Comparing LED and Tungsten Halogen Lamps
First published: 17th January 2012
Tungsten halogen lights have been available since the 1970's, they are an improvement on tungsten incandescent bulbs, and can provide intense illumination in a small package. Light Emitting Diodes (LEDs) of sufficient power for lighting applications have come down in price in recent years, and are being aggressively marketed as a replacement, particularly for spotlights and even tubular fluorescents. Tungsten halogen lights are said to be about 20% more efficient than ordinary tungsten ones, and LEDs are supposed to be highly efficient.
I had a circular ceiling light fitting (luminaire) with a 150W R7s tungsten halogen tube in frequent use, estimated to be consuming HK$500 of electricity per year. Dimming is a requirement, so a fluorescent would be unsuitable, even if a suitable circular one could be obtained. I wanted to investigate retro-fitting the luminaire, replacing the tungsten halogen with LEDs.
Four LED rings (12cm, 10cm, 8cm, and 6cm) and a 4.5cm LED disc were obtained. These have 5mm square surface mount LEDs (type 3020 or similar), arranged in strings of three with current-limiting resistors, so they can be used with a 12V supply. Additionally, the LED disc has a bridge rectifier, so it can be used with an AC supply.
These were mounted concentrically, and connected common anode. This array was powered using a 12V 2.1A switched mode power supply, passing though a 3 channel infra-red controller. The controller was designed for use with red, green and blue LEDs, for colour control, but all the LEDs used were white. The disc and 8cm ring were connected to one channel, the 6cm and 10cm ring to a second channel, and the 12cm ring to the third channel. Thus, different numbers of LEDs could be switched on by pressing different colour buttons on the IR controller.
Materials and Methods
The same equipment as described in Comparing Tubular Fluorescents was used. Additionally, the multi-meter was used to measure the D.C. current consumed by the LEDs.
Tests on LED Rings and Disc
Initial measurements were made of the voltages and currents of the array components.
|Single, surface mount LED||3V|
The rings use a 150Ω surface mount resistor for each string, the disc uses three 220Ω resistors in parallel for each string. The resistor loss is calculated from the measured current, and the bridge rectifier loss was measured.
|Size (cm)||Total LED||Strings||Resistor||Tot. I (mA)||I/string||bridge rectifier (V)||resistor loss (V)||LED Power|
Considering the whole array of four rings and the disc, there are 123 LEDs in total, drawing 807mA, which is 9.7W at 12V. However, only 7.3W is consumed by the LEDs themselves, the rest is lost as heat in the resistors and bridge rectifier. The array looses 25% of the input power in this way.
A single LED in the array is consuming between 52 and 67mW.
The tungsten halogen luminaire is in active use, and was not removed from it's location, there was therefore no measurement of the current or power consumption of the tungsten halogen light. The glass diffuser was in place for the test. The LED array was held adjacent to the tungsten halogen luminaire, and the PV cells were located directly below the lights. The LED array gave directional light, and cast harsh shadows. In addition to measuring when all the LEDs were on, the current and power consumption of the switched mode power supply was measured when all the LEDs were off. Two 100W incandescent bulbs were used as the bias load.
Considering the small differences in the rotation time, and consequently large relative measurement error, it was not considered worthwhile to measure the rotation time for the various sub-arrays.
|Distance from light fitting to PV cells||200cm|
|Number of LEDs||Light intensity (PV cell output, mV)||Current (Clamp meter, A)||Rotation time, test load + bias load (s)||Rotation time, bias load (s)|
|Light||Power (W)1||Lighting efficiency2|
|150W tungsten halogen||150||100%|
|LED array "off"||2.2||0%|
|LED array, maximum||11||115%|
Notably, the light output from the tungsten halogen luminaire is 12 times the output of the LED array. On this basis, an array of about 1476 5mm LEDs would be required for the same light output. If they are mounted at the same density as the current rings, this would require a circle 45cm in diameter.
|Switched mode power supply efficiency3||88%|
- 1 The tungsten halogen power is assumed to be the rated power of the tube. The LED array power is calculated from the clamp meter readings, the joule meter readings would give results even less favourable to the LED array, and might have a large degree of error due to the timing accuracy.
- 2 The lighting efficiency takes into account the light produced as well as the power used. Expressed as a percentage of the tungsten halogen value.
- 3 The switched mode power supply efficiency is calculated from the power input, derrived from the clamp meter reading, and the power output, calculated from the current passing through the LED array.
The power loss in the switched mode power supply is significant, and the power loss in the current-limiting resistors is high. How would this type of LED compare to the tungsten halogen, if these losses could be eliminated?
Using 1476 LEDs to produce the same light level, each consuming 67mW, the total LED power would be 99W, or 66% of the tungsten halogen rated power.
Sources of Error
All the sources of error mentioned in Comparing Tubular Fluorescents apply. Additionally, the tungsten halogen luminaire was measured with the glass diffuser in place, which might be expected to reduce the light output somewhat.
- Enough 5mm LEDs to provide the same light level as a single 150W tungsten halogen tube would be 45cm in diameter, and would not fit in the current luminaire.
- The power saving by using these LEDs powered in this manner is marginal, and would not justify the expense of so many LEDs.
- The power consumption in the switched mode power supply when the LEDs are off also reduces the overall energy savings, unless users are careful to always switch off the power supply when the light is not in use, instead of merely using the infra-red control to cut the power to the LEDs.
The use of simple current limiting resistors is the main cause of losses. An improved design would use closer matching between the forward voltage of the LED strings and the power supply. However, this would increase the sensitivity of the circuit to voltage fluctuations. A small decrease in input voltage to below the forward voltage of the LEDs would switch off the light, yet a small increase could send the LEDs into thermal runaway, with expensive results.
Active current limitation would seem to be indicated, but each string of LEDs would need dedicated control, adding expense. Making the LED strings longer, and the circuit input voltage higher, would reduce the number of strings required, counteracting this expense. It would also make the voltage drop across the control circuit a smaller fraction of the input voltage, reducing the power loss.
Achieving dimming while retaining the power efficiency would also be a challenge. A pulse-width modulated approach while limiting maximum current might be good.
- Large, very high power LEDs (1W or more) are probably better suited to general lighting applications.
- While LEDs appear energy-efficient on paper, the benefits can be easily lost by simplistic circuit design, especially for current limiting.