| It lies in how HF ballast work:
- It rectify the mains and by using the advantage of PFC stage convert it to ~400..450VDC. With this range and generic PFC controllers, the mains voltage might be anything between 90 and 277VAC or even 90..400VDC. This is intrinsic (normal) behavior of the basic active PFC circuits (even if ballast companies tend to name it as "extra feature" and give it weird names). Any decently made PFC should allow this AC voltage range, otherwise it is somehow ripped off to save cost. Only the DC rating might be missing, as DC supply is extra challenge for fusing - so operating AC only rated things on DC is fire hazard, even if it technically work, because fuses may not work correctly.
- The DC bus supply the main inverter, usually in half-bridge configuration, what generate from it high frequency square wave of the same Vpp as is the DC bus (so the amplitude is half). If it is necessary to supply really high voltage lamp (or the DC bus voltage is low due to omitted active PFC stage; when the lamp arc voltage is more then ~1/2 of the DC bus voltage; but it not hard limit, rather an optimum threshold, see below), full bridge is used to deliver twice the HF AC voltage.
The frequency might be controlled for passing trough startup sequence (real programmed start ballasts) or adaptive (regulate voltages or currents in different stages; note, then the simple selfoscillating circuit has adaptive frequency from it's nature)
- The HF AC voltage is then led via series ballasting inductor and DC decoupling capacitor to the lamp. The inductor provide an impedance to limit the lamp current (exactly as in "magnetic" mains frequency series choke ballast). As the frequency is under ballast control, the current might be regulated simply by changing the frequency. Or the frequency might be made adaptive to keep the current constant, even when other parameters change (mainly DC bus voltage by it's ripple). The fluctuating frequency is in fact a good thing, as it spread the ballast frequency spectrum, so does not allow parasitic resonance phenomenons to take place inside the lamp. Note, then this behavior is intrinsic to the basic "ferrite ring" controlled selfoscillating inverter - so was here way before "patented anti-striation features". In fact these "features" had to be reintroduced, as lot of "high end feature rich" ballast controllers were able to drive the lamp only by constant (preprogrammed) frequency, so resonance issues (striations,...) inside lamps popped up.
This coil might have auxiliary windings to provide source for voltage mode heater connection, if used (i think this was patented some time ago by some Israeli company, at least for CFL ballasts, but i don't know, if the patent is still effective). Heater circuits are then independent from arc supply, what have some advantages (better temperature control, ability to work with broken filament), but even disadvantages (filament can no more serve as EOL fuse, so the circuit need extra EOL protection)
- Parallel to the lamp is connected a capacitor ("resonant capacitor"), what form with the ballasting inductor a series LC resonator responsible for building up high voltage for lamp igntion. In ballasts with higher arc voltage this voltage boost is used even in normal operation - it allow the arc voltage to approach or even cross the "Vbus/2" limit (mentioned above). On US InstantStart circuits or circuits with voltage mode heater supply it is connected directly inside the ballast box, so it's current is not seen by the lamp at all. On current mode heater connection (EU style) it is in series with lamp filaments. On programmed start ballasts this provide the preheat current path, on instant start it only uses the filament as a fuse to protect the ballast (the high voltage strike the lamp way sooner then the filament has any chance to warm up, so even if the filament receive power during starting, such ballast is still instant start)
During an ignition attempt the current trough this capacitor become very high, what on simple selfoscillating ballasts might lead to component overheat, if the lamp does not ignite. In EU-style connection this current blow up the filament, so interrupt the circuit, so shut down the ballast.
And here come the trick with independent series operation:
Assume you want to drive multiple lamps in series in the US-style IS circuit (resonant capacitor is inside the ballast, so only one lead to each lamp end, or with voltage mode heater). You will need quite high voltage for the series chain ignition, so really high voltage rated resonant capacitor.
So you split it up into multiple pieces in series. And if you choose one capacitor per lamp and interconnect intermediate nodes with corresponding points between lamps and design the inverter frequency so, the ignition resonant current is under control, and all other components so it does not overstress them, interesting feature pop up: By removing (or failure) of one lamp, the rest stay lit. How it is possible? The resonant capacitor parallel to the failing lamp provide the current path.
Lot of ballasts use one common inverter for multiple lamp circuits (ballast inductor + resonant capacitor + lamp), but this does not allow to control the maximum ignition voltage, as the resonance curve is there very steep, so it is very sensitive for component tolerances. Together with the fact, then two lamps might operate easily in series and the energy directives (at least the US one) allow maximum two lamps to be dependent on each other, the two-lamp ballast in series connection (the easiest to implement and the most robust against component tolerances) is the frequent "configuration of choice" on fully protected programmed start ballasts.
|
