Ballasts for fluorescent lamps

Basics and technical details

Important note regarding this page

In new lighting installations it may be assumed that the time of the fluorescent lamp is over since the LED technology has matured to become the standard solution. In existing installations, however, fluorescent lamps may still be encountered for decades.

Accordingly, this page is no longer updated or maintained, but it can still provide many a useful hint with respect to existing installations and is therefore still held available here, albeit it represents the state of the art and the knowledge from about 2010.

For operating a fluorescent lamp, a ballast is required. Why is this so?

Gas discharge lamps, this meaning lamps using the principle to make a gas electrically conductive and thereby light emitting, are a relatively old technique. Especially fluorescent lamps represent a very widespread lighting system. It is not possible to apply the line voltage directly to such lamp, be it AC or DC, a higher or a lower magnitude. Traditionally these lamps have always been operated on AC mains by means of a so-called magnetic ballast, which is nothing more than a reactor or choke, for limiting the lamp current. In recent years, as power electronics techniques came up, an alternative way of operation was introduced, the so-called electronic ballast, which converts the incoming mains frequency into a much higher frequency, usually in the range of 20 kHz to 80 kHz, to operate the lamp with.

The magnetic ballast method creates a huge amount of inductive reactive power, significantly exceeding the magnitude of active power, but this reactive power can easily and cheaply be compensated without risk of any interferences, if done adequately (see section 5). The electronic ballast does not – or should not – produce substantial amounts of fundamental reactive power (displacement power factor DPF or cosφ). It need not but may be designed to operate on different mains frequencies, including DC, and different voltages, thereby also compensating any input voltage variances. The decisive argument put forward for its use is, however, the energy saving achieved, not so much by lower internal losses in the ballast itself, but rather by an efficiency improvement of the lamp when operated at the high frequency supplied from the output terminals of such electronic ballast. For this reason they feed less power into the lamp than a magnetic ballast does. However, electronic ballasts are several times more expensive than the plain passive magnetic models and much more susceptible to certain disturbances and are likely to become themselves a source of disturbances. Unlike the magnetic ballasts, which as a law of physics can follow only one principle of working and only one basic design, power electronics provide a lush choice of design variants and working principles to design electronic circuits for operating fluorescent lamps.

To top

Introductory notes and the basics of physics

Fig. 1.1: Characteristics of a 58 W fluorescent lighting tube, measured with DC and approximated with an empirical formula
Fig. 1.1: Characteristics of a 58 W fluorescent lighting tube, measured with DC and approximated with an empirical formula

Gases are generally not electrically conductive but may become so under certain conditions, just as any insulant becomes in a way conductive as soon as the breakdown voltage is exceeded. With gases the initiation of conductivity proceeds in three steps. Getting the procedure started at all requires the presence of at least a few charge carriers, traces of which are always present in atmospheric air and other gases, mostly showing up as ions, but also as free electrons. Their quantity is normally too low to get a current flowing. The dielectric strength of gases, however, drops as pressure drops. This looks like a contradiction at first sight, for less gas per volume of course also contains fewer charge carriers per volume, assuming the relative content remains the same. Strangely enough, fewer charge carriers indeed induce conductivity sooner, whenever this happens in a more indirect manner: You have to bear in mind that the conductivity is generated because the ions see themselves exposed to a force in the electric field and are accelerated (Greek ionein = to migrate). Of course »migration« is a severe »disexaggeration« in this context. In fact the »migration« speed of the particles is to be measured in kilometres per second. Ions may have been detected in aqueous solutions first, where indeed they just creep along at less than one millimetre per second like electrons in metallic conductors, wherefore they may have been called ions.

If the charged particles reach their target and get into touch with the electrode, they give away their charge and become a neutral molecule or atom, respectively the free electron is absorbed by the electrode metal. The way to get there, however, is not an easy one. Hardly have the charged particles gained some speed, they collide with other, uncharged particles and need to gain momentum again. If the density of air or other gas is very high, the next collision will occur rather soon before the ion has gained any nameworthy kinetic energy. But as density decreases the average free length of flight increases and thereby also the likelihood for the ion to gain enough kinetic energy to hit one or more electrons out of the next struck gas molecule, or to smash same gas molecule and thus generate two or more new charge carriers. As soon as at average each charge carrier before reaching the respective electrode has generated more than one new charge carrier an avalanche effect starts, and this explains why for operating fluorescent lamps and for light arc welding appropriate measures have to be taken to restrict the current flow: A plasma has been generated, which is to say a mixture of gas molecules in their original unchanged state with substantial shares of ions and free electrons. These individually travel from one electrode to the other, the positive ones in the opposite direction as the negative ones, forming the current flow, which in atmospheric air now only more takes some 30 V to maintain, at high current densities even less than that. In this state the plasma protects by contracting and separating itself from the surrounding air through the magnetic forces of the current, which enhances the current density and reduces the heat dissipation. It must not be forgotten, however, that at these extremely high temperatures a lot of heat is dissipated through heat radiation. Heat dissipation through radiation increases by an exponent of four with absolute temperature!

But this is already the final stage of conductivity in a gas. The first stage occurs at very low current densities around 10 nA/mm² and without any light emission. The second is the glow discharge stage at current densities up to about 1 mA/mm² and is thereby the one that is used in electric luminaires from the glow lamp to the fluorescent lamp. The working principle is the same in both types of luminaires. In the fluorescent lamp the luminous section of the gas column is artificially very much extended. The light itself is ultraviolet and therefore invisible but causes the fluorescent layer inside the tube to shine. So by varying the composition of the layer the colour of the light can be varied. The third stage then is the one called light arc, ranging up to some 10 A/mm².

What all of the three stages of gas discharge have in common is that the voltage required to sustain the current flow drops as current increases. Ohm's Law seems to be perverted into its opposite. With some justification you could speak of a »negative resistance«, for the differential quotient du/di indeed is negative (Fig. 2.1). However, this seems comprehensible in this case, for the higher the current, the more charge carriers are generated.

To top