Controlling HID lamps by intelligent power electronics

Transcription

1 Controlling HID lamps by intelligent power electronics Geert Deconinck, Peter Tant K.U.Leuven-ESAT 8 November 2007

2 Outline discharge lamps role of ballasts for discharge lamps variable frequency high-voltage power supply for hot-restrike modelling of HID lamps cold breakdown experiments hot restrike experiments conclusions 2

3 Discharge lamps breakdown and arc anode between electrodes in tube collisions ionising / elastic / inelastic collisions Planck s law discrete spectrum h. f = W W 2 1 kathode ν 3

4 Discharge voltage vs. discharge current SPANNING OVER DE ONTLADING [V] overgangsgebied 300 ontlading Spanningsval in de positieve zuil abnormale glimontladingen boog Stroom in de ontlading [A]

5 Low pressure discharge lamps fluorescent lamps (TL) mercury, sodium, lm/w, 8000 hr Fluorescerend poeder Straling in het zichtbare gebied Ultraviolette straling compact fluorescent lamps energy saving lm/w, hr Kwikatoom Elektronen Elektrode (gloeidraad) 5

6 High pressure discharge lamps higher luminance compact discharge tube high intensity discharge (HID) lamps typical lm/w, up to hr Molydeen bandje Elektrode + emitter Kwarts ontladingsbuis Vulling (77mm kwik & 20 torr argon) Weerstand 10kΩ + hulpelektrode 6

7 HID lamp 7

8 Outline discharge lamps role of ballasts for discharge lamps variable frequency high-voltage power supply for hot-restrike modelling of HID lamps cold breakdown experiments hot restrike experiments conclusions 8

9 Role of control gear ballasts provide power supply correct starting and operating voltage and current o initiate & sustain arc discharge between lamp electrodes ignition: high voltage required (kv) limit current to correct levels discharge lamps have negative resistance ballasts, auxiliaries 9

10 Starter and ballast for TL-lamp starter condensator elektrode ballast 10

11 Ballast characteristics ballast factor BF = commercial ballast light output laboratory reference ballast light output power factor PF = total power input voltage input current lamp current crest factor total harmonic distortion peak current CF = RMS current 11

12 Ballast types passive magnetic ballasts core & coil at net frequency active electronic ballasts at higher frequency often integrated starter 12

13 Electronic ballast netfilter oscillator C1 L1 C2 HF C3 net gelijkrichter 13

14 Electronic ballasts operate at higher frequencies khz for low-pressure discharge lamps Hz for low wattage HID lamps khz for high wattage HID lamps higher frequency allows smaller size of coils avoid interference and resonance in arc no stroboscopic effects smaller, lighter, more efficient more ionised gas o flux % above 10 khz 14

15 Electronic ballasts compensate lamp characteristics at start-up: ignition (breakdown) + warm-up in steady-state sometimes separate start-up device higher voltage is less statistical lag time often many consequent start-up pulses typical HID ballast PFC (power factor correction) + H-bridge typically 400 Hz (no resonance) blockwave 15

16 Electronic ballast advantages: lamp protection can allow protection of lamp e.g. at end of life, to ensure that if inner tube breaks, no external arc is established based on measuring low or erratic voltages output short-circuit protection thermal protection within ballast internal fusing 16

17 Electronic ballast advantages (ctd( ctd.) better colour output colour output depends on operating point (power) o (e.g. ceramic HID) maintaining current for optimal operating point o e.g. ±200K over lamp life o also when lamp is ageing o also for incoming voltage changes (surges / sags) allows dimming continuous dimming for 50%-100% of lamp power o automatically after 15 warm-up period allows integration with domotics (IED) 17

18 Electronic ballasts disadvantages higher capital cost sometimes lower power quality (depends on components, e.g. PFC) harmonics filters required o but also for magnetic ballasts interference o filters required 18

19 Outline discharge lamps role of ballasts for discharge lamps variable frequency high-voltage power supply for hot-restrike modelling of HID lamps cold breakdown experiments hot restrike experiments conclusions 19

20 Power supply for HID lamps HID lamps require a high ignition voltage 1 to 4 kv in cold condition up to several tens of kv in hot condition, hot-restrike trend mercury-free HID lamps: higher ignition voltages characterization of (cold lamp) ignition properties = statistical analysis characterization of hot-restrike properties ballast design o output voltage, output voltage for a given restrike time given ballast: estimation of restrike time, 20

21 Approach power electronics power supply continuous sine-wave output voltage adjustable frequency (<300 khz) variable amplitude ( <15 kv) low harmonic contents, no switching noise research purposes control and protection mechanisms automated measurements of hot-restrike characteristics 21

22 Test setup 22

23 Test setup asymmetrical H-bridge LC resonance circuit comprising T, L and C high sinusoidal voltage across C 23

24 Test setup lamp connected in parallel with C high-bandwidth, high-voltage 1:1000 probe Rogowski coil current sensor 24

25 Test setup switching rate controlled by pulse generator adjust to resonance frequency of LC circuit 25

26 Test setup DC bus voltage output voltage amplitude programmable waveform generator 26

27 Test setup optional resistor R lim limits breakdown current (omitted when LC tank energy is small) 27

28 Test setup DSO: records voltage, current and timestamp at each breakdown 28

29 Test setup Res. Diss. Res. detect the first breakdown event, and inhibit further control pulses Res. Diss. Off ENABLE 29

30 Test setup lamp ballast in series with the igniter circuit 30

31 Outline discharge lamps role of ballasts for discharge lamps variable frequency high-voltage power supply for hot-restrike modelling of HID lamps cold breakdown experiments hot restrike experiments conclusions 31

32 Test procedure cold breakdown experiments amplitude waveform generator produces repeating linear ramps ramp rate (kv/s) when breakdown occurs: a scope image is recorded further pulses are blocked after given sample time (5s), voltage ramp restarts 32

33 Measurement results cold breakdown experiments context 39 W metal halide lamp room temperature, f RES = 50 khz ramp rate = 762 V/s (slow) 300 measurement samples 33

34 Measurement results cold breakdown experiments discussion distribution of breakdown voltage: long right tail (not a normal distribution). a free electron must be available statistical time lag between exceeding min. V BD and actual breakdown 34

35 Measurement results cold breakdown experiments 762 V/s 1550 V/s 35

36 Outline discharge lamps role of ballasts for discharge lamps variable frequency high-voltage power supply for hot-restrike modelling of HID lamps cold breakdown experiments hot restrike experiments conclusions 36

37 Test procedure hot restrike experiments lamp burns at nominal power for 15 min. at t = 0, the lamp is switched off output voltage rises until lamp ignites when breakdown occurs: a scope image is recorded further pulses are blocked 37

38 Measurement results hot restrike experiments 39W metal halide arc tube only f RES = 50 khz, ramp rate = 4.4 kv/s (slow) - High initial V BD - High statistical spread < Steady state V BD Steady state V BD 38

39 39W MHD lamp Measurement results hot restrike experiments arc tube + jacket, single-ended f RES = 50 khz, ramp rate = 4.4 kv/s (slow) External breakdown < Steady state V BD Steady state V BD 39

40 Measurement results hot restrike experiments 39W MHD lamp f RES = 100 khz, ramp rate = 348 V/ms (high) 40

41 Outline discharge lamps role of ballasts for discharge lamps variable frequency high-voltage power supply for hot-restrike modelling of HID lamps cold breakdown experiments hot restrike experiments conclusions 41

42 Conclusions versatile & simple power supply for testing purposes output: high voltage & continuous wave avoid saturation of output inductors avoid excessive power dissipation in output capacitor multiple, subsequent lamp breakdowns avoided lamp temperature and electrodes are affected detection of breakdown voltage ramp rate is an important parameter lower ramp rate = o lower mean breakdown voltage o less statistical spread 42

43 Questions? 43