Emergency lighting cycles without electrode heating before lamp ignition usually damage fluorescent lamps.
Emergency lighting cycles usually damage fluorescent lamps as most emergency gear does not provide a programmed electrode heating cycle before lamp ignition or external heating during low power emergency arc discharge. These shortcomings are amplified by South Africa’s currently fragile grid, causing excessive emergency lamp damage in lighting installations.
![]() Fig. 1: Blue glow. |
![]() Fig. 2: Stable electrode. |
In the perfect scenario of a fluorescent lamp strike, electrode thermionic emission would need to be attained before lamp ignition. This is known as a “warm start”. A warm start is necessary to prevent electrode damage by providing a protective electron cloud around the electrode and thereby ensuring long lamp life. Improper lamp ignition produces a momentary blueish cathode glow adjacent to the electrode. This phenomenon is clearly visible to the naked eye and is shown in Figs. 1 and 2.
Fig. 1 shows a longer bluish glow that indicates damage to the electrode via ion bombardment. This momentary event causes considerable electrode damage that can render the electrode useless in mere seconds. It should be noted that emergency cold strikes only occur if the lamp was not powered before the mains failure (non-maintained mode or if the maintained lamp was switched off).
Fig. 3: Measured voltage required to attain thermionic emission for the various lamps.
Electrode damage is characterised by lamp end blackening. This blackening is a consequence of both vaporisation and ejection of electrode oxide onto the inside of the lamp end glass.
Furthermore, external cathode heating is required for all fluorescent lamps that operate below full power. This will ensure that the electrodes are protected against ion bombardment by an electron cloud which is normally provided by Joule heating during a full power discharge. This provision is hardly ever met with emergency fluorescent lighting due to expense and an inevitable reduction in lighting duration. T5 fluorescent lamps are most susceptible to electrode damage due to their fragile filament construction.
Fig. 4: The measured resistance of the electrodes of various
lamps at room temperature.
Research
Cosine Developments conducted extensive research into this phenomenon to establish the viability of a programmed start emergency ballast. Interestingly, the research also revealed why some common warm start ballasts actually damage electrodes, resulting in lower-than-expected lamp life. The following T5 lamps were used to verify this hypothesis:
Fig. 5: Power needed to allow thermionic emission.
Firstly, it was essential to first characterise each electrode to ensure optimal starting scenarios. There are two methods of driving the heaters: constant voltage or constant current. Constant current is universally favoured, especially by those ballasts using positive temperature co-efficient (PTCs) resistors.
The onset of thermionic emission, a dull red glow, is visible through the bulb wall. The electrode resistance increases with increasing dissipation (tungsten has a positive temperature co-efficient), so it is important to establish their resistance at both room temperature and at the threshold of thermionic emission. Each electrode of every lamp was energised to establish the onset of thermionic emission. Fig. 3 shows the measured voltage required to attain thermionic emission for the various lamps.
Fig. 6: A typical electrode warm-up time.
It is clear that the electrode characteristics vary across lamp wattages and follow no obvious trend. Fig. 4 shows the measured resistance of the electrodes of various lamps at room temperature and during thermionic emission.
As expected, the resistance at thermionic emission is three to four times that measured at room temperature. The power required could be computed from this data. The results are shown in Fig. 5.
Note that the power threshold varies from 0,56 to 1,45 W to ensure thermionic emission. This power would cause excessive battery drain and reduce the emergency illumination period substantially. We therefore opted to drive the lamp with DC current, which means that only one electrode (the cathode) requires heating power. This halves both the starting and static heating power drain from the battery. Mercury electrophoresis (migration of mercury towards the cathode) due to the polarising DC current will be reversed after the re-application of mains power.
Fig. 7: Time chart of startup.
These optimal conditions for cathode heating assume zero lamp voltage during the heating period. Circuitry was therefore constructed to apply heating power to one electrode while holding off the lamp voltage. External heating power can be significantly less than starting electrode power due to Joule heating from the arc current. This bonus can limit battery static drain to reasonable values.
It is also worth considering the thermal time constant of the electrodes. In our tests, we were examining the lowest possible power to achieve emission. This value would require the longest delay before ignition. Fig. 6 shows a typical electrode warm-up time with minimum applied power (three seconds). In this case, the electrode voltage was measured when current was 170 mA.
Prototype performance
After considering all this data, we developed a test unit with a starting scenario of three seconds of heater power, with zero lamp voltage followed by lamp ignition with reduced external electrode heating, as shown in Fig. 7.
Fig. 8: Starting cycle of a well-known European brand ballast.
The lowest effective heater voltage seems to be 4 V. The external heater power drain during arc discharge is approximately
50 mW, not much of an overhead.
The test results were very encouraging. A normal emergency ballast destroyed the lamp after one week’s continuous emergency cycle (using a mains derived power supply). The prototype warm-start unit had not caused any visible lamp damage after continuously powering the lamp for three weeks.
It should be borne in mind that the optimal starting scenario has a very narrow window. We measured a well-known European brand ballast whose starting cycle causes dramatic reduction in lamp life by being excessive (see Fig. 8).
The maximum rms voltage measured was 7,3 V. The maximum current measured was 880 mA and the maximum power dissipated by the electrode for 1,5 s up is therefore 6,5 W. Clearly, this lamp would have lasted longer without any pre-heat at all.
Contact Stirling Marais, Cosine Developments, Tel 031 579 -2172, stirling@cosine.co.za