| The electronic uses way simpler connection and magnetics, because there is one more variable to control what is happening: The frequency, so the possibility to use higher Q resonances while still sufficing with wide tolerance (so cheap) components.
Another principal difference is, the capacitances needed to achieve some resonance boost are very small, so do not pose arc stability problems (the common discharge negative dynamic resistance characteristics mean the discharge tends to be unstable and/or extinguish and cause huge current spikes when connected to a capacitor beyond certain value, the high frequency means all capacitors are way below that).
So as with mains frequency ballasts the voltages are given only by the turn ratio of the windings, you are quite stuck with the voltage aimed to be a kind of compromise between reliably igniting after the electrodes warm up, vs not igniting when they are not warmed up. That leads to unreliable starting at cold temperatures or cold electrode starting at high ambient temperatures, affecting lamp life. And for the filament heating you are stuck with warming them just enough to get the emission, but not enough to not overheat them or not losing too much power on them after ignition.
With the HF ballast you may easily form a series resonant circuit to boost the circuit current for filament preheat at the same time with generating the voltage you want either for the lamp to not ignite or to reliably ignite it, then operate with the coil as a ballasting impedance once the lamp ignited, all that by just manipulating the frequency.
And in that way you can easily make the ballast a "programmed start" - first you operate them just at frequency when the current is rather large (2xIarc is rather easy to achieve), but the voltage by far insufficient to ignite the arc, mainly on a cold electrodes (like 1.5x the normal arc voltage), then you sweep the frequency towards the resonance so while passing it you can easily get 5..10x the normal arc voltage for ignition, reliably igniting even a stubborn lamp (e.g. at low temperature, contaminated surface, long time unused,...), once the lamp ignites the arc reduces the Q so much the extra voltage and circuit current disappears practically instantly once it ignites (this usually leads to current below the normal operation), then the frequency sweeps down further so the current, now with the arc present, rises to the operating.
So all that could be done by just starting at one frewuency, staying there for the 1 second or so of preheat time and then sweep over ~100ms down towards normal operating frequency and you are done.
Or with even the simplest selfoscillating ballasts the power stage forms an oscillator, running at the series resonance of the lamp circuit, so building up the voltage for ignition and preheating current. But then there is either current dependent element (a saturating feedback transformer core), which shifts the frequency up so the current (and indirectly the lamp voltage) stays at a reasonable level, or a voltage drop across an extra capacitor reaches the DC bus, so via diodes the circuit gets loaded by feeding the power back to the DC bus feeding the oscillator circuit, so limitting the voltages and currents. In this state the filaments get heated up by really high currents (3..5x Inominal), but their temperature is automatically limited by the lamp igniting at the moment when they reach emission, as the ignition reduces the Q of the resonant circuit so much there won't be any voltage, nor current boost anymore. But with the selfoscillating types the control over the voltage during the pre-ignition phase is very poor, so usually the lamp cold starts instantly. But with cold cathodes the circuit still has rather high Q, so high filament current boost, which makes this damaging operation last for very short time, so leading to comparable switching cycle wear to the magnetic RS ballasts even when it is instant starting.
The later method is very common in disposable CFL's (with just the feedback core saturation control) and the cheap electronic ballasts. The problem with these is any protection against EOL lamps becomes rather complex, so often many designs rely to the filament melting and breaking once the electrodes lose emission so the preheating stage takes way too long. But a bit "stronger" lamp and/or even seal failure (so the air gives filaments strong cooling) means the filament do not break so the large currents and voltages overheat and destroy the ballast.
|
