But there is L of _some_ value here....

For the HV, how about making just the HV for striking independently from a DC capacitive multiplier, injected to the lamp "high" end through a diode, in a moment when both switching mosfets are OFF so it is like

1. preheat

2. pause inverter & let the HV strike the lamp

3. run

The L_of_some_value would have the same impedance in the resonance as the capacitor, so the same reactive power.

And for the separate HV pulse (first pretend it would be sufficient to start the lamp), just a diode won't be enough.
At first the transistor on the other side would be exposed to that HV as well. That means you will need way higher voltage rating for that. And higher voltage rating means it has higher losses during normal operation (both because of higher voltage drop, as well as because higher voltage devices use to be slower, so higher switching losses). Plus you would need a separate HV source for that pulse as well.

And for the starting: Just an ignition pulse is not sufficient for most lamps, it needs to be followed by some significant current fed into the lamp, when the arc voltage drop is still high (because the ionization is still building up). And for that the 150V from the halfbridge is not enough (without any resonance boost or so) and the ignitor spike would be of too short time to create anything else than just the initial ionization.
Normally the resonant circuit supplies the current the whole time the voltage drops from the 600V down to the 70V operation. With the classic magnetic coil the coild delivers the current for some 100's us from just the first kick (because of the large inductance) and then the 230V OCV is sufficient to feed that as well...


Indeed, there may be designed some complex contraptions to operate all lamps according to the "PL-S" standard, but the complexity, size, cost and losses would make such creations worthless, mainly when comparing to the mains frequency series choke...
So if something would be supposed to replace that choke, it is only a simple, single stage inverter working as  resonant preheat and start ballast. And that would require some restriction on the capacitance to work (so modifying the standard). Anything more complex becomes just pointless...

I thnk of some up converter can be made as a secondary output on the main inverter etc) charging a capacitor to ~1KV for the ignition source. The energy getting into the lamp would depend on the capacitance - choose big one and the lamp will strike completely

And how about instead of discharging that capacitor into the lamp, discharging it into a tap in L while one of the switching transistors is shorted (so no HV problem there), to act similar to HID ignitor

Precharging the capacitor: If the capacitor would be so large to have enough energy to form the arc, it would be at the same too large to allow stable arc operation. So you would need some extra series resistor. And that will become lossy normal lamp HF operation.
Plus still the transistors would have to be able to sustain that voltage.

Coil with the tap:
You have made just a single, few us pulse. What about the next 100us for the arc to build up?
Plus the eventual capacitor inside the lamp would effectively short out that pulse...
You are thinking just about a capacitor-less lamp, but the ballast will have to work both with and without the capacitor parallel to the starter. And those "with capacitor" are the majority of them (all made by reputable makers are like that)...

And if it will suffice with "just the majority" of lamps, the simple resonant inverter, relying on the capacitor inside of the lamp, will work the best. But it wouldn't achieve the "ENEC" certificate, because it would not support all lamps meeting the standard. But it will still be way better than a complex creation supporting just the minority of those lamps, moreover the lowest quality ones...


By the way the resonance is exactly the method, which let the HV to build up across the lamp without exposing the transistors to that. In exchange they have to handle the resonance current (Ires = Vres/Zo). When that is reasonable, they can handle it for few seconds before overheating.

Capacitor discharge directly into lamp :

Place a mosfet etc that keep it disconencted, except when charging and discharging. High voltage one, but since it is OFF in normal working it does not affect losses



Capacitor discharge into tap :

Place a mosfet etc that pulses it into the tap

Actually scratch that, just pulse the tap section of the coil across the main filtering electrolytic, using one of the "run" mosfets at the beginning of the coil, and one HV mosfet at the tap

Since the "run" mosfet is shorted the voltage across it is 0, and across the other one is equal to the rectified mains voltage - not too high. The HV mosfet will stay open in normal operation so no losses there either

Capacitor discharge directly into lamp :

Place a mosfet etc that keep it disconencted, except when charging and discharging. High voltage one, but since it is OFF in normal working it does not affect losses

But then the current will take the least resistance - into the halfbridge via the ballasting inductor...

Capacitor discharge into tap :

Place a mosfet etc that pulses it into the tap

Actually scratch that, just pulse the tap section of the coil across the main filtering electrolytic, using one of the "run" mosfets at the beginning of the coil, and one HV mosfet at the tap

Since the "run" mosfet is shorted the voltage across it is 0, and across the other one is equal to the rectified mains voltage - not too high. The HV mosfet will stay open in normal operation so no losses there either

The inductor of a HF ballast allows just very short time before reaching it's saturation, that would be way insufficient for the lamp to build any useful plasma. Therefore normally the ballast use a sinmewave between the saturation limits, so effectively a train of pulses. And so we are back to the resonance...

Plus even when such short pulse would suffice to ignite the lamp, you should not forget there can be (and in most cases will be) a capacitor directly across the lamp. And that capacitor will effectively short out any of such HV pulses (HID ignitors have regularly way larger pulse transformer and still are limited to about 200pF of load; here the load is 2.2..6.8nF, that is even above the range of long range ignitors requiring the full 50Hz ballast as the transformer, just because of the required pulse length, to operate into such capacitance; and HID ignitors are still not able to strike a fl;uorescent, the pulse is just not enough)

What about place a diode into between coil and lamp, so that the pulse train can just chgarge up the capacitor in the starter up to the voltage at which the lamp strikes, then continue it to build up the plasma. Then move to normal working, when the diode shorted by a mosfet

What about place a diode into between coil and lamp, so that the pulse train can just chgarge up the capacitor in the starter up to the voltage at which the lamp strikes, then continue it to build up the plasma. Then move to normal working, when the diode shorted by a mosfet

At first: The MOSFET would have to be high voltage rated. That means it will be rather expensive and have rather high Ron, so high losses during normal operation (it will have to carry the normal lamp current).

At second: The control part would have to feed the gate with something like 10V towards the source, all that floating at an AC potential of ~200Vrms (for the glow phase).
Third: How the starter would respond to that? Don't forget it is still there... With the resonant start the lamp ignition and run up is faster than the time the starter needs to close in the first time, but here the charging would get a bit longer.
Plus how to preheat the lamp filaments? With the resonant starting the 0.5A resonant current provides quite a lot of heating power, but once there is no capacitor in the lamp?

Of course, you may add components so long it may somehow work. We are talking about roughly 10 components just for that single MOSFET. Plus at least 5..10other components to generate the HV. And many of them will dissipate power during normal operation, so reduce the overall efficiency.
All that addition seems to be more complex than the full resonant start ballast. That means extra space, cost and losses. Do you really think something like that could be still commercially viable? With the competition of a simple series choke (the efficiency benefit gone, size benefit gone) or the use of the 4 pin lamps instead (no compatibility problem)?

Yea i see the point, though still thinking of whether it is somehow possible...

Back to the start. My gripe with the 4 pin PL-S is the width of the base with the side pins. This prevents making a cover-cap socket for it, like this socket for the 2 pin....


And it looks like nobody care much to make ultra lowloss chokes for them (allthough thats not technically a problem)

That socket looks like there should have been a amtching ballast for these in the form of being hng by the cable or with the thread for the chandelier tubing (the same as many E27 sockets ave). The question indeed is, whether these ballast ever made it to the market.

By the way my parents had E27 ballast adapters featuring exactly this socket. However quite soon after I have gutted these sockets and put them into a former electronic CFL base and use that for the next 15 years as PL-S adapters. With the resonant start relying on the capacitor within the PL-S and working perfectly with Tesla's "DZ" series, as well as Philips PL-S and Osram Dulux-S.
I made more such setups using regular PL-S sockets (as fixture conversion to these CFL's), even two setups operating three PL-S 7W on a single inverter, it worked well too. Of course, unlike the original adapter, my conversions wouldn't be able to attain the ENEC, but that is, what convinced me the problem of an electronic (HF) ballast compatibility is just to specify the capacitor and made a restriction to just the glowbottle starter as part of the standard...
The reason for that conversion were the notorious violent explosions of the incandescents there (why to bother with CFL's in the first place) and then the mass of the magnetic adapters - I was affraid the fixtures won't be able to support that.
And one side note: The tubes tend to last about three times compare to their normal life with the rated ballasts with the build in glowbottle preheat (both types of conversion were operated at the same time).

BJB (another manufacturer) indeed made the 4 pin PL-C version of this socket with built in ballast at some point (the PL-C does not have the side pins, so it fit in this sort of socket). But that is the exception...

I am interested in this sort of socket for lantern construction, where the choke can be placed somewhere else in the lantern - such as lamp hanging with a shade and choke is in the ceiling connection box