Fullscreen HTML5Med Res (??MB)HTML4

Slash your FA OFFICE LIGHTIN Most offices, factories and shops waste a lot of money in power for lighting. How would you like to save up to 50% of your lighting power bill . . . and get even more light into the bargain? We show you how – and have the measurements to back it up! I n a typical commercial building, lighting accounts for about 30% of total energy consumption. Air-conditioning amounts to about 50%, while the remainder is taken up by office equipment such as computers, copiers, printers and so on. So how can we reduce overall power consumption? Over the next few years this question will become far more pressing as electricity tariffs increase dramatically. Since air-conditioning is the main energy user, it behoves the building or office manager to ensure that everything has been done to minimise energy use. In particular, attention must be paid to anything which generates heat in summer, which means the air conditioner has to work even harder – and use even more energy – to overcome. Regular servicing of the air-conditioning system (especially cleaning the filters, which you can usually do yourself), monitoring of temperature settings in winter and summer, ensuring that doors are kept closed to stop drafts and so on are all important. Ultimately, measures like window tinting and double-glazing can provide further energy savings but the initial investment will be a lot higher. But cutting lighting energy use is the focus of this article. This came about for the very practical reason that all the fluoro tubes in the SILICON CHIP offices obviously needed replacing. In most offices, the approach would be to have all the tubes and starters replaced, together with cleaning the diffusers. That should be done every few years as a matter of course but this simple approach will not provide any energy savings. We were looking for significant savings. Our first step was to measure the light levels around the office and it must be stated that they ranged from just adequate to poor. In an office of about 12 x 8 metres, the levels ranged from under 150 lux to about 280 lux, at best. The average level was about 210 lux. The whole area is lit by 12 twin-36W fittings, more correctly referred to as recessed luminaires (or in the trade as “troffers”) which have prismatic diffusers. In our case, as we moved into this building in 2004, the tubes were probably at least six years, or around 16000 hours old – well overdue for replacement. Not only were the tubes noticeably down in emission but the prismatic diffusers were quite dirty. The second step was to replace the old tubes in two of the luminaires in my office area with new GE Cool White tubes which have a colour temperature of 5000°K. These tubes Inside a traditional (iron ballast) fluorescent fitting, as found in hundreds of thousands of offices, factories and shops around the country. The two ballasts are in the middle, the power factor correction capacitor is on the right, while the starters are mounted on the ends. Note the blackening of the tube ends – a sure sign these tubes are on the way out. 12  Silicon Chip siliconchip.com.au ACTORY/ NG bill! are significantly whiter than the old tubes which have the distinct greenish hue (or “cast”) of conventional fluorescent tubes. With the new tubes fitted, the light measurement went from 210 lux to 320 lux. This was much brighter but then we decided to try some Mirabella Tri-Phosphor tubes, again with a colour temperature of 5000°K. These were much brighter again; too bright in fact. So we opted to have just one Mirabella Tri-phosphor in each fitting. This gave a light reading of 270 lux, well above what we had started with. But we also wanted to try NEC quad-phosphor tubes which are claimed to be 15% brighter than tri-phosphor tubes. While slowly becoming more popular, they’re not the easiest things in the world to buy (as yet, they’re not in your local supermarket), nor are they cheap. But we found them in a Bunnings Hardware store and purchased a couple to try out. Incidentally, we also found out while shopping around that it is getting almost impossible to buy older, singlephosphor tubes any more. The vast majority of tubes on the shelves of both supermarkets and hardware stores were in fact tri-phosphor. So at least that’s a good start! We’re getting a bit ahead of ourselves here but we were so happy with our tests we bit the bullet and purchased a box of 25 NEC quad phosphor 37W tubes from our local electrical wholesalers, John R Turk, here in Brookvale. The cost was $316.25 including GST, or $12.65 per tube. This might seem expensive for 25 fluoro tubes but it is much cheaper than buying them retail. With a single NEC quad-phosphor tube in each luminaire, the light was up dramatically to around 310 lux. This was great so we then did the same for six twin-36W By LEO SIMPSON luminaires – fitting one quad-phosphor tubes for the old tubes and cleaning all the prismatic diffusers and whitepainted surfaces of the light fitting. This brought about a dramatic change. The final light reading on my desk was now 330 lux. By contrast, here’s a modern fluorescent luminaire fitted with a single electronic ballast (centre-right). Note the absence of a power factor correction capacitor and starters – they’re not needed with the electronic ballast. The downside of this particular fitting is that it cannot drive a single tube – you must have two fitted. But they should last longer. siliconchip.com.au May 2010  13 Fig.1: The operation of a conventional ballasted fluorescent light. The yellow trace is the incoming 230VAC waveform while the blue trace shows the waveform across the fluorescent tube. The pink waveform is the tube current while the purple waveform is the product of the tube voltage and current (power is 36W). Fig.2: These scope waveforms are taken from a twin36W electronic ballast luminaire. The green trace is the incoming 230VAC waveform while the yellow trace shows the overall current. The purple waveform is the product of the voltage and current (power is 77W). That’s towards the low end of the relevant Australian standard for office lighting (see separate panel: “What is the correct office light level”) but I found that it was more than enough for normal work. In fact, I found anything much greater than about 300 lux started to become a problem, especially when trying to read glossy or even semi-gloss (ie, coated) magazine pages. 250VAC power factor correction capacitor in the luminaires. This has been selected to correct the power factor of two ballasts in the luminaire and it over-compensates when just one ballast (ie, one tube) is in use. So in practical terms, we were able to reduce the power consumption of a twin 36W luminaire from around 90W to about 48W. The extra wattage compared with the power rating of the fluorescent tubes themselves (37W) is the amount of power wasted in the ferromagnetic ballasts. Pull out a tube! All of the tests so far had involved twin-36W luminaires using conventional ferromagnetic ballasts and starters. And by removing one lamp from each fitting, we obtained a reduction in power of just under 50%. Why not exactly 50%? The discrepancy is due to the 8F BALLAST ACTIVE 230V AC NEUTRAL PFC TUBE FILAMENT Traditional fluorescent light connection STARTER STARTUP ACTIVE 230V AC FILAMENT ~350V DC NEUTRAL Electronic ballast fluorescent light connection OSCILLATOR AC CAPACITOR INDUCTOR TUBE RESONANT CAPACITOR Compare the traditional fluoro lighting circuit to one with an electronic ballast. While the electronic ballast looks much more complicated, there is no starter nor power factor correction capacitor. Most electronic ballasts power two 36W tubes (not one as shown here for simplicity). The tube filaments in an electronic ballast circuit are essentially ignored – there is certainly no need to heat them. 14  Silicon Chip Small office tests In order to do a more controlled test, we decided to repeat the procedure in a separate office measuring 3.2 metres square (10.24 square metres), occupied by one Ross Tester. It was lit by two twin-36W luminaires, although one tube in one fitting was missing. The initial light measurement was 270 lux and that rose to 340 lux, with four tubes fitted. Removing the rather dusty prismatic diffusers increased the reading to 350 lux. Cutting to the chase, we removed the tubes from both fittings and fitted two NEC 37W quad phosphor 5000°K tubes to the luminaire immediately above the centrally placed desk. Now, with the lux meter sitting on the desk (as it was for the previous measurements), the reading jumped to 380 lux and then increased again to 420 lux with the prismatic diffuser cleaned and replaced. (Obviously the diffuser concentrates the light downward, hence the increase in brightness with it fitted). Pretty good eh? So, we have reduced lighting energy use in that small office from around 180W (with four tubes fitted) to 90 watts. But wait, there’s more! Electronic ballasts While simply replacing tubes with the new quad phosphor units makes a lot of sense, it is not without problems. First, starters, starter sockets and tombstones (the sockets in which the tubes sit) can become quite brittle with age and siliconchip.com.au Fig.3: by contrast, these waveforms are taken from a twin36W conventional ballast luminaire. Notice that the purple power waveform is 87.85W, substantially higher than for the electronic ballast fitting. Fig.4: the high frequency operation of a single tube in a twin-36W electronic ballasted fitting. It is being driven at 32kHz and gives about 9 or 10% more light output. the mere act of changing tubes or starters can fracture them. Then what do you do? If you’re doing the changeover yourself in a small office, you might consider replacing dud sockets but it is a time-consuming and dirty job. It would certainly not be practical to have the job done by an electrician, as the labour costs would be high. In any case, suitable tombstone sockets may not be readily available (most fittings these days have snap-in sockets made to fit a specific punch-out). In this case, you would simply replace the entire luminaire with one using an electronic ballast. These are now readily available at lighting wholesalers and are actually cheaper than the identical fittings with conventional ballasts. The catch is that typical twin-36W luminaires usually have one electronic ballast driving two tubes; you cannot operate them with a single tube. And while it is possible to purchase luminaires with a ballast driving a single tube, they will not necessarily fit into the same space occupied by the twin-36W fitting. But depending on the spacing of the twin-36W luminaires, it may be possible to fit electronic ballast versions and leave some fittings without any tubes. And that is what we did in the small office just discussed. We substituted a twin 36W luminaire with electronic ballast for the two conventional luminaires. This leads to two further benefits. First, electronic ballast luminaires are far more efficient than those with conventional ballasts. Compared with the 90 or so watts drawn by a twin-36W conventional ballast version, the electronic version only draws 77 watts, a power saving of 15%. Better still, the light output can be expected to increase by about 9%. Now why is that? It’s not magic. In a conventional ballasted fluorescent light fitting, the light output from the tube varies more or less sinusoidally at 100Hz, ie, double the 50Hz mains supply siliconchip.com.au These two graphs show the spectra of warm white (3000°K) versus “natural” 5000°K tubes. Notice that there is far more output at the “blue” end of the spectrum for the 5000°K tubes. (Courtesy Nelson Lamps Australia, distributors of NEC fluorescent tubes.) May 2010  15 Fig.5: this is the same test set-up as in Fig.4 but the scope is set to display the power waveform (purple trace) and shows a result of 34W. Depending on how this measurement is taken, it can vary from around 40W to less than 25W but the averaged long-term value is around 34W. Fig.6: this waveform shows the light output from a ballasted fluorescent tube, measured with a phototransistor. Notice that the light output is modulated at 100Hz, ie, with peaks corresponding to twice the 50Hz mains supply frequency. In effect, the lamp is extinguished 100 times a second but the persistence of the tube phosphor smooths this out so that flicker is normally not noticeable. By contrast, electronic ballasts run the tubes at much higher frequencies. In the case of the units we purchased, the drive frequency is around 32kHz. In effect, there is less variation in the UV radiation from the mercury vapour in the tube and the phosphors provide further smoothing. We have included some scope grabs of the typical light variation from a conventional ballast fitting and one fitted with electronic ballast. Interestingly, the light from the electronic luminaire still exhibits 100Hz modulation, overlaid with a much smaller modulation at 32kHz. However, the 100Hz modulation is about half that from the conventional ballast fitting. So what happened is that when we changed to an electronic ballast luminaire in Ross Tester’s office, the light reading increased to around 460 lux. By this stage Ross was asking whether he should be issued with sunglasses and skin cream for protection from UV exposure. It certainly is quite bright . . . but he got short shrift! Electronic ballasts have other side benefits as well. First, there is no apparent light flicker. Second, there is no audible hum or buzz which can be a problem with conventional ballasts. Third, there is none of that flick, flick, flickity flick when the lights are first turned on. Finally, because no starters are involved, they don’t need to be replaced when they fail (another saving!) and the ends of the tubes do not blacken so much as they age. However, there are two minor drawbacks with using electronic ballasts. The first is that if the ambient temperature is less than 5°C, the tubes may not start reliably. This is a problem with all fluorescent lights but apparently it is more pronounced in those fittings which have electronic ballasts. We have not tested this aspect – not only was it a balmy autumn (albeit damp) when we did our tests but the last time it regularly got to less than 5° here on the northern beaches of Sydney, dinosaurs were dropping dead. Secondly, the twin-36W electronic ballast luminaire we tested produced significantly more interference to AM radio reception than a conventional ballasted fitting. We have included some scope grabs of the interference waveform from each type. Even in this un-retouched photo of a twin fluoro fitting (which really doesn’t do it justice!) you can readily see the light level difference between a 4200°K tri-phosphor (the tube on the bottom) and 5000°K quad-phosphor (on top). While the “warmer” colour temperature of the tri-phosphor accounts for some of this difference, the quad phosphor is much brighter. 16  Silicon Chip siliconchip.com.au Fig.7: this shows the light output from a fluorescent tube driven by an electronic ballast. Notice that the light output is actually higher but still modulated at 100Hz and by about 50% less. It also shows very slight modulation at 32kHz. This could be a problem if you live in an area where AM reception is weak. But you’d probably already know this from interference from all your switchmode supplies! Colour temperature All our tests involved fluorescent tubes with a 5000°K colour temperature. They are noticeably brighter than Cool White 4200°K or other colour ratings. Their colour rendering is also quite reasonable with a Colour Rendering Index (CRI) of 84. In practice, we found the NEC quad-phosphor Natural 5000°K tubes to be far superior to all other tubes, especially those labelled Warm White (3000°K). Total power savings In all, we replaced the two old tubes in each of 17 twin-36W luminaires with 17 NEC quad phosphor tubes. The total power saving (including the changeover to one electronic ballast twin-36W fitting) is around 730 watts. Considering that these lights are typically on for 10 hours a day or around 2500 hours per annum, that means a saving of 1825kWh per annum. At the current tariff of 22c/kWh, that is a saving of over $400. Neglecting labour cost for the exercise (because we would have had to replace many tubes anyway), that means the payback is less than one year. And the whole office is considerably brighter into the bargain. By the way, we have been told by a distributor that the wholesale cost of NEC quad-phosphor fluorescent tubes is now actually less than their equivalent tri-phosphor tubes. So retail prices of quad-phosphor tubes should be coming down quite soon. T8 versus T5 tubes In none of this discussion have we mentioned T5 fluorescent tubes. Hmm, what’s this about T8 and T5 tubes? Simply put, this nomenclature refers to the diameter of the tubes in eighths of an inch. siliconchip.com.au Fig.8: this is the same waveform as in Fig.7 but with the scope settings changed to highlight the slight 32kHz modulation in the light output. What is the correct office light level? Insufficient light level in the workplace can cause eye strain, headaches and possibly induces migraine. And accidents occur more often when workers have difficulty seeing at less than optimum levels. By the same token, excess light levels are also not good for the health and safety of workers. Again, headaches, fatigue, stress and possibly migraines have been blamed on excessive light levels in the office. Try reading a newspaper in direct sunlight, for example and you will agree it’s not comfortable! Even if there were no ill-effects from excess light levels in the office, they of course waste a lot of energy. And that’s becoming even more important as electricity costs keep rising. As you might expect, there is an Australian standard covering the amount of light required in an office. AS1680.2.2 suggests that for ordinary office tasks, the lighting level should be in the range of 300-400 lux at desk (task) level and 160 lux as a background. The simple action of moving a desk (to accommodate other furniture or fittings, for example) may mean that what was acceptable light level is no longer enough. Or vice versa of course. Some applications will require stronger lighting – intensive manual graphic arts (ie, not on computer screen) such as drafting, illustrating, etc, may require illumination of up to 750 lux. Tasks involving constant reading, especially from small type, working with material that is not sharp (eg, poor photocopies) and similar has a suggested minimum of 600 lux. Older workers, too, usually require stronger lighting (ie more illumination) than younger workers. And speaking of computer monitors, every graphic artist will attest that too much illumination (whether from lighting or natural sources) on a monitor can drastically alter results, especially where colour rendition and lightness/darkness in image manipulation is concerned. May 2010  17 Making the measurements Quite a few different measurements were made in preparing this article. First, we measured the light output in the offices with a Digitech light meter available from Jaycar (Cat QM-1587). This will measure in lux or foot-candles. In parallel with that, we used Digitech (Jaycar QM-1580) digital multimeter which includes ranges for measuring lux. The various oscilloscope measurements posed a number of problems. First, most oscilloscopes can only handle limited input voltages and we wanted to measure the 230VAC mains voltage waveforms. These will overload any normal scope with a maximum input range of 5V/div when used with a 10:1 divider probe. Our solution is to use our 100:1 divider probe which is a PMK PHV621, made in Germany. When plugged into an oscilloscope with probe sensing, the maximum input range becomes 500V/div. For current and most of the scope measurements in this article we used a setting of 100V/div. It was used to monitor the incoming 230VAC 50Hz sine waveform. We also wanted to monitor the voltage across the fluorescent tubes themselves in both the conventional ballast and electronic ballast luminaires. This presents several problems. First, we need to measure the current in the Active-Neutral circuit and this really needs to be done with an active current probe which can be isolated from the mains circuit. To that end, we used an Agilent 1147A current probe which has a bandwidth from DC to 50MHz and a continuous current rating of 15A (50A peak). Conventional current probes for oscilloscopes simply do not have sufficient bandwidth to measure the fluorescent tube operating frequencies which can range well above 35kHz in luminaires which have electronic ballasts. The Agilent 1147A current probe is a hybrid unit combining a Hall Effect sensor for DC measurements and a current transformer for AC measurements. Its output is 0.1V/A and if connected to an oscilloscope such as the Agilent 5000/6000/7000 series, it will be automatically sensed and the trace will show amps/div rather than volts/div. A further complication arose because we wanted to measure and display the voltage waveform directly across the fluorescent tubes. This is difficult enough in a conventional ballasted fitting but is more complicated in fittings with electronic ballasts which operate two tubes from the one ballast. The solution is to use an active differential probe and in this case we used a Pintek DP-25. This can handle a maximum voltage of 1000V RMS on its differential inputs while the maximum voltage between each input and ground is 600V RMS, ie, more than adequate to handle the voltages involved when making scope measurements on the 230VAC mains supply. It has three ranges: x 20, x50 and x200. For further information on the three items described above, the high voltage probe, current probe and active differential probe can be obtained from Trio-Smartcal, 3 Byfield Street, North Ryde, NSW 2113. Phone 1300 134 091. www.triosmartcal.com.au Other measurements We also measured power consumption of the various luminaire fittings and this was done with our own Appliance Energy Meter which was featured as a constructional project in the July & August 2004 issues. In addition, we compared the level and modulation of the light output of the conventional ballast and electronic ballast fittings. This was done using a standard phototransistor with a 10kΩ collector load resistor. These scope waveforms (Figs. 6, 7 & 8) are measured with the same reference level. We confirmed that not only is the light output from the electronic ballast fitting higher than the conventional ballast fitting but that the 100Hz modulation was about half. Finally, as a crude measurement of electromagnetic interference from the two types of luminaire, we used a standard portable AM radio while the interference waveforms were taken from a small coil of wire in close proximity to the respective fluorescent tubes. 18  Silicon Chip siliconchip.com.au Fig.9: the 50Hz interference signal radiated from a fluorescent tube with conventional ballast. It will be heard as a characteristic loud buzz in an AM radio. Hence, a T8 tube is nominally eight eighths or one inch in diameter and a T5 tube is 5/8-inch. The newer T5 tubes are claimed to be more efficient than T8 tubes and can only be run with electronic ballasts. However, if T8 tubes are similarly run with electronic ballasts, there is no difference in efficiency in terms of lumens/watt. In any case, T5 luminaires and T5 tubes are currently a great deal more expensive than T8s. There is no point in changing over. And why not LED replacements? Some readers may wonder why we have not considered LED replacements for fluorescent tubes. After all, they are available overseas, even in a “drop-in” package; that is, the same size and shape as a conventional fluoro tube and capable of being driven in the same fittings. The simple answer is that while the very best of them can only just match the efficiency of NEC quad-phosphor tubes (around 100lm/W), they are extremely expensive. In the next few years that is bound to change. The next step? The next step in the power saving saga is to eliminate those wasteful halogen downlights in our office. We will bring you more in due course. SC Here’s what to look for on the fluorescent tubes themselves. Top is the 3000°K NEC tri-phosphor, while below is the brighter 5000°K NEC quadphosphor. Both are rated at 37W and both are “T8”, or oneinch diameter, tubes. Note the absence of the “HG” (mercury) marking on the quad phosphor. siliconchip.com.au Fig.10: the 50Hz interference signal radiated from a fluorescent tube with electronic ballast. It will produce a lot more interference to AM radio reception. Simple steps to start saving (1) At the very least, remove and wash the diffuser and while it is out, wipe over both the fluorescent tubes and the inside of the fitting with a damp cloth. This won’t save you any power but you won’t be wasting any of the light output from what you’ve got. (2) Better still, do No.1 but at the same time, replace the old tubes with tri-phosphor tubes. Again, you’ll get even more light output for the same power. (3) Much better again, replace the old tubes with quad-phosphor tubes. You may well find (as we did) that you only need one quad-phosphor in each fitting. That’s an immediate power saving of 50% or more AND more light output than the pair of old tubes. (4) Best, replace the whole light fitting (usually called a “troffer” in the trade) with one fitted with an electronic ballast AND a quad-phosphor tube. You can normally do this quite legally yourself because these days, the vast majority of commercial/ industrial fluorescent lighting fixtures are fitted with a standard 3-pin power plug which mates with a 3-pin socket on the lighting circuit inside the false ceiling.     You might well ask “why not keep the existing fitting but simply replace the ballast with an electronic type?” We asked the same question of our wholesalers and found that the electronic ballast cost almost as much as a complete fitting (within a couple of dollars!) . . . and then you have to pay an electrician to replace it because that’s something you cannot legally do yourself! So it is economically unviable. May 2010  19