The working principle with magnetic ballasts

Fig. 2.1: Conventional glow starters
Fig. 2.1: Conventional glow starters

When commonplace line voltage, 230 V 50 Hz or something similar, is applied to a fluorescent light tube, normally nothing will happen. The withstand voltage of the gas inside, usually low pressure mercury vapour, 1.3 mg in a 58 W tube, is higher. When the filaments are being heated, they start to emit additional electrons, but this still does not suffice to reduce the breakdown voltage below the regular periodic peaks of the mains alternating voltage.

 

 

Fig. 2.2: Electronic starters are available for all possible situations of application
Fig. 2.2: Electronic starters are available for all possible situations of application

With a cold 8 W T5 type lamp (room temperature) self-ignition without any sort of firing was observed at 480 V TRMS (≈680 V peak). This value could be reduced to 380 V by pre-heating the filaments with a separate transformer. A 58 W tube was found to start off from the cold state at 1300 V sine wave, dropping to 550 V with pre-heated filaments.

A further reduction occurs when the voltage is applied abruptly, from 0 to full, instead of slowly increasing it by means of a variable transformer, but still self-start at 230 V 50 Hz does not occ

Important for getting started: The correct starter

Therefore a starter is connected in parallel with the lamp, usually the commonplace glow starter (Fig. 2.1), with any luck an electronic starter (Fig. 2.2). The basic wiring is given in Fig. 2.3. When applying the mains voltage a glow discharge is initiated inside the glow starter (Fig. 2.4) which heats up the bimetallic contacts and causes them to close (Fig. 2.5).

Now current flows from the mains via the ballast, the cathode filament, the starter and the second filament. This way the cathodes are pre-heated. But since the glow discharge has solely been shorted by the bimetallic contact, the bimetallic contact cools down and opens again few seconds after closing. By interrupting the current through the (relatively great) inductance of the ballast a substantial voltage surge is generated across the ends of the fluorescent lamp, starting a current flow through the tube (Fig. 2.6).

Fig. 2.3: Wiring diagram of a fluorescent lamp with magnetic ballast and glow starter
Fig. 2.3: Wiring diagram of a fluorescent lamp with magnetic ballast and glow starter
Fig. 2.5: …the contact shorts out the glow discharge, while a current limited by the ballast is flowing through the filaments&hellip
Fig. 2.5: …the contact shorts out the glow discharge, while a current limited by the ballast is flowing through the filaments&hellip

To top

Fig. 2.4: A glow discharge heats up the bimetallic contacts&hellip
Fig. 2.4: A glow discharge heats up the bimetallic contacts&hellip
Fig. 2.6: …the contact cools down again, which causes it to open, and a self-induction impulse fires the lamp – hopefully!
Fig. 2.6: …the contact cools down again, which causes it to open, and a self-induction impulse fires the lamp – hopefully!

The starting process with the usual suboptimal glow starter

Fig. 2.7: Glow discharge heats up bimetallic contacts, contacts short out glow discharge, cool down again and open
Fig. 2.7: Glow discharge heats up bimetallic contacts, contacts short out glow discharge, cool down again and open

At least this is what you would hope for. In fact, the luminaire is fed with AC, and whether the instantaneous current value at the instance of ignition, that is, of contact opening, is high enough right at that moment to generate a sufficiently high voltage impulse is an open question. But 'not now' does not mean 'never ever', since now, if the strike is not successful, the full voltage comes to be applied across the starter's terminals, glow discharge starts again, and a few seconds later the next firing attempt follows and so on until some very fine second the instant of firing coincides with a sufficient instantaneous current amplitude. Only then a small current flow through the lamp is initiated, which immediately generates more charge carriers so that the avalanche effect of conductivity increase according to Fig. 1.1 of the gas inside the tube is started.

The ballast's inductive resistance now prevents that on account of this conductivity increase also the current increases with avalanche effect right up to the big bang. The voltage across the starter, which at any instance is identical with the voltage drop across the lamp, is now so small that no new glow discharge is initiated in it. At least preliminarily this is so. As the lamp ages, the operating voltage drop across it gradually increases until at some moment it is so high that glow discharge inside the starter does start again (reclosing voltage): The starter is triggered even though the lamp is still in operation and shorts it out. Thereby the lamp is turned off – and of course it is ignited right again. There you have your flashing thunderstorm.

Fig. 2.8: …and when this game has been going on for long enough, this can be clearly seen when opening the starter (left; on the right an unused sample with filtering capacitor)
Fig. 2.8: …and when this game has been going on for long enough, this can be clearly seen when opening the starter (left; on the right an unused sample with filtering capacitor)

So statistically, this primitive, incredible technique called glow starter replaces any one start of a given lamp with several starting attempts, while especially the number of ignitions is reported to be a crucial lamp ageing factor.

During pre-heating, the current exceeds the rated lamp current by about 35%, since it flows only through the reactor (Fig. 3.9, bottom right) – and also through both of the filaments so as to pre-heat them. Their voltage drop, however, is low, only some 10 V, while the great voltage drop across the lamp is shorted out.

With an old poor quality ballast that obviously operated way too close to the range of magnetic saturation, if not right within, the current during cathode heat-up rises clearly more than mentioned 35% above the rated 0.67 A, namely as high as 1.15 A. The heating power of each filament reaches 13.5 W, which makes the filaments shine in a bright white even without any voltage between the two of them applied. This provides more likelihood to get started because the instantaneous current amplitude at the instance of contact opening is more likely to exceed the necessary minimum for ignition, which then also lies lower because of the plenty of free electrons emitt

Fig. 2.9: In the end of a day (or a week or a month) the contacts weld together, and the lamp persists in permanent pre-heating operation, and even with an optimised low-loss magnetic ballast a 58 W lamp still consumes a useless 33 W; with an old inefficient magnetic ballast dimensioned for 220 V and now being operated at 230 V the useless power even rises to 65 W!
Fig. 2.9: In the end of a day (or a week or a month) the contacts weld together, and the lamp persists in permanent pre-heating operation, and even with an optimised low-loss magnetic ballast a 58 W lamp still consumes a useless 33 W; with an old inefficient magnetic ballast dimensioned for 220 V and now being operated at 230 V the useless power even rises to 65 W!

To top

The starting process with the optimal electronic starter

Unfortunately a substantial excess in pre-heating current also adds to the ageing impact of start-ups if current is really excessive, while pre-heating is basically essential to reduce the wear effect of starting procedures. The much better choice is a combination with an energy efficient ballast, which by design operates still more or less within the linear range of the core material even during pre-heat, and an electronic starter, since electronic starters:

  • Start after optimum pre-heat time for maximum lamp life.
  • Start at a defined point of the phase (current peak), so each firing is successful, no flickering.
  • No replacement of starters, unlike recommended to do with conventional glow starters along with each lamp replacement.
  • No residual current through the filtering capacitor as contained in a conventional glow starter.
Fig. 2.10: Starting voltage pulse (bottom left), inrush and warm-up currents on a 58 W lamp, rows 1 and 2 with low quality magnetic ballast, rows 3 and 4 with energy efficient magnetic ballast, rows 1 and 3 without and rows 2 and
Fig. 2.10: Starting voltage pulse (bottom left), inrush and warm-up currents on a 58 W lamp, rows 1 and 2 with low quality magnetic ballast, rows 3 and 4 with energy efficient magnetic ballast, rows 1 and 3 without and rows 2 and

Also, improved glow starters already provide a 20% lifetime expectancy increase, but of course the glow technology cannot offer any of the other advantages of electronic starters. All the more amazing it does appear, though, that this polished-up version is being offered by an international lamp and electronics manufacturer, unlike electronic starters, as one should have expected. Howsoever, it makes the lamp lifetime advantage of electronic ballasts dwindle to some 10% or 20%, anyway.

It would lead too far to delve into the electronic details of such starters at this point. The working principle, after all, is the same as with conventional ones: A normally closed contact that opens a certain time lag after powering. Fig. 3.9 shows the current during firing and during warm-up. In rows 2 and 4 a capacitor was connected in series with the lamp and ballast that substantially reduces the warm-up current, which is often argued to be a disadvantage. However, if this is a disadvantage then the circuit, especially the ballast, is poorly designed, which is not attributable to the basically brilliant method of serial compensation (in a so-called lead-lag connection, see section 4). Indeed the mentioned high pre-heat current with a poor quality ballast drops as low as 0.676 A when serial compensation is applied, which is only more 60% of the prior value. Yet, the same measurements when carried out with a high efficiency ballast provide readings like 0.994 A in the lagging circuit (without compensation) and 0.698 A in the leading circuit (with serial compensation), so the ratio between the two is still 70%.

As a comparison the starting procedure was recorded with the transients recording function of a power analyzer, more precisely speaking the voltage across the ends of the lighting tube was recorded. So the voltage input terminals of the meter were connected to the poles of the starter, which are permanently connected to one of the filaments each.

It needs mentioning at this point that the meter was a usual power quality analyzer with a transients capture function among others, not a dedicated transients recorder. This requires that the device always needs to wait for at least a few mains frequency periods to recognize the amplitude, frequency and waveform of the regular line voltage. Only then can it decide what deviates far enough from this to be called a transient. Everything which – depending on setting – deviates by 20%, 50%, 100% or 200% from its expected instantaneous value is supposed to be a transient and is recorded as such. Unfortunately the threshold value can only be fixed to these per cent values but not to absolute voltage amplitudes. At the first instance of switching a fluorescent lamp on, however, the starter represents a closed switch, shorting the input terminals of the meter, which caused the automatic range selection to step down to the lowest range of 4 V. This did not enable the recording of the transient. Apart from this, the recorder continuously recorded transients, about 2 per second, even when set to 200%, since the voltage across the input terminals was practically 0, and 200% of nearly nothing is still nearly nothing. Any minor coupled disturbance was therefore recorded as an assumed transient.

Fig. 2.11: Wiring diagram of a fluorescent lamp with magnetic ballast and electronic starter
Fig. 2.11: Wiring diagram of a fluorescent lamp with magnetic ballast and electronic starter

To top

Fig. 2.12: Test circuit for recording the lamp firing transient
Fig. 2.12: Test circuit for recording the lamp firing transient

As a comparison the starting procedure was recorded with the transients recording function of a power analyzer, more precisely speaking the voltage across the ends of the lighting tube was recorded. So the voltage input terminals of the meter were connected to the poles of the starter, which are permanently connected to one of the filaments each.

Fig. 2.13: Starting a T8 lamp 58 W with conventional starter
Fig. 2.13: Starting a T8 lamp 58 W with conventional starter

Fig. 2.14, however, provides evidence of how clean such a starting process can be and always will be using an electronic starter. You can take this recording as often as you like, and it will always look alike: There is a high, narrow peak at a precisely defined point of time. For this reason this recording is displayed twice – these are two different views of the same event. In the right view the cursor line was merely moved to the right. This provides the advantage of making the very narrow peak visible at all, whenever it is difficult, since it is very high but extremely narrow, just as it is supposed to be. The left view, on the other hand, provides the opportunity to read (at the top of the screen) that the impulse ranges from -0.32 kV to 1.36 kV. That's enough – and quite sure causes less conducted and radiated disturbances than the multitude of blurred impulses of the glow starter.

Fig. 2.14: Starting a T8 lamp 58 W with electronic starter
Fig. 2.14: Starting a T8 lamp 58 W with electronic starter

To top

The starting process and the lamp lifetime expectancy

Fig. 2.15: Fluorescent lamps in a residential washroom with magnetic ballasts and electronic starters: Only one lamp replacement within well over 30 years
Fig. 2.15: Fluorescent lamps in a residential washroom with magnetic ballasts and electronic starters: Only one lamp replacement within well over 30 years

In the washroom of a single-family home new lamps were installed around 1970 (Fig. 2.15). The new luminaires were refurbished with electronic starters in their very early days. Since then, one of the lamps has been replaced once and the other one never ever. Neither has any of the starters, of course. You can see that the old 38 mm diameter T12 lamp is still in place and working (Fig. 2.16), while these were taken over by the 28 mm thick T8 lamps already in 1980. The washroom is at the same time the passage to another cellar room so that the lights are switched relatively often, at average about 5 times a day, while the average operation time is low, barely an hour a day. So this lamp has done around 10,000 service hours and 50,000 starts to date.

The lifetime of fluorescent lamps is usually identified by means of the switching cycle given in IEC 60081, according to which the systems are operated for 2:45 h and turned off for 15 min. It says there that the starters used in such a test have to comply with the IEC 60155. However, this standard refers to glow starters, and hence the lamp lifetime rating with magnetic ballasts refers to ignition by means of glow starters.

For the practical lamp life, or for the impact from the number of starts upon same, respectively, optimal pre-heating is crucial. The operation of one and the same lamp on an electronic warm-start ballast is sometimes rated with a 30%, sometimes with a 50% longer lifetime. Ratings for immediate-start electronic ballasts are generally not given, albeit these represent the generic implementation of electronic ballasts. With magnetic ballasts, however, the inverse procedure is always preferred and the – unfortunately very low – standard of the glow starter chosen to be the normative benchmark.

Regardless of this, some experts actually do regard the electronic ballast’s »warm start« as rather »lukewarm«. For an optimal lamp life, they say, the cathodes would need to be heated for at least 2 s, and nor was the pre-heat current normally what it should be, but such a long pre-heat time seems to be an intolerable intrusion to the contemporary user.

Obviously the same applies to the contemporary manufacturer, since according to Philips a triphosphor fluorescent lamp lasts for about 60,000 h without any switching on and off, while in the standard test, say 8 switch-on and switch-off cycles a day, it does 15,000 h. A halophosphate lamp reaches about half of these values, while special long-life brands may be able to more than double this life span – with all the appropriate caution in the context of such ratings, for it takes several years to collect several years of operating experience. The open question always is whether a recently accomplished experience is still topical by then, or whether the respective version of the respective type of product is already off the shelves again.

The further question is how long a fluorescent lamp really does last when optimally pre-heated. Unquestionably, however, the commonplace glow starter falls far below this optimum, since, as explained earlier, it replaces one starting process with several starting attempts and thereby multiplies the number of ignitions. Even the one out of several attempts finally leading to success may have run in an optimal manner only in a few exceptional cases. Rather, in the majority of cases, the instantaneous current amplitude at the instance of ignition may have sufficed only sparsely rather than with a rampant reserve for successful ignition, which again places some additional wear impact upon the filaments. The inevitable radio interference filtering capacitor inside the glow starter attenuates the voltage peak, rounds it, lops it and hence reduces its efficacy. So how high is then the improvement potential of starters »other than glow starters« according to IEC 60927, such as electronic starters, which, dependent on ambient temperature and other parameters, always fire after the optimal pre-heat time and always at the peak of the current, hence always successfully at first attempt? Commensurate data from industry is lacking here, too. Not even interested parties such as the manufacturers of such starters have it because these companies are small and such measurements are quite extensive and accordingly expensive.

But as the most common lamp types last roughly 4 times as long when operated in continuous mode as they do in the standardized test cycle, this facilitates the conclusion that with the switching frequency of above mentioned standard of 1 on / off period per every 3 h only 1/4 of the lamp wear is due to the »normal« generation of light but 3/4 account to the switching. Beyond, there are at least a few hints and experiences pointing towards the far superior lamp preservation through electronic starters.

For instance, on the demonstration panel by one of the manufacturers, which is regularly exhibited at trade fairs, the lamp operated with a glow starter (Fig. 2.17 bottom right) has to be replaced after every second fair – along with the starter. These lamps are operated according to the extremely fast test cycle according to IEC 60155 for testing the starters, i. e. 40 s on / 20 s off. Hence, the respective lamp has done about 6,000 ignitions when done. By then either one of the filaments is interrupted, or the starter’s contacts are welded together, so that the fuse of the »fused« starter trips, or the lamp only more flashes and flickers instead of burning continuously, which will also trip the fuse and set the lamp dark.

The lamp with electronic starter, however, has survived 25 exhibitions in the meantime, totaling up to no less than 120,000 starts (with only some 1,300 h of operation, though), still being fully functional with hardly blackened ends (Fig. 2.17 top left). Deviating from this, the lamp test according to IEC 60081 with a net operating time of 2:45 h per every gross 3 h of test duration, executed with glow starters and hence leading to a lifetime of 15,000 h, totals to an overall test duration of 16,364 h. So this duration will include 5,455 starts, obviously being responsible for 3/4 of the overall ageing effect. Let’s add another 1/4 on top, and the fluorescent lamp would theoretically be »done« after 7,264 glow starts without having burnt at all. Of course it always needs to have burnt for a moment in the respective time intervals between ignitions – at least long enough to verify that it did fire successfully. This complies quite precisely with the observation on the demo panel. Assuming it was not »about 6,000 starts« but precisely 7,264 starts within 81 h of operation, then this point, together with the two points »5,455 starts with 15,000 h« and »60,000 h with just one start« comes to lie exactly on the black line given in Fig. 2.18.

Fig. 2.16: Removal of the cover reveals: One of the two lamps is still of the old T12 type (38 mm diameter) and is still working well
Fig. 2.16: Removal of the cover reveals: One of the two lamps is still of the old T12 type (38 mm diameter) and is still working well

Regardless of this, some experts actually do regard the electronic ballast’s »warm start« as rather »lukewarm«. For an optimal lamp life, they say, the cathodes would need to be heated for at least 2 s, and nor was the pre-heat current normally what it should be, but such a long pre-heat time seems to be an intolerable intrusion to the contemporary user.

Obviously the same applies to the contemporary manufacturer, since according to Philips a triphosphor fluorescent lamp lasts for about 60,000 h without any switching on and off, while in the standard test, say 8 switch-on and switch-off cycles a day, it does 15,000 h. A halophosphate lamp reaches about half of these values, while special long-life brands may be able to more than double this life span – with all the appropriate caution in the context of such ratings, for it takes several years to collect several years of operating experience. The open question always is whether a recently accomplished experience is still topical by then, or whether the respective version of the respective type of product is already off the shelves again.

The further question is how long a fluorescent lamp really does last when optimally pre-heated. Unquestionably, however, the commonplace glow starter falls far below this optimum, since, as explained earlier, it replaces one starting process with several starting attempts and thereby multiplies the number of ignitions. Even the one out of several attempts finally leading to success may have run in an optimal manner only in a few exceptional cases. Rather, in the majority of cases, the instantaneous current amplitude at the instance of ignition may have sufficed only sparsely rather than with a rampant reserve for successful ignition, which again places some additional wear impact upon the filaments. The inevitable radio interference filtering capacitor inside the glow starter attenuates the voltage peak, rounds it, lops it and hence reduces its efficacy. So how high is then the improvement potential of starters »other than glow starters« according to IEC 60927, such as electronic starters, which, dependent on ambient temperature and other parameters, always fire after the optimal pre-heat time and always at the peak of the current, hence always successfully at first attempt? Commensurate data from industry is lacking here, too. Not even interested parties such as the manufacturers of such starters have it because these companies are small and such measurements are quite extensive and accordingly expensive.

But as the most common lamp types last roughly 4 times as long when operated in continuous mode as they do in the standardized test cycle, this facilitates the conclusion that with the switching frequency of above mentioned standard of 1 on / off period per every 3 h only 1/4 of the lamp wear is due to the »normal« generation of light but 3/4 account to the switching. Beyond, there are at least a few hints and experiences pointing towards the far superior lamp preservation through electronic starters.

For instance, on the demonstration panel by one of the manufacturers, which is regularly exhibited at trade fairs, the lamp operated with a glow starter (Fig. 2.17 bottom right) has to be replaced after every second fair – along with the starter. These lamps are operated according to the extremely fast test cycle according to IEC 60155 for testing the starters, i. e. 40 s on / 20 s off. Hence, the respective lamp has done about 6,000 ignitions when done. By then either one of the filaments is interrupted, or the starter’s contacts are welded together, so that the fuse of the »fused« starter trips, or the lamp only more flashes and flickers instead of burning continuously, which will also trip the fuse and set the lamp dark.

The lamp with electronic starter, however, has survived 25 exhibitions in the meantime, totaling up to no less than 120,000 starts (with only some 1,300 h of operation, though), still being fully functional with hardly blackened ends (Fig. 2.17 top left). Deviating from this, the lamp test according to IEC 60081 with a net operating time of 2:45 h per every gross 3 h of test duration, executed with glow starters and hence leading to a lifetime of 15,000 h, totals to an overall test duration of 16,364 h. So this duration will include 5,455 starts, obviously being responsible for 3/4 of the overall ageing effect. Let’s add another 1/4 on top, and the fluorescent lamp would theoretically be »done« after 7,264 glow starts without having burnt at all. Of course it always needs to have burnt for a moment in the respective time intervals between ignitions – at least long enough to verify that it did fire successfully. This complies quite precisely with the observation on the demo panel. Assuming it was not »about 6,000 starts« but precisely 7,264 starts within 81 h of operation, then this point, together with the two points »5,455 starts with 15,000 h« and »60,000 h with just one start« comes to lie exactly on the black line given in Fig. 2.18.

Fig. 2.17: Trade fair demo panel – at the bottom right the lamp as well as the starter have already been replaced some 12 to 14 times, the one top left not yet ever
Fig. 2.17: Trade fair demo panel – at the bottom right the lamp as well as the starter have already been replaced some 12 to 14 times, the one top left not yet ever

Hence, despite all caution applicable for lack of data, the available observations facilitate the conclusion that, in the standardized cycle including one switch-on and one switch-off per every 3 h, the major part of the ageing impact upon generic fluorescent lamps is due to the respective number of firings. Using the same lamp and the same magnetic ballast, however, the switching frequency obviously has as much as no effect at all upon the lamp life as soon as the glow starter is replaced with an electronic starter! This even remains valid if the »unscientific« ac hoc values used here should be by a factor of, say, 4 in error – including the exemplary prices named here.

An information brochure by the lamp and lighting section of ZVEI includes such a benchmark plotting of light flux curves of T8 fluorescent lamps operated on magnetic and electronic ballasts. Strangely enough both of the charts do not only resemble each other but appear to be absolutely identical right down to the last pixel. One tends to assume a typographical error, a swap of pictures, but why has nobody realized this yet since 2005? The other charts at least exhibit some divergence from each other. Care has been taken with these ones, however, to use a mean value out of a lead-lag compensation with 50% inductive and 50% capacitive lamps. This appears awkward, though, because at exactly this very time this very ZVEI issued the recommendation not to use the lead-lag compensation any longer (rather than to adapt the capacitance ratings for serial compensation to contemporary conditions – such as to include the shift of the voltage rating from 220 V to 230 V, the lamp power rating e. g. from 65 W to 58 W, as well as much narrower tolerance margins with lamps, starters and magnetic ballasts – see Section 5). Without this, however, the new lamps are substantially overloaded with the old capacitors.

Had separate charts been given for the inductive and the capacitive circuits of a lead-lag configuration, as would have been the correct approach, it would have turned out straight away why it was intended to abolish the lead-lag configuration. However, simultaneously it would have become obvious that the inductive circuit, optionally equipped with parallel compensation as recommended by ZVEI ever since, provides approximately the same lifetime expectancy as the warm-start electronic ballast.

Fig. 2.18: Lifetime expectancy of generic triphosphor fluorescent lamps plotted against the number of starts and the types of ballasts and starters used
Fig. 2.18: Lifetime expectancy of generic triphosphor fluorescent lamps plotted against the number of starts and the types of ballasts and starters used

In an old report by Philips from 1995, however, which on top of all was designated »for internal use only«, the systems still used to be recorded separately. It gives rather precisely 16,000 h of lamp life when operated on an electronic ballast – with a relatively low uncertainty margin of about ±10%. In the uncompensated magnetic ballasts group they achieved quite precisely 15,000 h. This is no more than a 7% advantage for the warm-start electronic ballast. In return, the group with the immediate-start electronic ballasts achieved only 13,000 h and hence even 15% less lifetime expectancy than could be identified with magnetic ballasts and the decrepit glow starter (Fig. 2.18)! The suboptimal magnetic ballast configuration with serial compensation and the oversized capacitance value achieved 12,000 h and thereby only insignificantly less. It is very likely that a better pre-heating than at present commonly used on the market would be possible to implement in contemporary electronic ballasts and would also be commercially reasonable to use with respect to life cycle costing for the whole installation. More information on when it is worthwhile to turn the light off or when to better leave it running will be given in section 7.1. An industry, however, supplying electronic ballasts as well as lamps may still calculate a positive balance selling one electronic ballast at 30 € per luminaire and in return lose 1/3 of its turnover with replacement lamps. On the other hand, replacing a throwaway starter at 30 cents with an electronic one coming at 3 € and with a similar lifetime expectancy as the magnetic ballast at 12 €, both lasting about as long as the whole building, and in return sacrificing the entire electronic ballasts market along with about 70% of the lamp replacements is quite obviously very economical for the user but not for the supplying industry.

To top

Operation

Fig. 2.19: Voltage and current of a 58 W fluorescent lamp in theory…
Fig. 2.19: Voltage and current of a 58 W fluorescent lamp in theory…

Under the assumption that the voltage be unaffected by any distorted currents from non-linear loads and therefore still sinusoidal, which you rarely ever find these days, the curves theoretically deducted from Fig. 1.1 and the underlying formula look like those in Fig. 2.17: Of course the extremely non-linear behaviour of the lighting tube distorts the voltage measured across the two filaments very much, for while the current is highest the voltage drop is lowest. Yet this voltage is unable to distort the current to a nameworthy extent because the distorting lamp load is connected in series with the very high inductance of the ballast, in this case »780 mH, which suppresses current distortion, respectively suppresses the flow of harmonic (higher frequency) currents. So the current curve looks nearly sinusoidal, apart from a crease at each zero crossing. Of course there is a long time lag between the voltage peak and the current peak, which means a high share of fundamental reactive power, but this is by far the minor problem. Real power quality problems arise when the current curves in a network become substantially distorted, say having high harmonic contents. Fig. 3.18 shows that these characteristic waveforms of the lamp current and the voltage across the lamp do not only occur in the theoretical model but also in practical measurements.

Fig. 2.20: …and in practice
Fig. 2.20: …and in practice
Fig. 2.21: An electronic ballast is not by default dimmable
Fig. 2.21: An electronic ballast is not by default dimmable

To top

Dimmability and its price

There have been various methods around to achieve the dimmability of fluorescent lamps with magnetic ballasts, ranging from phase angle control to an inverter feeding a whole lamp arrangement with a variable frequency. The problems were, especially in the former case, the increasing flicker when dimmed down low and to keep the lamp from extinguishing completely. Methods such as longitudinal electrodes paralleling the lamp and permanent filament heating were the more or less satisfactory solutions, the latter of which decreases the energy efficiency and makes dimmability doubtful if used in order to cut electricity costs. The variable frequency was not so much different from the use of electronic ballasts today, only the frequencies were lower and the magnetic ballast on top of electronic frequency inversion still needed. At 50 Hz, the full power was fed into the lamp, while as frequency rose, impedance of the ballast became higher, so the current dropped, while the voltage across the complete luminaire remained largely stable. Therefore and on account of the higher frequency the light flux was also more stable than with phase angle control, but after all the method was not so much cheaper than equipping each lamp with an individual dimmable electronic ballast. A new dimming technique for magnetic ballasts is presently being developed in Canada), which seems quite promising but is not yet commercially available on the market. So until today, if dimmability is required, the choice is still an electronic ballast, while the stand-by consumption of these (note section 6.1) remains an issue! Moreover, the dimming feature does not come together with an electronic ballast without mentioning, as is sometimes believed, but rather doubles the price once again (Table 2.1), which is already very high in comparison to even a high-quality magnetic ballast. The quoted prices per piece are valid for a quantity of one unit package, which is usually about 20 pieces, and possible rebates for larger lots range from 0% to a maximum of 50%. For OEM equipment being traded to the luminaire industry in tremendous piece numbers a substantially higher rebate may be possible. The unfortunate consequence of this is that the luminaires are then equipped with electronic ballasts by default, and customers who do not purchase huge quantities will be served with magnetic ballast luminaires not even on demand, however justified their desire may be (see section 6).

The prices of electronic ballasts are valid for those with a warm start as well as the so-called cut-off features, which is the only fair comparison. A cold start electronic ballast without cut-off technology comes at 47.50 €; from Vossloh-Schwabe. Both warm start capability and cut-off technology are appreciated as a valuable extra with electronic ballasts, while they come without mentioning, enforced by the principle, with magnetic ballasts, be it with electronic starters or with the poor conventional glow starters.

Table 2.1: Catalogue prices for a 230 V, 50 Hz, 58 W ballast
Table 2.1: Catalogue prices for a 230 V, 50 Hz, 58 W ballast

The prices of electronic ballasts are valid for those with a warm start as well as the so-called cut-off features, which is the only fair comparison. A cold start electronic ballast without cut-off technology comes at 47.50 €; from Vossloh-Schwabe. Both warm start capability and cut-off technology are appreciated as a valuable extra with electronic ballasts, while they come without mentioning, enforced by the principle, with magnetic ballasts, be it with electronic starters or with the poor conventional glow starters.