will be localized. Under
`−1 of lamp
`UHP conditions, Pin is particularly high ((cid:1)10 W A
`current), and the underlying plasma–electrode interaction is
`less well understood than in other discharges. Therefore, we
`have determined Pin from the measured electrode temperature
`for a variety of conditions.
`For example, apart from depending on current, Pin is found
`to be a function of the shape of the electrode front. In figure 15,
`electrode temperature and heat load have been measured for
`one electrode at a current of 1.6 A, before (left) and after
`(right) the growth of a tip on the electrode front (the growth
`is favoured by the Philips pulse current scheme discussed in
`section 2.8). In the former case, Pin = 16 W, while Pin = 13 W
`for the latter—the sharper front reduces Pin. This effect must
`be taken into account to predict electrode temperature and to
`assess design alternatives. Also, the shape dependence helps
`to estimate where the heat load hits the electrode.
`To summarize, knowing the magnitude and distribution
`of Pin, we can predict the electrode temperatures for different
`designs and select
`the one with the maximum expected
`electrode stability and lifetime.
`
`2.7. Optical system performance
`
`The optical requirements of a projection system can be
`translated back to a requirement for the lamp via the etendue
`formalism [5, 9, 11, 18]. Etendue, sometimes called optical
`invariant, is a quantity describing the ability of an optical
`system to conduct light. In any optical system, the etendue
`can be at best conserved but never decreased. Etendue at any
`location in the optical train is defined as the product of the
`beam cross-sectional area at that location and the solid angle
`spanned by all rays passing through that cross-sectional area.
`In a properly designed projection system the display is
`limiting light throughput by its size and acceptance angle. The
`etendue of the projection system is then given by the product
`
`3000
`
`Energetiq Ex. 2087, page 6 - IPR2015-01377
`
`

`

`Review: UHP lamp systems for projection applications
`
`standard 90Hz
`
`300-500 Hz operation
`
`Philips UHP-S
`
`1.0000
`
`0.1000
`
`0.0100
`
`0.0010
`
`0.0001
`
`probability
`
`0
`
`35
`
`70
`
`105
`
`140
`
`175
`
`210
`
`burning hours
`
`Figure 19. Probability of arc jumps of more than 100 µm during
`lamp life for different operation modes. The red area indicates the
`Philips proprietary UHP-STM operation mode. With the other
`operation modes, long periods of very unstable operation will occur.
`The photos in figure 18 have been taken at 13 and 61 h burning time
`for this lamp at 90 Hz operation.
`
`The recordings of the arc attachment position shown
`before can be used to calculate a ‘jump width’, where the
`probability P (µ) of a jump width µ is defined as the relative
`number of measurements showing a jump width of at least µ.
`A typical development of arc jump probability during lamp
`life is shown in figure 19 for different operation modes. With
`the standard 90 or 300–500 Hz square wave current operation,
`the arc jump activity varies between relatively quiet periods
`and probabilities up to 30% for jumps >100 µm, depending
`on the constant change of the electrode surface. Quiet periods
`are always correlated with a pointed front surface, while arc
`jumping indicates a flat structure. The operation mode related
`to the red area shown in figure 19 will be discussed later in
`this section.
`As experiments with large numbers of lamps have
`proved, any lamp will show arc attachment instabilities in an
`unpredictable way during several hundred hours, no matter
`whether it is more or less stable at the beginning of lamp life.
`Using special electrode shapes like a pointed shape does
`not work in practice for short arc lamps. During the lamp’s
`lifetime, the large electrode surface modifications will change
`any initial electrode surface into a rough and sometimes flat
`structure. A lamp design which allows stable arc attachment
`only for the first 10–100 h is no solution for projection
`applications.
`For a proper optical design of the projection system the arc
`jump widths have to be known. A cumulative distribution of
`the probability of arc jumps is shown in figure 20. The figure
`indicates that in the standard situation, jumps of up to 100 µm
`occur in 13% of the burning time. Jumps of above 300 µm do
`not occur.
`The moving arc affects the light distribution on the
`display and therefore causes brightness variations on the
`screen. With typical jump frequencies of 0.01–10 Hz they
`attract the attention of the observer. Even with a uniform
`illumination, fluctuations of the total screen brightness may
`be the consequence of arc instabilities. It is nearly impossible
`to define a general threshold above which brightness variations
`are perceived as disturbing. We experienced visual (and
`therefore disturbing) effects for brightness variations above
`1% at any place of the screen.
`
`3001
`
`Figure 18. Electrode shape (grey) of the same lamp after 13 and
`61 h of burning. The contours and black dots indicate the arc
`positions as described in the text.
`
`deliver a high current into a very small discharge volume of
`below 1 mm3 in the case of the UHP arc. The stability of these
`electrodes in direct contact with a 7000 K hot plasma is one of
`the challenges for a short arc lamp with long life.
`In the UHP lamp the electrodes are operated at
`temperatures close to the melting point of tungsten (above
`3500 K). No matter which perfect electrode shape is used
`initially, a rough electrode front surface develops after some
`burning time. The arc plasma will change its attachment point
`on this surface frequently. The stability of arc attachment
`depends on the actual microstructure of the electrode front
`and varies, therefore, in frequency and amplitude during lamp
`life. Such a moving arc affects the light distribution on the
`display and therefore causes disturbing brightness variations
`on the screen [10, 12].
`UHP lamps were observed during their total burning time
`with a CCD camera set-up. The position of the arc and the
`two hot spots in front of the electrodes are recorded every few
`seconds. A detailed photo of the arc and electrodes is stored
`every hour. This experiment allows the tracking of any arc
`movements directly and correlating them with the status of the
`electrode surface.
`Figure 18 shows typical pictures of the electrodes after
`some ten hours, burning time, taken from the same lamp.
`The shape of the electrodes is shown as grey areas while the
`black contours indicate the brightness distribution of the arc
`at the same time. The varying positions of arc hot spots were
`recorded throughout the last hour of burning before taking the
`pictures and are indicated by black dots at the electrode front.
`After 13 h, the arc has attached on a lot of different
`positions distributed over the flat front part of the electrode.
`After some further burning hours, the electrode surface has
`changed to the more pointed shape shown in the second photo.
`The arc position is much more stable at this time. However,
`this situation will not last for long but will return to the previous
`one later on.
`
`Energetiq Ex. 2087, page 7 - IPR2015-01377
`
`

`

` standard situation
`
`stabilised
`
`50
`
`100
`
`150
`
`200
`
`250
`
`jump width [µm]
`
`Figure 22. Relative lamp current versus time which guarantees a
`stable arc attachment.
`
`G Derra et al
`
`0.50
`
`0.40
`
`0.30
`
`0.20
`
`0.10
`
`cumulative probability
`
`0.00
`
`0
`
`Figure 20. Cumulative arc jump distribution in standard and
`UHP-STM operation mode.
`
`Figure 21. Simulation of the effect on total screen brightness of an
`arc jump of 100 µm. The arrows indicate etendue values of typical
`LCD and DLP based display systems with diagonal size given in
`inches.
`
`The arc attachment instabilities occur with all kinds of
`short arc lamps at most times in life. Sophisticated electrode
`designs only help for the first hours in life and are therefore no
`solution. To avoid complaints in the market, all set makers use
`some kind of optical integration to suppress the visibility of arc
`jumping. The general working principle of optical integrators
`is to split the beam of light from the lamp into a large number
`of sub-beams which are superimposed on the display. Any
`fluctuations from beam to beam are averaged out.
`Two types of integrators are most common: lens array
`integrators [15], mostly used in combination with LCDs
`and rod integrators [16] normally used with DLPTMdisplays
`(‘Digital Light Projection’, trademark of Texas Instruments.
`DLP displays are arrays of small micro mirrors which can be
`tilted for image generation).
`Optical integration has been a good measure against the
`visibility of arc jumping for the past years. For display
`sizes smaller than 1 inch, however, even if the integration still
`preserves a good homogeneity, arc instabilities become visible
`again as a fluctuation of total image brightness [10]. Figure 21
`shows the simulated effect on screen brightness in the case of
`a typical arc jump width of 100 and 200 µm (cf figure 20).
`Hence,
`the feasibility of optical integration becomes
`limited for
`small etendue,
`i.e.
`small displays.
`An
`
`3002
`
`alternative solution, preferably at the source itself, is urgently
`needed.
`The physics of the arc attachment modes in high pressure
`arc lamps is a very complex problem and no general theoretical
`model is available so far. Detailed investigations, however, led
`to a way to stabilize the lamp arc directly without reducing the
`overall performance of the lamp.
`The stabilization of the arc during the total lifetime is based
`on an electronic drive scheme and was invented by Philips
`It was realized in the UHP-STM electronic
`in the mid-90s.
`driver using a specially designed lamp current [2], shown
`schematically in figure 22.
`The solution is closely related to the interaction of driving
`mode and arc attachment. Contrary to the standard ‘block’
`form current shape, the application of the special current form
`with extra current pulses at the end of each half-wave (see
`figure 22) results in a perfectly stable arc attachment on the
`electrode over the whole lifetime [10]. The arc position is
`fixed and the disturbing effects of a moving arc are successfully
`avoided.
`Figure 19 shows the probability of arc jumping with
`the standard block form current and the new current form
`for arc stabilization. Any instability is suppressed by more
`than a factor of 1000 and no disturbing visual effects remain.
`The cumulative probability shown in figure 20 indicates that
`the stabilization suppresses all arc jump widths down to our
`measuring accuracy of 20 µm.
`The large effect on the arc caused by the small pulse
`on the lamp current shows the close relation between lamp
`and electronics and the importance of a combined lamp and
`electronics development. The relation to the display system
`has to be addressed too: synchronization of the pulse current
`with the display frame frequency is implemented in the
`electronics. For all projection systems, it is, therefore, possible
`to use the extra light boost caused by the extra current.
`Because of its clear advantage with respect to picture
`stability over the whole lamp life, all other properties of the
`lamp being as good as before, the new current form has b