Freddy Manders, Marco Haverlag Philips Lighting Eindhoven

Paul Aben, Job Beckers, Winfred Stoffels Technical University of Eindhoven

Optical study of the breakdown phenomenon in High Intensity Discharge lamps

Outline

Introduction HID lamps Background approach

Experimental Set-up Power supply Lamps under investigation

Experimental results DC ignition of low aspect lamps DC ignition of High aspect lamps AC ignition

Conclusions

Introduction

Experimental set-up

Experimental results

Conclusions

Philips

One of the biggest lamp manufactures in the world

In Eindhoven pre-development of all lighting sources

• Halogen• Fluorescent• Compact fluorescent• LED• HID • etc

USED FOR:

High efficiency High colour rendering

PRACTICAL APPLICATIONS

Stadiums Tennis courts Parking lots Automotive Shopping malls Beamers

Burner is filled with:

- Starting gas (Noble gases, e.g. Xe & Ar)

- Buffer gas (e.g. mercury)

- Radiation emitting substance (e.g. Na, Ce, Dy, Ca, …)

Introduction

Experimental set-up

Experimental results

Conclusions

Background

• HID lamps Ignition voltage • ~ 4 kV (300 mBar, pulse ignition, standard HID lamps)• ~ 18 kV (10 Bar Xe, pulse ignition, Automotive lamps)

• Issues• 1 st electron, small lamps ( Automotive, UHP)• insulation materials (cables, lamp cap, etc)• Safety issues, bigger lamps• More expensive electronics

• The lamp should always ignited

We need to lower the ignition voltage of the gas discharge

Introduction

Experimental set-up

Experimental results

Conclusions

Approach

Try to understand how breakdown in HID lamps happens

Study breakdown process in model lamps with an ICCD camera and change a few parameters

• Gas type (Ar / Xe)• Gas pressure (300 mBar, 700 mBar )• Length / diameter of the lamp• Positive / negative voltage• Voltage source (Pulse, AC)

Introduction

Experimental set-up

Experimental results

Conclusions

Experimental set-up

Introduction

Experimental set-up

Experimental results

Conclusions

Experimental set-upLamps

Voltage source

Pulse source of 5 kV with a rise time of 10 nsec.

AC voltage source from 25 kHz up to 3 MHz

-4

-3

-2

-1

0

1

2

3

1 1 0 0 1 2 0 0 1 3 0 0 1 4 0 0 1 5 0 0 1 6 0 0 1 7 0 0 1 8 0 0 1 9 0 0 1

• Low aspect ratio burners– Argon, p = 300 mbar, d =

7 mm

• High aspect ratio burners– Diameter = 4 mm– Argon and Xenon– electrode distance: 1.5 cm

and 2.7 cm– 300 mbar and 700 mbar

Introduction

Experimental set-up

Experimental results

Conclusions

Vo

ltage

(kV

)

Experimental results

DC pulse ignition of low aspect lamps

DC pulse ignition of high aspect lamps

AC ignition

Introduction

Experimental set-up

Experimental results

Conclusions

Low aspect ratio: Argon 300 mBar, -4.0 kV

0 max

6 ns

7 ns

9 ns

10 ns5 ns

12 ns

13 ns

14 ns

15 ns

0 ns

3 ns

4 ns

Introduction

Experimental set-up

Experimental results

Conclusions

Grounded side

High voltage

side

21 ns

19 ns

18 ns

17 ns

Low aspect ratio: Argon 300 mBar, -4.0 kV

• High voltage ionization expands spherical• High voltage ionization is not uniform, channels are visible, streamers.• Grounded emission becomes, because of the interaction with the wall

9 nsec

Introduction

Experimental set-up

Experimental results

Conclusions

-4.0 kV -3.0 kV -2.5 kV

7 nsec

11 nsec

16 nsec

17 nsec

21 nsec

16 nsec

29 nsec

40 nsec

51 nsec

59 nsec

20 nsec

41 nsec

59 nsec

73 nsec

99 nsec

Low aspect ratio: Argon 300 mBar

Introduction

Experimental set-up

Experimental results

Conclusions

Low Aspect ratio: Argon 300 mBar, -2.5 kV

• High voltage emission region expands spherical• High voltage emission region is diffuse, burner is almost

completely filled with ionization• No grounded electrode emission is visible

59 nsec

Introduction

Experimental set-up

Experimental results

Conclusions

7 nsec

11 nsec

16 nsec

17 nsec

2 nsec

6 nsec

8 nsec

11 ns

39 nsec

60 nsec

99 nsec

128 nsec

-4.0 kV +4.0 kV +2.5 kV

Low aspect ratio: Argon 300 mBar

Introduction

Experimental set-up

Experimental results

Conclusions

Low aspect ratio: Argon 300 mBar

• In all 3 pictures mechanism is streamer• Negative streamers are in general more diffuse.• +4 kV: clearly streamer mechanism• +2.5 kV: slower and more diffuse

8 nsec16 nsec 99 nsec

+4 kV-4 kV +2.5 kV

Introduction

Experimental set-up

Experimental results

Conclusions

Conclusions low aspect ratio lamps

Introduction

Experimental set-up

Experimental results

Conclusions

• At high negative over voltage (-4 kV) discharge is a streamer

• For lower voltage there is a transition to a Fast Ionisation wave (-2 kV)

• Negative streamers are more diffuse then positive

• Discharge always starts at powered electrode because of interaction with wall

High aspect ratio burners

•Argon, p = 300 mbar, d = 1.5 cm, V = +4kV

   

52 ns 94 ns

8 ns 58 ns 99 ns

15 ns 69 ns 107 ns

23 ns 75 ns 109 ns

25 ns 84 ns 121 ns

39 ns 85 ns

93 ns48 ns  

   

0 nsIntroduction

Experimental set-up

Experimental results

Conclusions

High aspect ratio burners: Ar300 mBar, +4kV

• streamer along burner wall• very little branching• grounded emission expands spherical and is diffuse

85 nsec

Introduction

Experimental set-up

Experimental results

Conclusions

Ar 300 mBar Ar 700 mBar Xe 300 mBar

23 ns

52 ns

75 ns

85 ns

121 ns

109 ns

99 ns

23 ns

51 ns

71 ns

85 ns

115 ns

98 ns

89 ns

22 ns

39 ns

57 ns

85 ns

140 ns

135 ns

106 ns

High aspect ratio burners: +4kV

Introduction

Experimental set-up

Experimental results

Conclusions

• Ar300: little branching, intense emission at grounded electrode

• Ar700: much branching• Xe300: much branching, very little emission at

grounded electrode

High aspect ratio burners: +4kV

85 nsec, Ar 700 mBar

85 nsec, Ar 300 mBar

85 nsec, Xe 300 mBar

Introduction

Experimental set-up

Experimental results

Conclusions

97 ns 269 ns

9 ns 108 ns 272 ns

20 ns 143 ns 301 ns

21 ns 188 ns 303 ns

29 ns 212 ns 313 ns

39 ns 242 ns

257 ns54 ns  

   

0 ns

320 ns

340 ns

High aspect ratio burners: Ar 700 mBar - 4kV

Introduction

Experimental set-up

Experimental results

Conclusions

• High voltage ionization (negative voltage) very diffuse and at a very low intensity

• At grounded electrode streamers are formed• Two different mechanisms during one discharge

257 nsec

Introduction

Experimental set-up

Experimental results

Conclusions

High aspect ratio burners: -4kV

Conclusion high aspect lamps

Introduction

Experimental set-up

Experimental results

Conclusions

• More interaction with the wall because the distance to the wall is smaller

• Positive discharge branch more then negative

• Negative discharge is more diffuse and slower

•The higher the pressure the more branching

• Xe branches more then Ar

Comparison of velocities at + 4 kV

velocity / (106 m/s)

d = 1.5 cm d = 2.7 cm

Ar300 1.9 ± 0.1 1.6 ± 0.1

Ar700 1.3 ± 0.1 1.1 ± 0.1

Xe300 1.7 ± 0.1 1.4 ±0.1

Xe700 1.2 ± 0.1 1.0 ± 0.1

Introduction

Experimental set-up

Experimental results

Conclusions

Introduction

Experimental set-up

Experimental results

Conclusions

1.2

1.2

1.2

1.2

velocity / (106 m/s)

d = 1.5 cm d = 2.7 cm

Ar300 1.9 ± 0.1 1.6 ± 0.1

Ar700 1.3 ± 0.1 1.1 ± 0.1

Xe300 1.7 ± 0.1 1.4 ±0.1

Xe700 1.2 ± 0.1 1.0 ± 0.1

Comparison of velocities at + 4 kV

Introduction

Experimental set-up

Experimental results

Conclusions

1.4

1.4

1.4

1.4

velocity / (106 m/s)

d = 1.5 cm d = 2.7 cm

Ar300 1.9 ± 0.1 1.6 ± 0.1

Ar700 1.3 ± 0.1 1.1 ± 0.1

Xe300 1.7 ± 0.1 1.4 ±0.1

Xe700 1.2 ± 0.1 1.0 ± 0.1

Comparison of velocities at + 4 kV

Introduction

Experimental set-up

Experimental results

Conclusions

velocity / (106 m/s)

d = 1.5 cm d = 2.7 cm

Ar300 1.9 ± 0.1 1.6 ± 0.1

Ar700 1.3 ± 0.1 1.1 ± 0.1

Xe300 1.7 ± 0.1 1.4 ±0.1

Xe700 1.2 ± 0.1 1.0 ± 0.1

1.2

1.1

1.1

1.1

Comparison of velocities at + 4 kV

Conclusions velocity measurements

Introduction

Experimental set-up

Experimental results

Conclusions

• Out of the camera pictures it is possible to calculate the velocity of the discharge. The speed are all in the order of 10^6 m/s which is normal for a streamer discharge

• The velocity results for the high aspect ratio lamps with 4 kV show that there are some relations:

The factor the velocity changed

Distance between electrodes changed from 1.5 cm to 2.7 cm

1.2

Gas pressure changed from 300 to 700 mBar

1.4

Gas type changed from Ar to Xe

1.1

Breakdown voltage for AC ignition for high aspect lamps

0 50 100 150 200 250 300 350 4001400

1600

1800

2000

2200

2400

2600

Ign

ition

vo

ltag

e (V

)

Frequency (kHz)

300 mbar Xenon 700 mbar Xenon

-29%

-26%

Introduction

Experimental set-up

Experimental results

Conclusions

For DC pulse ignition ~ 4 kV is needed to ignited the 700 mBar Xe lamps so AC lowers the breakdown voltage with ~ 50 %

Experimental results (AC ignition - Xenon)

50 100 150 200 250 300 350 400 450

0.70

0.75

0.80

0.85

0.90

0.95

1.00

1.05N

orm

alize

d ig

nit

ion

vo

ltag

e

Frequency (kHz)

300 mbar Xenon 700 mbar Xenon

Two regimes in which the burners can ignite

Introduction

Experimental set-up

Experimental results

Conclusions

Experimental results of 300 mBar Xe (AC ignition)

Low frequencies: Discharge travels via the wall Looks like pictures of pulse ignition (streamer-like channels, branching)

High frequencies: Discharge travels through the gas Ionisation channel has the shape of a streamer …

30kHz

80kHz

140kHz

200kHz

300kHz

400kHz

Introduction

Experimental set-up

Experimental results

Conclusions

700 mBar of Xe

Maximum velocity of ionisation front: smv 4

max 10)3.04.1(

Shape looks like a streamer-like discharge Maximum velocity of the ionisation channel is in the order of those in Townsend discharges.

Contradiction:

300 mBar Xenon at 200 kHz

Introduction

Experimental set-up

Experimental results

Conclusions

Introduction

Experimental set-up

Experimental results

Conclusions

Conclusions on AC ignition

AC ignition voltage about 50-60% lower than pulse ignition voltage Ignition voltage is a decreasing function of frequency

At relatively low frequencies the ionisation channel travels along the wall At relatively high frequencies the ionisation channel travels through the gas

Ionisation channel builds up step-wise over many periods Ionisation channel only grows during voltage maximum

Possible explanation: Due to alternating E-field more charged particles stay in the volume, and are able to ionise for a longer time.

The higher the frequency, the more charged particles stay in the volume.

Thank you for your attention!