Some fluorescent lights last a long time. Today I replaced one that lasted over 20 years with frequent use. The cost difference between rubbish and a good reliable fluorescent light is negligible. We need consumer laws to protect as from the resource wasting short lived and dangerous rubbish.
The light was used 4 - 6 hours per day when people are around and sometimes left on for 12 hours or more as a security light when everyone is out. A rough average of 50 hours usage per week multiplied by 50 weeks per year is 2500 hours per year; multiplied by 20 years is 250000 hours of use.
Update Sept 2010
I replaced a Philips Energy Saver 18 watt after 2 or 3 years of use. How long did it last? 3 hours a day, 4 days a week, 50 weeks a year, is 600 hours a year, less than 2000 hours in total. The globe is bent as if the base melted from heat. The globe was one of two in a fitting designed for 2 * 60 watt, or 120 watt in total. The energy savers are supposed to produce the same light and less heat but this energy saver appears to have produced more heat than a 60 watt incandescent globe.
I replaced the Philips Energy Saver 18 watt with a 15 watt globe from a different brand and the 15 watt globe produces more light. Modern energy saving globes are closer to what the original energy saving globes promised but are still not independently tested and certified for either energy use, colour, brightness, or long life.
The first thing to kill the light is loss of metal from the hot elements at each end. Slight changes in the design end thickness of the elements can change their life by ten or a hundred fold.
Do you save the light by switching it off?
Fluorescent lights are also damaged by switching off and on. This light was usually switched on and off once per day. Some days it was switched on and off a dozen or more times. A guess would be 20 times per week or a 1000 times per year, which is 20000 times during the 20 years of use. Manufacturers do not state the minimum number of times you can switch your light on and off.
Old style fluorescent lights had electrical starter circuits that burnt out quickly from switching off then on. You were better off leaving lights on instead of switching them off for a short time. Electronic starter circuits are more reliable. Occasionally a manufacturer will mention something about switching lights off then on. There is not compulsory labelling.
Think of a cupboard with a shelf light operated by a door switch. You might open the cupboard 50 times per day. One manufacturer stated their fluorescent light was good for 30000 starts. That is 600 days or less than 2 years. An incandescent or LED light would last many times longer.
New designs should be better
No matter what you do with the old style lights, they have to fit in an existing fitting with a separate ballast and starter element, limiting your design options. New fluorescent lights have the ballast and starter element built into the light giving you the freedom to design long lasting lights. The problem is a really long lasting light would put the manufacturers out of business. Deliberate bad design helps corporate profits.
Price is no guide
Some of the cheap new fluorescent lights last only a couple of thousand hours, about the same time as the old incandescent lights they replace. Many do not reach the design life printed on the packet and, even if they did reach the design life, they can be so expensive, they will still be more expensive than incandescent lights. The cost saving quoted for electricity is usually based on an expensive electricity rate higher than what we pay. In Australia, they would save money in remote country towns but not in the big cities.
Some of the new fluorescent lights quote 15000 hours and sell at a price where they pay for themselves in 5000 hours, giving you a good benefit. The difference between failing after a couple of thousand hours and lasting 15, 20, or 50 thousand hours is only a few cents worth of material and a few seconds of extra quality control. The retail price is absolutely useless as a guide to what you get. Some of the most expensive fluorescent lights are the least reliable. Some of the most reliable and long lived are little difference in price from the cheapest.
One of the replacement elements for the old light, the starter, is marked as containing no radioactivity. There was never any warning about radioactivity on the old starters but now we are told the new ones are not radioactive. This is typical of industry. Some smoke detectors had radioactive elements and there was no warning on the packet. Now there are smoke detectors using a more modern and accurate detection system. The new ones are marked as not being radioactive but the old radioactive ones are still on sale with no warning.
The new light is rated as producing 3250 lumens from 36 watts, or 90 lumens per watt of power consumed. Many fluorescent lights produce an uneven spread of light that does not work effectively for human eyes, which means you have to use a far more powerful light to see properly. The Philips tube says the light output has a CRI greater then 80.
CRI is a colour rendering index. Incandescent lights have an index of 100. Conventional fluorescents have indexes as low as 50. The best modern fluorescent lights with a triple phosphor mix are in the 80s. When you photograph human skin or anything with a pastel colour, such as apricots, you want a CRI of 100 for perfect reproduction. Primary colours used on posters, toys, and some furniture will look good at a lower CRI.
The new Philips fluorescent light is the first one to mention a CRI. Lots of lights mention colour temperatures or lumens but both are useless without a high CRI. The CRI should be a compulsory marking on all fluorescent lights.
What was life like with the old fluorescent light that I replaced? The light was in a kitchen. Cooking during the day was easy. Cooking at night sometimes resulted in food of slightly the wrong colour. Tomato appeared slightly darker than in real life. Tomato based dishes and sauces appeared very dark under the fluorescent light compared to daylight. I always cooked based on taste and smell, not colour, when cooking at night.
When I renovated my dining room, I changed the lights from energy saving fluorescent lights back to incandescent lights so my guests could enjoy the colour of their food.
In Australia there is no compulsory labelling of fluorescent lights for colour accuracy. I do not know of a compulsory standard test for life expectancy.
The CRI standard was always vague. There was a new recommendation in 1999 but no standard. What we need are standard colour indexes for accuracy and for safety.
Colour Index Accuracy should compare the light output across the visible range, from 380nm through to 740nm, to a theoretical perfect even source. The test should be in small increments, about one percent increments or 3.6nm. Make it bands of 4nm or 5mn for easy measurement. Measure the variation of each band from a perfect 100. The variations will be positive percentages. Add up all the percentages and divide by the number of bands.
A small variation, perhaps 3 percent either way across only half the range, will produce a variation of only 1.5 percent overall. A large variation, 10 percent across several bands, will still produce only a small variation because it is divided down by by all the bands. If we set a target of 100 for 100 percent accuracy in every band and 0 percent as a 100 percent failure in every band, the difference between a good light and a bad light will be trivial. What we need is an additional factor to amplify the bad.
The worst of the current lights have a variation of 50 percent across a significant number of bands. I suggest variations be quadrupled so that a variation of 50 percent across a significant number of bands is considered a complete failure. A variation of 50 percent across 20 percent of the bands would normally produce a total variation of 10 or a CIA of 90 percent. Quadrupling the result would produce a variation of 40 or a CIA of 60%, a more obvious failure.
People might not mind a CIA of 90 when the light is slightly cheaper. Converting the CIA to 60 out of 100 would show people the accuracy is big enough to produce a significant visual defect.
Colour Index Safety would be a measurement of holes in the visual spectrum that are dangerous for people with colour vision impairment. People with partial loss of sensitivity to a specific colour should not have to work in an environment where the lights are also weak in that colour because the combination can lead to those people missing safety indicators and text highlighted in colour.
Colour Index Ultra would be a measurement of dangerous ultraviolet light. You do not want ultra violet light without knowing the strength. Fair skinned people need to protect themselves against excess exposure. Some poorly designed fluorescent lights produce enough fluorescent light to damage fair skin. There should be a test and a warning, in this case the CIU.
Ultraviolet lights are used for growing plants and some aquariums. In dark parts of our planet, ultraviolet lights are used to supplement sunlight during the winter, especially for people who work indoors all day. Ultraviolet lights are useful and we need a standard labelling to give people an indication of their exposure.
The ideal indication would produce a figure similar to the UV exposure warnings published for sunlight over summer. You could then use the same decision making process for wearing protective clothing, skin cream, and sun glasses.
You cannot easily prove the number of hours a light is used. Hours based guarantees are useless. I suggest a date of manufacture then a number of years from then. There are 8760 hours in a year. I suggest 2 years is a reasonable guarantee period given the easy of making a light last long past the 17520 hours in 2 years. The date of manufacture should be stamped on the light in the factory. Replacement at your local shop should be automatic without having to produce a receipt. So long as they sell that brand, they should replace any light of that brand.
A compulsory minimum lifetime would force manufactures to spend that extra 2 cents at the factory.
The design of modern lights gives manufacturers the chance to produce a better product. Unfortunately in Australia, and many other countries, there are no labelling requirements or minimum guarantees to force manufacturers to compete on quality, efficiency, long life, or colour accuracy. You have to do all the research.