Automotive Incandescent Lamp Drive for 42V Powernet Martin O Hara & Dr Peter Miller

Transcription

1 Automotive Incandescent Lamp Drive for 42V Powernet Martin O Hara & Dr Peter Miller Taylors Road, Stotfold, England, SG5 4AY 1

2 Incandescent Lamp Driver 42V Filament too long and fragile 12V Filament cheap and available 2

3 Background 36V bulbs are unlikely to be used because their thin filaments mean lifetime would be unacceptably low in an automotive environment 12V bulbs are readily available. A Dc-Dc converter to step 42V down to 14V seems an expensive solution for bulbs (which could easily consume >300W) It would therefore be nice if a cost effective way could be found to power 12V bulbs from the 42V supply. 3

4 We want a cheap, low risk solution, so what do we already know? Operation at 50 and 60Hz AC is very common for bulbs and requirements are well understood Hz AC operation is claimed by VORSTENBOSCH ELECTRO to give longer life than DC operation. Flicker, audible noise and EMI are not issues at 50-60Hz AC. 4

5 Hz AC means bulb flicker is at 120Hz power in bulb at 60Hz 60Hz AC Voltage So for a PWM drive, 120Hz is identical to 60Hz AC 5

6 1 0.8 So we propose an ~120Hz PWM drive power in bulb at 60Hz 120Hz PWM Voltage 60Hz AC Voltage

7 120Hz PWM Match filament temperature (light output and colour) to DC case. Filament thermal mass is such that temperature is reasonably stable estimated ~ +/-35 o C ripple around ~2500 K burn temperature. 7

8 Filament Temperature Measurement 8

9 Thermal Test Schematic Switching Circuit PSU HP6268 B Pyrometer Pulse Gen W C 9

10 Automotive Bulb Temperature-Power Curve Steady State Operation 10

11 What about the switch on surges? When a bulb is cold it has a low resistance 5W lamp is ~3.6Ohm at ambient, ~ 35Ohm when lit. When first switched on a large surge flows This will be 3 times as large with a simple 36V PWM compared to 12V DC. This large current could rupture the filament. 11

12 Temperature-Resistance Curves 12

13 Steady State Inrush at 12V 13

14 So we need to limit switch on peak current. First we created an electro-thermal simulation model for bulbs. We then used this to investigate various options The aim was to approximately match turn on times while keeping peak surge current <= that seen at DC. 14

15 Electro-Thermal Model 15

16 Glow Time and In-Rush Current 16

17 5W bulb PWM drive - switch on Lumens(%) Time (ms) Note: DC drive graph is almost identical. 17

18 PWM drive is known to cause EMI (?) Rate of change of current is key factor in determining EMI. Operation at ~120Hz (8.3ms period) means edges can be relatively slow without significantly effecting efficiency (the slower the edges the lower the efficiency). Can meet CISPR class 5 requirements (the hardest). While achieving >98% efficiency. Dc-Dc converters range from 90-98% efficiency. 18

19 PWM Pulse Shaping 19

20 Wideband Conducted EMI Curves 20

21 Conducted EMI Curves Band A Band B Band C Band D Band E 21

22 PWM Drive Efficiency Bulb Rating (W) DC Loss (W) Switching Loss (W) Total Loss (W) Efficiency (%) No heat sinking required in sample circuits. 22

23 Lamp Life Given the reduced peak inrush current and similar warm up profile it was not felt switching would be an issue. Given the same operating temperature and low ripple it was not expected that vibration or shock would be issues. Noise was not an found to be an issue at 120Hz. Overvoltage protection was built into the drive circuit. Testing is the only way to see the effect of PWM drive on lamp life. Life and lumens maintenance needed investigation. 23

24 4 bulbs tested at a time Lamp life testing 2 run at DC, 2 run from PWM Light output recorded on a data logger. PWM set to match light output (for average bulbs ). 2 accelerated voltages (same acceleration factors used on 42V and 12V [15.5 & 16V DC] ), continuous burn. Most testing on 5W bulbs to avoid need for forced cooling. Results referenced back to 12V hours using T 12 = T Vcc Vcc

25 Life Test Schematic PSU 42V 100kW LDR1 Bulb1 PWM1 Drive Pulse Gen LDR2 Bulb2 PWM2 Drive PSU 14V LDR3 Bulb3 LDR4 Bulb4 240kW 22 Channel ADC 10kW 25

26 Bulbs tested Rheinmetall Elektronik 12V 5W BA15S EBT No

27 Obvious Checks LDR s within +/-10% for same light intensity. First failures Each test has 4 bulbs, two DC two PWM. DC drive socket 1:55% first failure, socket 2: 45%. PWM drive socket 1:55% first failure, socket 2:45% Failures appear to be reasonable normal distributions. No pattern visible from 2 accelerated voltages. 27

28 5W Bulbs Weibull Plot DC Mean 4538hrs PWM Mean V hours 28

29 Lumen Maintenance Over Life PWM DC 29

30 NOTCHING PHENOMENON Life of a lamp is usually controlled by the evaporation of a normal tungsten filament when the lamp is operated in a stationary state. However, lamps with low current and long life sometimes come to the end of life because of other factors besides the evaporation of a filament. That is the phenomenon called "FILAMENT NOTCHING", and especially in the case of DC lighting, they are easily affected. "FILAMENT NOTCHING" is the stairs-like or sawtooth-wave-form uneven phenomenon which appears on the surface of a tungsten filament after lighting for a while. In the case of AC lighting, it appears in a part of the surface of a temperature gradient. In the case of DC lighting, it appears on the whole surface. Accordingly, it decreases the diameter of a filament unevenly. Consequently, the thinned part of a filament becomes abnormally high temperature and promotes the evaporation of tungsten,then life becomes shorter. Also, the thinner the diameter of filament is, the more a lamp tends to be affected? see next slide for SEM photographs. Average life in DC lighting is usually about 50% of that in AC lighting. Ref. 30

31 FILAMENT NOTCHING PWM drive. DC drive. 31

32 New capabilities Light up rate can be controlled slower rates may result in longer life, particularly for frequently lit bulbs could potentially reduce light up time (eg for brake lamps) may be able to speed up lighting and keep 12V dc lifetimes. Overvoltage protection is standard ( kill voltage tests become redundant). Quite simple to regulate the bulb voltage. This is good in practice, but makes accelerated life tests impossible. 32

33 Conclusions PWM driving of 12V bulbs from 42V is practical Key off lights (e.g. parking lights) either require a higher AmpHour 36V battery (and perhaps no 12V battery?) or some simple logic in the electronics to power them from the 12V battery when the engine is not running. Cost effective Cheaper than centralised Dc-Dc converter Low electrical and audible noise May even increase bulb life testing to date shows ~2X gain in life. 33

34 Automotive Incandescent Lamp Drive for 42V Powernet 34