Possible disturbances with magnetic ballasts

As mentioned before, magnetic ballasts provide a mature and long-proven technique unlikely to cause any trouble or damages to other power consumers or the supplying voltage. One possible source of disturbance, which is more likely today to cause damage or malfunction to modern sensitive equipment than was the case in the past, is the voltage peak generated by self-induction in its high inductance at the instance of turn-off. Normally this will not be a problem, since lamps are hardly ever operated in parallel with such equipment on the same circuit with a common switch, but in one case it did happen. This rather uncommon damage could only occur on account of this exotic constellation (Fig. 4.1) but it should be mentioned that it may be a bit less exotic to parallel magnetic ballast fluorescent lamps with electronic halogen lamp transformers. Cases have been reported where the latter have repeatedly been destroyed by the turn-off self-induction surges of the former, and a special surge protector has been developed. Yet the problem could as well have been avoided by paralleling the two lamp-ballast-units with a capacitor. An appropriately dimensioned compensation capacitor will form a resonance frequency equal to the mains frequency, and the AC will therefore softly sway out after the supply voltage is turned off. With a smaller capacitor the resonance frequency is higher, and the turn-off voltage peak is »only« substantially attenuated, not entirely avoided, but the height of peak is not as crucial for the likelihood of disturbances as the rise time edge, which is attenuated very much through even a small capacitance.

Emission of disturbances from magnetic ballasts

Fig. 3.1: Unhealthy parallel connection of an electronic load and a highly inductive load on one common power switch
Fig. 3.1: Unhealthy parallel connection of an electronic load and a highly inductive load on one common power switch

In another case an old Commodore computer locked up every second time the 18 W fluorescent lamp in the bathroom of an old residential building was turned on. The home was TN-C wired, without a dedicated earth / protective conductor but only two cores in all single-phase supply lines and an interconnection between the neutral and protective earth connectors inside each single socket. This alone may have been the cause for the trouble or at least may have contributed to it, but anyway a capacitor connected in parallel with the ballast and lamp solved the problem.

It remains to be amended that there is no voltage surge when turning on the current in an inductance. The mentioned lock-ups in fact did not occur when pressing the light switch but when the starter tried to fire the lamp, which is basically a turn-off process of a reactor current, intentionally generating a voltage surge to get the lamp started (section 2.1).

Fig. 3.2: Symmetric ballast
Fig. 3.2: Symmetric ballast

In some cases, sensitive equipment may be disturbed by the high frequency emissions that the lamp as a gas discharge device emits, even if operated at mains frequency. In these cases it sometimes helps to just swap the polarity. Care should be taken that the ballast is always connected to the phase and the lamp to the neutral as an earthed conductor, not vice versa. This reduces the likelihood of described trouble. If it still occurs, a so-called symmetric ballast may help, the inductance of which is split in two halves, each of them to be connected to one end of the tube (Fig. 3.2). Beyond, only the usual commonplace filters will help, whenever these, if used excessively, may cause a leakage current problem. Inrush currents, however, are generally not a problem with magnetic ballasts. Their inrush currents are not that high. Further attenuation can be achieved when serial compensation is applied (bottom of Fig. 2.1), while parallel compensation (chapter 4) adds the inrush current of the capacitor, which has very steep rise time edges and may therefore very well become a problem.

When talking about the harmonic disturbances of electronic ballasts it is frequently alleged that magnetic ballasts also cause current harmonics, while this is not really so. The ballast itself is a linear element if designed properly, so as not to let the core material enter the range of magnetic saturation under normal operating conditions, which would be highly disadvantageous from a power quality as well as an energy efficiency viewpoint. Rather, the non-linear behaviour of the lamp itself causes an extreme magnitude of voltage distortion (Fig. 2.7, Fig. 2.8) but which on account of the high inductance of the ballast causes only little current distortion. So no nameworthy disturbance appears across the terminals of the luminaire. The harmonic load on the neutral conductor with lamps spread equally across the three phases is correspondingly low, in the case of 58 W lamps the simulation reveals about 35% of the phase current (Fig. 3.3).

Fig. 3.3: Operating 3 fluorescent 58 W lamps with magnetic ballasts on 3 phases, sum of the phase currents forming the neutral current
Fig. 3.3: Operating 3 fluorescent 58 W lamps with magnetic ballasts on 3 phases, sum of the phase currents forming the neutral current

Sometimes noise is mentioned as a type of disturbance from magnetic ballasts but this, if it occurs, is a case of faulty lamp design or fabrication. A faultless ballast alone does not produce any noise, but if it is fixed to a metal sheet surface in the luminaire this has to be done adequately: Tightly but including washers made of rubber or plastics. Otherwise mains frequency humming or buzzing may occur.

As another disturbance the inevitable permanent flicker at double the mains frequency is often mentioned. In some locations, where rotary machines are worked with, this can become dangerous on account of a stroboscopic effect that may make the rotating machinery appear to stand still or at least cause heavy optical misconception of its rotary speed or even the direction of its rotation. This, however, can easily be avoided by spreading lights equally across the three phases of the supply or by simply applying lead-lag connection (section 4).

Apart from this, it remains to be noted that with TV sets the 100 Hz technique is regarded as the latest flicker free development.

To top

Susceptibility of magnetic ballasts to disturbances

Fig. 3.4: Test configuration to provoke a voltage sag
Fig. 3.4: Test configuration to provoke a voltage sag

This is a short chapter. Fluorescent lamps operated with magnetic ballasts are almost entirely unsusceptible to commonplace network disturbances. The high inductance connected in series with the lamp suppresses surges, peaks and harmonics, i. e. if the likes of those are present in the line voltage they will be able to drive only a fraction of the proportional current through the lamp.

Fig. 3.5: Current peak caused by an asymmetric voltage dip
Fig. 3.5: Current peak caused by an asymmetric voltage dip

What may cause trouble – not really damages but flickering of the light – are voltage dips. Especially if these cover more or less one semi wave, during this semi wave the current will drop over-proportionally with the voltage dip: Since current starts to drop during the voltage dip, voltage drop across the lamp rises (Fig. 1.1) and leads to an acceleration and amplification of the current decrease. Subsequently, the full flux density of a normal current peak is by far not reached, which turns the next (opposite) current semi-wave into a milder form of starting process with excess current peak (Fig. 3.5) if the voltage is normal during that next semi wave. This way a positive flicker may occur, i. e. excess brightness above normal peak value, even though the voltage is normal during this particular semi wave and has even been sagging the semi-wave before. If the inrush current or other current peak caused by some nearby device (Fig. 3.4) happens to hit more or less equal parts of two subsequent semi-waves, only a normal current sag with consequential brightness sag occurs, but also slightly amplified beyond the proportional magnitude because of the non-linear behaviour (Fig. 3.6).

Basically the same occurs when DC impact causes a slight voltage asymmetry in the network. Old hair dryers, when operated at half power, normally use only one semi wave, and when the network resistance is high, a fluorescent lamp with magnetic ballast operated on the same circuit may flicker visibly. After all a measurement showed that a direct voltage content of 6 V, representing 2.7% of the line voltage rating, caused a direct current of 92 mA to flow through a ballast and lamp circuit, representing 18.1% of the rated lamp current.

Fig. 3.6: Current dip caused by an approximately symmetric voltage dip
Fig. 3.6: Current dip caused by an approximately symmetric voltage dip

The »negative resistance« of the lamp also leads to an over-proportional variance of brightness with deviation from rated voltage, while electronic ballasts including compact fluorescent lamps (CFL) claim to compensate this by means of their electronic control. What remains left of this promise will be discussed in section 6.2. Still, the loss of brightness at undervoltage is a lot less than with incandescent lamps, the efficiencies of which, performing poorly anyway, drop dramatically when operated below the rated power input.

But this is virtually all that may happen with magnetic ballasts. Adequate means to reduce the voltage flicker – and over-dimensioning of cables and especially transformers is in many cases enough to achieve this – just need to be provided, which will be necessary for other sensitive devices anyway and as a second effect reduce energy losses. Damages or failures of lamps or ballasts on account of poor power quality do not occur.

Reliability of magnetic ballasts

Therefore, this chapter is even shorter than the previous one. The ballast itself hardly ever becomes damaged through surges or over-voltages because of its simple and sturdy structure and because it has to be designed to withstand its own self-induction pulse anyway. Surely it happens that a magnetic ballast fails on account of shorted turns in the winding, which produce excess heat and thereby further turn shortings, then current increase, even more excess heat and so on. It may take some weeks, however, before this process reaches this avalanche state and finally blows the fuse. By then the fault may have been detected because of charred smell or uncommon noises, but in any case the ballast (and the adjacent lamp which is overloaded by the excess current) will under no conditions cause a fire during failure. This is the only type of failure that ever occurs with magnetic ballasts, and it is really the exception.</p>