Hello:

I want to see the internals of a typical american electronic rapidstart ballast, because of the following reasons:

One of the members of LG, said me that an SRS ballast (Like Eltam "Rapidstart" ballast for 40W T12), uses the capacitor to start the lamp (No capacitor connected = no ballast).
Electronic ballasts also uses a capacitor to start the lamp, and here is the similarity between the operation principle of an electronic and an SRS ballasts.
By contrast, in a true rapidstart ballast (Like the american HPF rapidstart ballasts and the Eltam and Shaingout rapidstarts for a single F96T12 HO-O 110W), the capacitor aren't used for starting at all but only for PFC much like in preheat (NPF rapidstart = Without cap, HPF rapidstart = with cap). Instead, they uses a step-up autotransformer to start the lamp.

Despite there are no ways to make a real electronic rapidstart ballast, by using the capacitor for starting, the fact that electronic rapidstart ballasts exists in the americas + I have evidences from some american memers from LG, that electronic rapidstarts have all sorts of the behaviors of a magnetic rapidstart (Same starting, same lamp EOL show, if one lamp extingushes in a multi-lamp fixture that have a multi-lamp ballast, the rest of the lamps glows dimly instead of extingush), brought me to a conclusion, that the american electronic ballasts (Either instant, rapid and programmed start), lacks of a starting capacitor that all of the 2xxV electronic ballasts have, and that they probably uses a different mechanism for starting.

So I wants to see the internals of an american electronic rapidstart ballast, to verify my long time guess, that the american electronic ballasts, like the magnetic Slimline, RS and Trigger Start ballasts uses a step-up autotransformer for starting, rather than a capacitor (Which may also explain their similar case appearance to the american magnetic autotransformer ballasts outside, and their larger size and shorter length compared to 2xxV electronic ballasts).

US RS ballasts:
There are two kinds: NPF and HPF, differing not only in power factor, but by the internal circuit (it is not only a question of power factor sorrection capacitor like with series chokes, it is really different way of working).

I found here a schematic of two lamp HPF RS ballast.
The re are two basic components:
1) Transformer, it have three functions:
  Generate 3.6V for lamp's filaments (three 3.6V windings, one connected to the Neutral side)
  Step up the OCV to ignite the lamps (made on purpose with high leakage inductance, to smooth out the lamp arc current)
  Compensate the phase of the ballasting reactance from the main ballasting capacitor (C1), so the primary current is kept minimum (this is done by keeping an air gap in the main magnetic circuit)
2) Ballasting capacitor (C1): It's primary function is to provide a series impedance, so regulate the current when the lamps are lit.

The C2 is there to "leak" the high voltage across the bottom lamp to help the ignition, the top one is supposed to ignite afterwards (or by the field from the grounded reflector acting as an external electrode)
C3 is there to suppress HF oscillations, so eliminate the eventual RF emission.
Resistors parallel to all capacitors are there to discharge them after power OFF, so they won't become a shock hazard for the technician.


The NPF ballasts are lacking the C1 and utilize only the leakage inductance as the main ballasting impedance. That mean all the reactive power have to be handled by the primary winding, what make there higher losses. Therefore such concept is used mainly with lower ballast factor (feeding the F40T12 by 25W in residential fixtures), where the overall lower arc current keep these losses under control (about the same as on the full power HPF unit of the same mass)

In both cases the OCV for lamp ignition come from the step-up transformer and filament heating from the auxiliary windings. The filaments remain powered, but as they heat up during operation, their resistance increase and so reduce the heating power. Moreover the heating voltage is nearly 90deg off-phase from the lamp arc current, so the arc current split between both terminals of the filament, reducing it's load. So the overall filament load is about the same as with the preheat start circuit (the lamp rating came from)


The 230V SRS lamp feed concept originate from the series choke, it only use two parallel wires for the ballast choke. When the lamp is not lit, the current into the PFC capacitor flow through each of the wire in the opposite direction, so practically cancel out their magnetic effect for the main flux loop in the core. Only the leakage inductance between these two wires is seen and it slightly rise the mains voltage for the lamp ignition (few percent) But as it flow through the capacitor, it heat up the lamp filaments.
When the lamp ignite, the arc would become rather low impedance, so the arc current split between the wires:
From the mains flow the almost real 40W+losses power (so about 0.22A on 240V and F40T12), from the capacitor flow it's phase compensation contributor (so about 0.35A).
In fact there is no real resonance, the name is a bit misleading there - the presence of the inductors and capacitor may suggest the resonance to someone not knowing exactly, how it really work...
And because the OCV is only about the same as the mains voltage, it become insufficient to ignite the thinner F36T8.


All HF electronic ballasts I've seen utilize the real resonance to build up the high voltage for lamp ignition.
Some of them utilize it even to boost the voltage for running lamps (e.g. 120V ballast with bridge rectifier and a halfbridge inverter feeding higher voltage lamps)
Some use the resonance capacitor (or part of it, in case the resonance is used to boost the voltage for lamp run) current to heat up the filaments for ignition, some other use auxiliary windings on the choke, some use no auxiliary heating at all (the real IS ballast - use only one wire per lamp end)
And except of the real IS ballasts (with no filament heating at all, so single wire per lamp end), all are technically either rapid start (heat up the filaments together with applying the high voltage for ignition; even when they appear to start instantly - the filament power is so high, it take fraction of a second to heat them up) or of programmed start (the application of the high voltage is delayed, so the electrodes fully warm up before the ignition)



Now with multilamp ballasts, the individual lamp circuits may be connected either in parallel, or the lamps may be in series on one branch.
With the "HPF RS" parallel connection, each branch have it's own ballasting impedance (so at least separate "C1"), in case of NPF ballasts (and with better quality HPF as well) the separate secondary winding, so the leakage inductance become independent for each lamp.
Here one branch may be made as "lead" (ballasted by the C1) and the other as "lag", ballasted only by the leakage inductance. As normally the reactive power cancel out in the magnetic domain, the primary winding is not loaded with it. With certain lamp and mains  voltage relation it may even happen, then the primary current drop to really low value, so does not dissipate at all. But such balance is then very sensitive for the real lamp arc voltage, so using lamps with different voltage may disrupt this balance, heat up the primary and that additional heat may lead to overheat of the ballast assembly.
This lead-lag I would expect mainly in 4-lamp ballasts (2s2p - 2 lamps in series in one branch, one lead and one lag branch)



The HF electronic then have the ballasting inductor and resonant capacitor individual for each lamp. Because their resonance may fifer due to component tolerances, such ballasts have to provide a frequency sweep for reliable lamp ignition, so to ensure each of the branches goes through the resonance and generate the ignition voltage. This is a bit of complication for the selfoscillating concepts, where one branch may take over the full frequency control and the second is then not able to ignite it's lamp (when the first lamp is faulty,...).

Series operation mean the arc circuits are connected in series, so two lamps suffice with single resonant circuit.
But as the overall voltage become higher, normally the resonant capacitors is split to two in series. When the resonance current is used to heat the electrodes, such split mean each of the capacitor is connected across "it's" lamp. Then the overall circuit have one resonance frequency, so it is easy to feed.