US3771013A

US3771013A - Lighting system including positive volt-ampere discharge lamp

Info

Publication number: US3771013A
Application number: US00147987A
Authority: US
Inventors: W Roche; J Waymouth
Original assignee: GTE Sylvania Inc
Current assignee: GTE Sylvania Inc
Priority date: 1971-05-28
Filing date: 1971-05-28
Publication date: 1973-11-06
Anticipated expiration: 1990-11-06
Also published as: JPS5522916B1

Abstract

A lighting system including a fluorescent lamp having a fill gas and pressure characterized by a high electron mobility for providing a static volt-ampere characteristic which includes a substantial region of positive slope. A power supply comprising an AC voltage divider, rectifier and filter capacitor is used to dynamically operate the lamp substantially within the positive region of its volt-ampere characteristic, and a series connected saturated transistor amplifier maintains the positive volt-ampere mode of operation by preventing the peak operating current of the lamp from exceeding a predetermined maximum level.

Description

[ NOV. 6, 1973 *[22] Filed:

[ LIGHTING SYSTEM INCLUDING POSITIVE VOLT-AMPERE DISCHARGE LAMP [75] Inventors: William J. Roche, Merrimac; John F.

Waymouth, Marblehead, both of Mass.

[73] Assignee: GTE Sylvania Incorporated,

Danvers, Mass.

May 28, 1971 [21] Appl. No.: 147,987

[52] US. Cl. 315/101, 3151241 [51] Int. Cl. H05b 39/00 [58] Field of Search 3l5/DlG. 7, 283,

[56] References Cited 3,629,648 12/1971 Brown et al. 315/DIG. 7

Primary Examiner--Nathan Kaufman Azt0rneyNorman J. OMalley, Edward J. Coleman and Joseph C. Ryan 57 ABSTRACT A lighting system including a fluorescent lamp having a fill gas and pressure characterized by a high electron mobility for providing a static volt-ampere characteristic which includes a substantial region of positive slope. A power supply comprising an AC voltage divider, rectifier and filter capacitor is used to dynamically operate the lamp substantially within the positive region of its volt-ampere characteristic, and a series connected saturated transistor amplifier maintains the positive voltampere mode of operation by preventing the peak op- UNITED STATES PATENTS eratmg current of the lamp from exceeding a predetermined maximum level. 3,448,335 6/1969 Gregory et al. 315/DIG. 7 y 3,457,458 7/1969 Paget et al. 315/DIG. 7 7 Claims, 7 Drawing Figures PAIENTEDuuv 6 I975 3771.013

I F 'I "2 i i FIG.2 FIG.3

WILLIAM J. ROCHE JOHN F. WAY MOUTH INVENTORS avg m ATTO N EY "BY 5 1973 f7 1 O1 3 SHEET 2 UP 3 44 8- Ac Voltage 'NPUT Multiplier 48 32 22 s i 2? I I8 1 t i I I #l U T myst c? Rectifier Filter t 3 4 T I I 28 Filter I Saturation mplifier 3s Bias 4o v FIG? WILLIAM J. ROCIHE JOHN E WAYMOUTH INV T EN 0 5 BY (20% ATTOR EY PATENTEDHUY 6 191a 3.771.013 SHEET 30F 3 300 v 1 n "5 f T Y 11 0/ X 200 0 U 11 m: O C P. 31 .2 3 n: .3 8

E D E 2 5xl0 l0 ELECTRON MOBILITY LL -(m%/OLT SEC.)

PVA MODE STATIC LAMP OPERATING 9 TRAJECTO CHARAC/l ERISTIC FIG 100 200 300 LAMP CURRE-NT (m0) WILLIAM J. ROCHE JOHN F. WAYMOUTH Lump Voltage (Volts) ATTOR EY LIGHTING SYSTEM INCLUDING POSITIVE 'VOLT-AMPERE DISCHARGE LAMP BACKGROUND OF THE INVENTION This invention relates generally to lighting systems including electric discharge lamps and more particularly to an improved lightingsystern in which the discharge lamp is operated in .a dynamic positive voltampere mode.

The starting and operation of electric discharge lamps, such as fluorescent lamps, presents two major problems. First, these and similar devices require a rel atively highvoltage to ignite an arc across the lamp as compared with :the voltage needed to maintain the are once it is ignited. Secondly, once thearcis ignitedit has anegative resistance characteristic causing it to tend to draw .increasinglymore. current .untilit reaches a runaway .condition. Both problems are generally solved by the use .of an inductive ballast in series with the lamp and the line terminals supplyingpowersto the-lamp. The voltage applied across the lamp by the ballast is adequate to maintain an ignited arc. The ballast steps up or gives an inductive .kick to the voltage, producing peak voltages above theline peak voltageand adequate .to strike thearc. The reactance of the ballast .then limitsthe current through thelampto its rated value.The objections to the useof ballasts are that they areheavy and bulky by reason of the large amount of copper winding .and iron core in theirconstructions. They:are expensive to make, and draw substantial power not use- .ful for lighting. Further, ballasts heat the environment of thelamp, andunless carefully constructed, generate .acousticand electromagnetic noise. In the case of DC ionized plasma. Hence, at very high power loadings,

.e,g. 5 to .25 kilowatts, a xenon llamp exhibits a positive resistance characteristic, since further increases in lamp current in the presence of theionized plasma can occur only with an increase in the electric field. In view Of this phenomenom'it is our understanding that for high power load applications, xenon lamps have been statically operated in a positive resistance mode withoutthe useof a customary type ballast. .Althoughit appears that such a mode of xenon lamp operation avoids the objectionable features of a conventional ballast,

orhter disadvantages are presented. Namely, positive resistance xenon lamp operation is limited to high power discharge lamp applications; it is restricted to a :single fill gas;.and it would appear to require a linevolt- .age regulator to assure acceptable lamp operation. It

will also be noted that the xenon :lamp still :requiresa high voltage .for ignition, .thus requiring someform of auxiliary circuitry.

SUMMARY OF THE INVENTION Accordingly,lit is an object of .thepresent invention to providean improved discharge lamp lighting system. the

It is another object of the invention to provide a lighting system in which an electric discharge lamp is operated in a substantially positive volt-ampere mode wherein the lamp discharge has substantially a dynamic 1 positive resistance characteristic and :thus does not require a ballast of the conventional type.

It is a further object of the invention to provide an electric discharge lamp having a static volt-ampere characteristic which includes a substantial region of positive slope well below the current level at which complete ionization occurs.

A further object of the invention is to provide a .circuit for operating a positive volt-ampere discharge lamp substantially within the positive region of its static volt-ampere characteristic for moderate power load applications.

It isyet another object of the invention to providea fluorescent lamp lighting system having significantly improved utilization efficiency.

Briefly, these objects are attained in a lighting system including an electric discharge lamp having a fill gas composition and pressure characterized by a high electron mobility to provide a static volt-ampere characteristic which includes a substantial region of positive slope. Thesystem further comprises a starter and a capacitive type voltage multiplier for igniting the lamp, a

power supply for dynamically operating the lamp substantially within the positive region of its volt-ampere characteristic, and means for preventing the peakoperating current of the lamp from exceeding a predetermined maximum level.

Ina preferred embodiment, the power supply comprisesa capacitive AC voltage divider, a full wave rectifier, a'filter capacitor, and means for connecting the filter capacitor and the lamp inan operating circuit loop. The filter capacitor is selected to maximize the average operating current of .the lamp by permitting closed loop current and voltage swings about the maximumof the static volt-ampere characteristic. The power supply and lamp then cooperate to provide dynamic stability by maintaining the equilibrium of this closed loop statically stable. A fail-safe transistor is serially connected in'the' lamp operating circuit loop andbiased to operate as a saturated amplifier when the operating current of the lamp lies below a predetermined maximum level and to clip lamp current swings exceeding the maximum level.

As a result of the dynamic positive volt-ampere mode of operation facilitated by the fluctuating DC voltage of the filter capacitor, the discharge lamp exhibits a dycurrent limiting resistor, the lighting system of the .invention exhibitsan extremely high power utilization efficiency. In addition, the present lighting system eliminates the acoustic and electromagnetic noise associated with inductive ballast devices, and provides a lamp operating circuit suitable for compact packaging. As compared to the statically operated xenon lamp system, the present invention has a much more general application as it provides a dynamic positive volt-ampere mode of operation at moderate power levels and allows flexibility in fill gas composition and pressure.

BRIEF DESCRIPTION OF THE DRAWINGS This invention will be more fully described hereinafter in conjunction with the accompanying drawings, in which:

FIG. 1 shows graphically the volt-ampere characteristics of a typical electric discharge lamp;

FIG. 2 graphically illustrates the effects of binary lamp voltage fluctuations and current perturbations on the negative slope of the characteristic curve of FIG. 1;

FIG. 3 graphically illustrates the effects of binary lamp voltage fluctuations and current perturbations on the positive slope of the characteristic curve of FIG. 1;

FIG. 4 is a block diagram of the lighting system according to the present invention;

FIG. 5 is a graph of maximum positive resistance lamp current vs. electron mobility;

FIG. 6 graphically illustrates the trajectory of a typical positive volt-ampere mode of operation of a system according to FIG. 4; and

FIG. 7 is a schematic circuit diagram of the system shown in FIG. 4.

DESCRIPTION OF PREFERRED EMBODIMENT To enable a full understanding of the construction and operation of the invention, an introductory discussion of the operating characteristics of electric discharge lamps will now be presented with the aid of FIGS. 1, 2 and 3.

FIG. 1 illustrates the volt-ampere characteristics of a typical electric discharge lamp. The solid line curve 10 represents the steady-state, DC, or static, characteristic and is the locus of those points (v,i) for which the rate of production of electron-ion pairs, n, exactly equals the rate of loss. For voltages above the steady-state characteristic, the ionization rate exceeds the loss rate, and the ionization density increases with time. For voltages below the steady-state characteristic, the loss rate exceeds the ionization rate and the electron density decreases with time. The dashed-line curves are contours of constant logarithmic time derivatives of ionization density, l/n.dn/dt, on the v, 1' plane. Thus, the steadystate volt-ampere characteristic 10 is the member (of the family of curves) for which l/n.dn/dt=0.

The region of conventional discharge lamp operation, commonly referred to as the negative voltampere characteristic, is that portion of the steadystate characteristic curve to the right of maximum, for which steady-state operating voltage decreases with increasing operating current. For that reason, electric discharge devices such as fluorescent lamps are said to exhibit a negative resistance characteristic and, therefore, require a current limiting impedance for proper operation.

For example, consider a lamp operating on a fixed voltage circuit at an instant of time at the point 12, having the coordinates (v i on the negative slope of curve 10. If a slight perturbation in current moves the operating point (v i +Ai) to the right, it will lie in the domain of l/n.dn/dt greater than zero, wherein electron density increases with time. At any fixed voltage, current is proportional to electron density, n, and therefore is increased with time also, moving the operating point (v i +Ai) further and further into the domain of positive dn/dt. Similarly, a slight perturbation of the opposite sign in current moves the operating point (v i Ai) into the region of negative dn/dt, and current decreases unidirectionally to zero.

The conventional ballast stabilizes the operating point of a discharge on the negative volt-ampere characteristic by including a series impedance such that if 1 increases by an amount Ai, the voltage across the discharge lamp terminals is reduced by an amount Av such that the point (v ,Av, i -l-Ai) falls below the steady state characteristic into the domain for which dn/dt is negative and electron density decreases with time. The current then decreases with time, restoring the operating point (m -Av, i +Ai) back toward the steady-state characteristic point (v i Similarly, a momentary negative perturbation Ai of current results in an increase in voltage +Av across the lamp terminals, moving the operating point from (v to (v +Av, i Ai) into the domain for which dn/dt is greater than zero. The resulting current increase with time restores the operating point back toward the DC characteristic (v i Thus, the conventional ballast circuit stabilizes the lamp operating point against current perturbations of either sign.

In the region where the volt-ampere characteristic has a positive slope, for example at point 14 on curve 10, however the lamp will run stably on a constant voltage circuit since increases Ai in current move the operating point into the region of negative dn/dt, returning the operating point toward the equilibrium value, and decreases Ai in current move the operating point into the region of positive dn.dt, resulting in an increase in current back toward the equilibrium point.

We have recognized that steady-state or DC stability is not actually required for lamp operation, but merely time average stability. For example, a circuit which alternately switches back and forth between two voltages, one of which lies in the domain of positive dn/dt, with the other in the domain of negative dn/dt, can trace out a closed loop in the v, i plane, as illustrated by A,B,C,D in FIG. 2 with respect to a negative sloping portion of curve 10. Starting at point A (voltage v current i,, and dn/dt 0) electron density and current increase to point B (voltage v current i at which point the voltage is switched down to v At (v i dn/dt 0, so that electron density and current decrease with time down to point D, at which point the voltage is switched to v Of course, in lieu of the rectangular loop for the simplified case of binary switching, a senusoidal voltage variation would produce a closed loop resembling a distorted ellipse. In either case, however, time average stability is maintained.

It is to be noted however, that if the voltage is fixed at v, or v the lamp circuit will not be stable.

Further, when operating a lamp in the negative voltampere mode, a minor perturbation on one cycle may introduce instability, as illustrated by the dashed lines in FIG. 2. For example, suppose the loop starts from a slightly higher current, as at point E. This starts the current increasing part of the loop from a point at which the value of l/n.dn/dt is, for example 1,100 sec." instead of 1,000 secf. Therefore, in a fixed time at v,,

the electron density will increase :along EF by more than AB, carrying the point at which the voltage is switched from v to v out to point P, further from B and E is from A. After switching to v at point G, the operating pointlies on the contour of l/n.dn/dt =800 sec. instead'of -l ,OOO secf, for example,so the :decrease of current during the transistion time is smaller than previously, carryingonlytoH and thence to I after switching from v to v,. .As a consequence, the loop does not close, and the starting pointfor a new loop at I lies on the contour of l/n.dn/dt 1,200 sec", for instance,and lis further from Athan was E. Thus, a perturbation fromequilibrium hasresultedin a shaftaway from, ratherthan back to the starting point, and this action continues to accelerate. It may readilybe-shownby similar arguments that perturbations of the opposite sign lead to accelerating shiftsin the opposite direction. Hence, acircuitbasedonthis principle can operate stably only if the switching between v and v is caused to take .place at fixed "values of current, i and i rather than at fixed times. This mode of operation, therefore,

requires active switching fordynamic stability.

In the positive-slope :region of characteristic curve 10, however, the aforementionedvoltage switching or fluctuating mode of operation is dynamically stable with respect toeither current ortime sensing, the latter not requiring active current control. This will now be discussed with reference to FIG. 3, in which the steepness of thetpositive slopingiportionof thecharacteristic curve hasbeenexaggerated for clarity. Considera similar perturbation in a circuit operating between i and v, in the loop ABCD. The perturbation starts a new loop at E instead of A. At E the value of l/n.dn/dt is,

for example, only 900 sec. instead of 1,000 sec. at

A, because the :point B isactually closer to thesteadystate characteristic than A. Accordingly, the increase in current BF during the fixed on time at v, is less than the increase AB, hencegpoint F is closer to B than E to A. Moreover, after switching down to v the transient point G is on a contour of, for example, l/n.dn/dt =-1 ,100 sec. instead of-l ,000 sec. 1 ,since it is farther from the steady-state curve than C. The resulting decrease in electron density and accompanying reduction in current along OH is greater than the decrease along CD. Thus, after switching back to v,, the point I must lie closer to A than B does, in a direction of restoring equilibrium. It may readily be shown by similar arguments that the same recovery action occurs in the case of a perturbation of the opposite sign.

The present invention takes advantage of both the positive and negative regions of the lamp characteristic for its operationLThat is, the lamp current loop encompasses regions of positive and negative 1/n.dn/dt on both sides of the maximum of the static curve of FIG. 1. The loop will close or remain dynamically stable so long as the equilibrium of the loop 'is statically stable, i.e., remains to the left of the maximum of the static curve of FIG. 1. This feature is a distinguishing facet of the invention in that a closed lamp current loop can be generated using passive circuit components, where the time duration of the current loop in regions of positive l/n.dn/dtto the rightofthe maximumon the static curve is the control factor, and not the current magnitude. This utilization of a passive circuit time constant in generating the lamp current loop will be further amplified in the remainder of the discussion.

According to one aspect-of the invention, an electric discharge lamp is provided which which the region of positive slope of the static volt-am1pere characteristic is to maximize power input to the lamp, it is desirableto operate at as high an-average current as possible.

Since the positive slope of thestatic lamp characteristic is very shallowin the region of the maximum, however, small perturbations in the line voltage can easily upset the equilibrium of the closed current loop as previously defined. Variations in the line voltage of only a few percent will shift the equilibrium of the closed current loop to the right of the maximum, resulting in an open-ended current loop and a current runaway as previously described. To compensate for this possibility,

the lamp power supply includes an AC voltage divider for significantly attenuating line voltage perturbations. In addition, to preclude the possibility of an undesired shift in the equilibrium of the closed current loop due to internal lamp instability, it is yet another aspect of the invention to include a fail-safe amplifier in the power supply circuit which insures that lamp current cannotexceedsome predetermined critical value.

Referringnow to FIG. 4, a lighting system according to the invention is shown as including a DC operated fluorescent lamp 16 having an anode probe 18 as one electrode and a cathode filament 20 as the second electrode. Typically, thelamp includes a sealed glassenvelope of tubular configuration, with the inside surface of the tube having a fluorescent phosphor coating. To support an electric discharge, the sealed glass tube is filled with an inert gas ormixture of gases at a pressure below atmospheric and a small amount of mercury, generally enough toprovide a low vapor pressure of between 3 and 15 microns during operation.

In accordance with the presentinvention, selected fill gas compositions and pressures are employed to provide a static volt-ampere characteristic having a substantially extended positive slope. A single gas parameter which takes into account both composition and pressure is the electron mobility of the fill gas. The electron mobility function, n, is partially determined by the mean free path of the electron ahd hence the lamp fill pressure. The electron mobility is also influenced by the energy lost in elastic type collisions multiplied by the frequency of these collisions. Both these factors are functions of the atomic weight of the gas atom.

The maximum positive resistance currents (I max.) for a wide spectrum of lamps are shown plotted as a function of ,u.,. in FIG. 5. The data points were obtained employing the following fill gas compositions and pressures in a 20WT12 fluorescent lamp:

Data point Lamp l 0.4 Torr. Ne:Ar/99 5:0.5 II 0.33 Torr. Ne:Ar/:30 Ill 0.35 Torr. Ar IV 1 Torr. Ne:Ar/99.5:0.5 V 1.5 Torr. Ne:Ar/70:3O VI 2.0 Torr. Ne:Ar/99.5:0.5 VII 0.36 Torr. ArzHe/zl5 VIII 2.0 Torr. Ne:Ar/70:30

IX 2.5 Torr. Ne:Ar/99.5:0.5 X 1.5 Torr. Ar

XI 2.5 Torr. Ne2Ar/70z30 XII 25 Torr. Ar

XIII 1.5 Torr. ArzHe/85zl5 XIV 2.5 Torr. ArzHe/85zl5 XV 3.2 Torr. Ar

Reducing the curve of FIG. to an equation, we can say that lamp 16 (FIG. 4) has a maximum positive resistance current which is related to the electron mobility function y as follows:

maz. 250 glo (1 2) 450 where he is the electron mobility of the gas at the current for a specified fill pressure. FIG. 5 demonstrates that the same value of I is possible with a number of gas compositions and pressures. Thus the lamp designer can select from a variety of combinations to suit his particular requirements. The gaspressure combinations yielding the highest values of and thus the highest values of I,,,,,,, are those involving low pressure neon; for example the fill gas of data point I provides a volt-ampere characteristic having a positive slope range which exceeds 275 milliamperes.

' In light of the previous discussion, a lamp possessing a high value of #8 will possess conditional dynamic stability in and of itself if it is operated from a specially adapted voltage supply. Hence, in accordance with another aspect of the invention, a power supply is provided for restricting the operation of lamp 16 to be substantially within the positive sloping region of its voltampere characteristic. As previously discussed, such a mode of operation, which shall hereinafter be referred to as the positive volt-ampere mode, provides the lamp with a positive resistance characteristic whereby conventional ballasting is not required.

In FIG. 4, the power supply is shown as comprising a voltage divider 22, a lamp voltage filter 26, a bias voltage filter 28 and a saturated amplifier 30 for inserting and maintaining the lamp in a positive volt-ampere mode. The divider network 22 has a pair of

input terminals

32 and 34 for connection to an AC power source, typically 120 volts 60 hertz line voltage, and reduces the AC input voltage to a level suitable for lamp operation. The reduced AC voltage is then converted to DC by rectifier 24 and stored in the filter 26. A lamp operating circuit loop is the provided by serially connecting lamp l6 andsaturating amplifier 30 across the output of filter 26.

During normal operation, the bias voltage for amplifier 30 is provided by circuit 36, which is connected to the rectifier output via filter 28. Separate filter 28 is employed for purposes of operating efficiency, as will be described hereinafter. Fail-safe amplifier 30 is biased to operate in a saturated condition when the operating current of lamp l6 lies below a predetermined maximum level and to clip lamp current swings exceeding that maximum level. In this manner, the power supply does not attempt to control or ballast lamp 16 within the positive volt-ampere mode, but merely monitors the boundaries of the positive volt-ampere mode to prevent a possible excursion into the negative voltampere region. The lamp operates in this fashion, drawing energy as required from the filter capacitor 26 and in effect controlling its own power consumption.

Lamp ignition is provided by means of a preheating circuit, a voltage multiplier, and a starter switch. The preheating circuit includes a switch 38, which when closed serially connects cathode electrode 20 and amplifier 30 across the rectifier 24. In addition an auxiliary bias circuit 40 is connected between switch 38 and amplifier 30 for changing the bias of the amplifier when switch 38 is closed to provide operation compatible with the preheating phase of the lamp ignition process.

The high starting voltage required for ignition is developed by a voltage multiplier having

input terminals

44 and 46 for connection to a source of AC power. Lamp 16 is connected across the output of the multiplier, with the anode 18 of the lamp being coupled to both the multiplier 42 and filter 26 via a starter switch 48. Preferably, switches 38 and 48 are functionally combined, as illustrated by the dashed line, in a doublepole momentary contact type switch, having normally open and normally closed positions.

Alternatively, switch 48 can be deleted entirely leaving switch 38 as the long starter switch, with some sacrifice in lamp life. For such a circuit modification, switch 38 can be replaced by an automatic starter, such as a glow bottle type switch.

When the AC supply voltage is connected across

terminals

44 and 46, prior to lamp operation, the voltage multiplier 42 will develop and store the required ignition voltage. In operation, the ignition phase is commenced by momentarily closing the normally open switch 38 to provide a current path for preheating the cathode filament 20. Simultaneously, closed switch 38 also activates bias circuit 40 to increase the bias applied to amplifier 30 to provide compatibility with the preheat current. Upon termination of the momentary preheat phase commencing when switch 38 is opened either manually or automatically, starter switch 48 will close to discharge through the lamp the high voltage stored in multiplier 42 and thereby ignite lamp 16. With lamp 16 conducting, multiplier 42 becomes inactive and subsequent operation of the lamp is sustained by the fluctuating DC output of filter 26. During this period, the saturated amplifier 30 functions to dampen the ignition transients and insert and maintain lamp operation in the positive volt-ampere mode by preventing the peak operating current of the lamp from exceeding a predetermined maximum level.

Filter 26 is selected to provide a predetermined amount of ripple for maximizing the average operating current of the lamp, in a closed loop as previously described. That is, filter 26 generates in conjunction with the lamp the desired time-varying voltage output which swings between the domains of l/n.dn/dt 0 and l/n.dn/dt 0 on both sides of the maximum in FIG. 1. A typical lamp trajectory (closed loop) provided by a system operating in a positive volt-ampere (PVA) mode is shown in FIG. 6. The curve is for a 20WTl2 lamp 24 inches in length and filled with 99.5% neon and 0.5% argon to a pressure of 2.5 Torr.

FIG. 7 schematically illustrates a preferred implementation of the lighting system shown in FIG. 4. The AC voltage supply 50 is connected to the power supply through a capacitor 52, which functions as the voltage divider 22 and thereby serves to match the voltage requirement of the lamp with the AC supply 50.

The value of capacitor 52 is determined by the following equation:

sz L/ L( m VL) where, P is the desired power level in the lamp, F is the AC supply frequency, V is the lamp operating voltage, V,', is the peak value of the AC supply voltage, and C, is the value of capacitor 52.

The reduced AC voltage so obtained appears at

nodes

1 and 2 and is then rectified by the full wave bridge circuit consisting of

diodes

54, 56, 58 and 60, which comprise the implementation of rectifier 24. Filter 26 is implemented by capacitor 62; hence, the rectified voltage appears across capacitor 62 at nodes 3 and 4, where it is filtered and then transferred to the lamp through two sets of diodes and a serially connected transistor biased to function as saturated amplifier 30. More specifically, the lamp operating circuit loop comprises the following components which are serially connected in the order named between nodes 3 and 4;

diodes

64 and 66, starter switch 48, lamp l6,

diodes

68 and 70, and transistor 72 having its collector electrode connected to the cathode of diode 70 and its emitter electrode connected to node 3.

To compensate for possible line voltage deviations, the ratio of the value of capacitor 62 (C to that of capacitor 52 (C is made as large as possible consistent with the overall circuit design constrains. Typically C is chosen to be at least an order of magnitude greater than C In this manner, perturbations in the line voltage will be attenuated by an amount log [C /(C +C )]db before reaching the lamp.

Transistor 72 is biased to operate in a saturated condition provided the lamp current does not exceed a preselected level. In the saturated condition, transistor 72 functions like a closed switch having essentially a zero impedance and, thus, essentially a zero voltage drop. The preselected current level, below which lamp current flows through transistor 72 unimpeded, is chosen to be conincident with the onset of the dynamic negative volt-ampere mode of operation; i.e., the empirically determined boundary between the dynamic positive volt-ampere and the dynamic negative volt-ampere regions. Since lamp 16 is designed to have an extended static positive volt-ampere characteristic, and since the power supply is designed to provide a fluctuating DC voltage at nodes 3 and 4 for operating the lamp about the maximum of the characteristic, the probability of violating the lamps dynamic positive volt-ampere boundary is minimal and the transistor will remain inactive and pass lamp current unrestricted.

To maintain transistor 72 in saturation for all lamp currents in the dynamic positive volt-ampere range, a base bias current must be supplied. The value of this base current is determined by the following equation:

'8' c max/B where, I is the maximum value of instantaneous lamp current allowable for maintaining the positive volt-ampere mode of operation, as predetermined empirically, and B is the common emitter current amplification factor for transistor 72.

During lamp operation, 1,, is supplied to the transistor from capacitor 74, which provides the function of auxiliary filter 28, through a base bias resistor 76, which performs the functions of bias circuit 36. The value of resistor 76 is determined by the following equation:

16 VH/IB 5 where, V is the peak capacitor voltage across capacitor 74, and I is the maximum desired base current as determined from equation (2).

It is desirable to supply 1,, from a separate capacitor 74 rather than from the main filter capacitor 62 for reasons of circuit efficiency. This results from the condition that the base current need not be constant to maintain transistor 72 in saturation since the lamp current will rise and fall with the small cylic voltage variations across capacitor 62. By selecting the value of capacitor 74 to be two orders of magnitude less than the value of capacitor 62, the voltage across capacitor 74 will rise and fall in synchronism with the lamp current. In this manner, the power loss in bias resistor 76 can be minimized with no sacrifice in transistor performance.

To isolate the operation of the bias circuit from the main filter 72, a blocking diode 78 is connected be tween nodes 4 and 5.

As with all practical filters, the DC voltage provided to the lamp across capacitor 62 at nodes 3 and 4 contains a certain amount of desired ripple. It is this AC ripple voltage superimposed on the DC voltage of capacitor 62 which allows a closedlloop mode of operation about the maximum of the static curve. The extent of the loop in terms of current swing is determined by the value of capacitor 62 in combination with the time varying resistance of the lamp; hence, there is provided a pseudo time constant C R(t), where C is the value of capacitor 62 and R(t) is the lamp resistance. Since R(t) is an extremely difficult function to define analytically, an emperical procedure is best suited for determining the optimum value of C R(t) in terms of maximizing the lamp power. This provides another design constraint on the value C in addition to that discussed previously with respect to its ratio to C for attenuating line voltage perturbations. In particular, filter capacitor 62 is selected to maximize the average operating current of the lamp by permitting current swings exceeding the lamp current at the maximum of the static curve in a manner whereby the fail safe transistor 72 is not actuated to clip the current swings. Of course, transistor 72 will still preserve the positive volt-ampere mode of operation by clipping any substantial lamp current swings exceeding the maximum level.

The circuit for preheating the lamp cathode electrode comprises switch 38, diode 80, and cathode filament 20, which are serially connected between node 5 and node 6. The function of bias circuit 40 is adapting the control of transistor 72 to the preheating phase is provided by diode 82 and resistor 84 which are serially connected between switch 38 and the base of transistor 72. Switch 38 is a normally open momentary contact type switch which is functionally coupled to the starter switch 48, as previously discussed with respect to FIG. 1. When switch 38 is closed during the lamp ignition process, a current path is provided from node 5 through switch 38 and diode 80, through the lamp cathode electrode 20 and returning through diodes 68 and and transistor 72 to node 3. Simultaneously, the closing of switch 38 also increased the base current in transistor 72 by providing low impedance shunt path via diode 82 and resistor 84 around the normal base resistor 76. The base current in transistor 72 must be increased during this phase of the starting process since in general the preheat current of most fluorescent coils exceeds the lamp operating current by approximately a 1.5:1 ratio.

The function of voltage multiplier 42 is provided by

capacitors

86, 88, 90 and 92. The multiplier input terminals comprise

nodes

1 and 2, and the multiplier voltage output is obtained at nodes 6 and 7. When the circuit is connected to the AC supply, and prior to the activation of the preheating circuit, a high voltage quickly builds up on the multiplier capacitors across the lamp. This capacitive voltage is limited in its energy content, since capacitive energy in general equals CV /2, and the capacitors of the multiplier circuit are made extremely small; for example C may differ from the value of each multiplier capacitor by a ratio of 150:1. Hence the lamp can not fully ignite until the cathode coil is raised in temperature to an emissive state. During the voltage build up in the multiplier circuit, there is only a minute current in the capacitor 52, since as previously stated, the multiplier capacitance is extremely small. As a consequence, the multiplier input at

nodes

1 and 2 sees essentially the full voltage value of the AC supply.

Upon the activation of the preheat circuit, when switch 38 is closed and switch 48 is simultaneously opened, the'high voltage of the multiplier circuit is temporarily removed from the lamp terminals at nodes 6 and 7 in order to prevent a premature ignition during the initial temperature rise in the cathode coil 20. Immediately following cathode preheating, however, when switch 48 is returned to the closed position and switch 38 is opened, the high voltage which was stored in the multiplier circuit is once again impressed across the lamp terminals at nodes 6 and 7, and, since the coils are at an emissive level, the lamp will ignite. After lamp ignition, the voltage requirement of the lamp is matched to the available supply voltage according to the relation of V and V, in equation (1).

During the charge build up phase of the multiplier, the high voltage is developed across capacitors 88 and 92. For example, if the peak AC input is 170 volts, each capacitor will develop a 340 volts charge which add together to yield a 680 volt starting potential appearing at nodes 6 and 7.

The circuit action resulting in the 340 volt potential across capacitors 88 and 92 can be described in the following manner.

Capacitors

86, 88, 90 and 92 will all be charged to the peak value of the AC supply 50 after one 60 Hertz oscillation. On succeeding cycles,

capacitors

86 and 90 will transfer charge to capacitors 88 and 92, respectively, and then be recharged in turn by the AC supply 50. The resulting voltage appearing across nodes 6. and 7 as a function of the number of cycles required for generation may be expressed as:

where, V, is the open circuit lamp voltage between nodes 6 and 7, V is the peak value of AC supply 50, and q is the cycle index number.

A closed form of the above series expression more suitable for analytic purposes is:

From equation (5) it is evident that V will attain approximately 97 percent of its maximum potential value after 5 cycles (q 5). Subsequent to lamp ignition (usually well within 5 cycles, or 85 milliseconds), the starting circuit remains dormant during nonnal lamp operation.

During normal operation of the circuit, the lamp 16 is connected across a zero impedance source, namely capacitor 62, and, therefore, must rely on its own plasma inertia to limit its own current. This the lamp does by means of its favorable gas composition and pressure. The gas composition and pressure are selected, as previously described, such that a closed loop hysteresis effect occurs between the lamp current and the electron density within the lamp, as illustrated with respect to FIG. 6. It is noted that a conventional negative volt-ampere type fluorescent lamp, i.e. a lamp having a positive characteristic limited to a range of aobut milliamperes, will not operate in an acceptable manner on the power supply described with respect to FIGS. 4 and 7.

By way of example, the following circuit parameters may be used in the embodiment of the lighting system shown in FIG. 7:

AC Supply 50 120 volts AC, 60 Hz. Capacitor 52 15 microfarad, 200 volts DC Capacitor 62 250 microfarad, 200 volts DC 0.1 microfarad, 200 volts DC 0.] microfarad, 400 volts DC I microfarad, 200 volts DC DTS 410 (Delco)

Capacitors

86 and 90 Capacitors 88 and 92 Capacitor 74 Transistor 72

Diodes

54, 56, 58, 60, 64 70, 78, and 82

Diodes

66 and 68 Resistor 76 Resistor 84 Switch 38-48 Si, 1A, 200 PlV Si, 1A, 400 PIV 10,000 ohms, 1 watt 330 ohms, 1 watt DP, momentary contact, NC and NO positions 24" Tl2 fluorescent lamp filled to pressure of 1.5

Torr with a gas mixture consisting of 70% neon and 30% argon by volume Lamp 16 NVA PVA Operation Operation 28.75 watts Pinput 26 watts 24.25 watts Lamp Power 24 watts 84.5% Efficiency 92.5% 400 ma Lamp current 350 ma 67.8 volts DC Lamp voltage 65.8 volts DC 0.6 lag Power Factor 0.4] lead It is apparent that the present invention represents an entirely different mode of operation for fluorescent lamps yielding significant advantages. In particular the PVA lighting system provides a means of operating fluorescent lamps without the requirement of the conventional current limiter or ballast. This reduces the size,

the system offers design flexibility and is capable of providing a PVA mode of operation for the more commonly employed lower power level applications.

Alternative ignition circuit implementations may be employed in lieu of the preheat circuit and multiplier shown. Further, in the event the efficiency provided by the use of auxiliary filter 28 is not important to a given application, the bias voltages may be obtained directly from filter 26. Also, saturated amplifiers other than the common emitter transistor arrangement may be employed, and of course, divider 22 is not required if a suitable voltage source is available. In fact, in applications where controlled operating conditions can otherwise be maintained, the fail-safe transistor switch may not be required.

We claim:

l. A lighting system comprising in combination: an electric discharge lamp containing a fill gas having a composition and pressure selected to provide a static volt-ampere characteristic having a substantial region of positive slope, the electron mobility function [1. of

said fill gas and the maximum positive resistance current I of said lamp substantially conforming to the relationship I,,,,,,, 250 log (11, 450; means for igniting said lamp; and a power supply for operating said lamp comprising a rectifier having input means for connection to a source of alternating current, a filter coupled to the output of said rectifier, means connecting said filter and said lamp in a direct current lamp operating circuit loop, said filter being selected to provide a predetermined amount of direct current voltage ripple in said lamp operating circuit loop to thereby maximize the average operating current of said lamp by permitting closed loop current and voltage swings about the maximum of said static volt-ampere characteristic, the extent of said current swings being determined by the reactanceof said filter and the time varying resistance of said lamp and being selected to maintain the equilibrium of said closed volt-ampere loop in said region of positive slope, and thus statically stable, to thereby operate said lamp in a dynamically stable positive voltampere mode, a fail-safe amplifier serially connected in said lamp operating circuit loop, and means for biasing said amplifier to operate in a saturated condition when the peak operating current of said lamp lies below a predetermined maximum level and to clip current swings exceeding said maximum level, said predetermined maximum level being chosen for inserting said lamp in said dynamically stable positive volt-ampere mode during ignition and subsequently maintaining said dynamically stable positive volt-ampere mode of operation.

2. A lighting system according to claim 1 wherein the input means of said rectifier includes a voltage divider, and said filter comprises a capacitor coupled across the output of said rectifier.

3. A lighting system according to claim 1 wherein said fill gas comprises neon at a fill pressure of not more than 2.5 Torr.

4. A lighting system according to claim 3 wherein said fill gas composition comprises not less than neon and not more than 30% argon.

5. A lighting system according to claim 2 wherein said lamp igniting means comprises a voltage multiplier having input means for connection to a source of alternating current and a starter switch connected between said multiplier and said lamp.

6. A lighting system according to claim 5 wherein said lamp has anode and cathode electrodes, said anode being coupled through said starter switch to both said multiplier and said filter capacitor, and wherein said lamp igniting means further includes means for preheating said cathode electrode comprising switching means for serially connecting said cathode electrode and said amplifier across the output of said rectifier, and means connected between said switching means and said amplifier for changing the bias of said amplifier when said switching means provides a closed circuit.

7. A lighting system according to claim 2 wherein said amplifier comprises a transistor has emitter and collector electrodes connected in said lamp circuit loop and a base electrode, and said means for biasing said transistor comprises a bias capacitor across the output of said rectifier, and a resistor connected between one terminal of said bias capacitor and the base electrode of said transistor, a blocking diode being connected between said filter capacitor and the: junction of said bias capacitor and said resistor, and said bias capacitor having a capacitance value substantially less than that of said filter capacitor whereby the power loss in said resistor is minimized.

Claims

1. A lighting system comprising in combination: an electric discharge lamp containing a fill gas having a composition and pressure selected to provide a static volt-ampere characteristic having a substantial region of positive slope, the electron mobility function Mu e of said fill gas and the maximum positive resistance current Imax. of said lamp substantially conforming to the relationship Imax. 250 log10 ( Mu e) - 450; means for igniting said lamp; and a power supply for operating said lamp comprising a rectifier having input means for connection to a source of alternating current, a filter coupled to the output of said rectifier, means connecting said filter and said lamp in a direct current lamp operating circuit loop, said filter being selected to provide a predetermined amount of direct current voltage ripple in said lamp operating circuit loop to thereby maximize the average operating current of said lamp by permitting closed loop current and voltage swings about the maximum of said static volt-ampere characteristic, the extent of said current swings being determined by the reactance of said filter and the time varying resistance of said lamp and being selected to maintain the equilibrium of said closed volt-ampere loop in said region of positive slope, and thus statically stable, to thereby operate said lamp in a dynamically stable positive volt-ampere mode, a fail-safe amplifier serially connected in said lamp operating circuit loop, and means for biasing said amplifier to operate in a saturated condition when the peak operating current of said lamp lies below a predetermined maximum level and to clip current swings exceeding said maximum level, said predetermined maximum level being chosen for inserting said lamp in said dynamically stable positive volt-ampere mode during ignition and subsequently maintaining said dynamically stable positive voltampere mode of operation.

4. A lighting system according to claim 3 wherein said fill gas composition comprises not less than 70% neon and not more than 30% argon.

7. A lighting system according to claim 2 wherein said amplifier comprises a transistor has emitter and collector electrodes connected in said lamp circuit loop and a base electrode, and said means for biasing said transistor comprises a bias capacitor across the output of said rectifier, and a resistor connected between one terminal of said bias capacitor and the base electrode of said transistor, a blocking diode being connected between said filter capacitor and the junction of said bias capacitor and said resistor, and said bias capacitor having a capacitance value substantially less than that of said filter capacitor whereby the power loss in said resistor is minimized.