A few years ago i Saw two really interesting videos on YouTube, a person welded a starter on a electronic ballast and on a electronic CFL bulb and converted It to preheat, but i dont Saw the video on YouTube anymore. The video never explained how to do It, simply shows the preheat cycle. If anybody has the video link please share It and if the video is real please share how to convert the electronic ballast in to preheat, This would be a cool project to simulate the earlier preheat CFL bulbs.

Are you sure it was a starter connected to a circuit with an electronic ballast? Wasn't it just some kind of warm start (preheating the electrodes before ignition) ballast type?
Or it was just the tube from a CFL connected to a magnetic choke and starter?

Because in my experience if you just connect a glowbottle starter parallel to the resonance capacitor (talking about a HF electronic ballast circuit; in order to resemble a preheat circuit), the starter usually does nothing at all (until the lamp EOL), because usually the lamp starts way faster than it takes for the starter electrodes to warm up to close. So the lamp operation is an instant start like the starter was not there at all.

In fact the here well known Arcotronic CFL addapter from NIGG has an electronic gear but it works with PL-S 5-11W G23 and PL-C 10-13W G24-d1 cfl which always contains a starter. This adapter was patented in 1986.

I found the video about the preheat electronic ballast: https://youtu.be/CHwop9w5tsU

As you can see the starter makes the function really well, how This can happen?

In fact the here well known Arcotronic CFL addapter from NIGG has an electronic gear but it works with PL-S 5-11W G23 and PL-C 10-13W G24-d1 cfl which always contains a starter. This adapter was patented in 1986.

I made a few of similar adapters myself (selfoscillating 2-transistor ballast circuit from blown out integrated CFL, with a PL-S socket), that is where my experience with this setup came from. First time powered on a new lamp "preheats", but the next cycles it pretty much always stsrted instantly.
And with this I also found "the hard way" that not all "PL-S"-like lamps have the capacitor with the starter. Many cheepeese did not and those then did not work with that electronic ballast (the capacitor there is needed to both close the feedback of the ballast oscillator circuit so it starts to do anything, plus it becomes the resonant capacitor to facilitate the starting voltage boost).

I found the video about the preheat electronic ballast: https://youtu.be/CHwop9w5tsU

As you can see the starter makes the function really well, how This can happen?

It is the combination of the time needed to heat up the filaments in the glow (cold cathode) mode being slower than the time needed to warm up the starter electrodes to close. So the lamp does ignite instantly, but before transitioning to hot cathode mode, closing stsrter shuts it down.

When did that with the PL-S, the lamp was faster, so the starter did not react.
Probably the heavier, higher current rated PL-L filaments are significantly slower or the starter faster on the higher PL-L current (~0.33A at HF or so) than the thinner filaments of PL-S and 0.14A ballast, the starters appear of the same size electrodes, so the higher current likely warms it up faster.


Technically there is nothing in such setup that would guarantee anything, regarding the "blink happy" vs "boring instant" start.
But definitely the starter presence won't save the electrodes from the cold cathode sputtering wear, as the powerful glow discharge will always happen even with the starter. The starter just shorts it out at about the same time the electrodes would reach emission without it, so the cold cathode discharge takes place about the same time either way.

What would happen if you just connect the starter directly to the lamp, and just not connect the wires going from the capacitor.

I have a cheap 18w PL-L desk lamp that has an electronic ballast with only 2 wires coming out, and the capacitor is directly across 2 pins of the socket. I in theory could convert it to true preheat by just removing the capacitor, and putting a glow bottle in its place.

Without the capacitor the ballast itself won't start, so everything would be dead. Assume the simple two transistor power oscillator circuit, you may identify it by the presence of about 10..15mm diameter toroidal core feedback transformer with 3 windings of about 4..8 turns each, often wound using standard colored plastic sleeved wires.
If the ballast uses some advanced control IC (it was a programmed start,...) it may not work at all, as the starter would likely trigger some of the protection features these IC's use to include.

But if you take a standard starter as sold, there is normally a capacitor inside the can connected parallel to the glowbottle, so that will take over the feedback/resonant capacitor function.
The ballasts are not that much picky about the exact value, it has to be in the ballpark of 5..15nF. Of course, its value has impact on the lamp starting, but nothing that severe. Only with really higher voltage lamps (PLL36 or longer), where the voltage has to be boosted by the resonance even during normal burn, it will have some impact on the exact current fed to the lamp, or too high value may feed high residual filament current. Compare the input power with the original cap and with the stsrter, if the stsrter isn't more than 10% higher, it will be OK.

Just confirming, the ballast won't work unless the capacitor is connected to the lamp?

Interesting so tiny equipment.

Exactly. It won't work without the capacitor.

What would happen if you just connect the starter directly to the lamp, and just not connect the wires going from the capacitor.

I have a cheap 18w PL-L desk lamp that has an electronic ballast with only 2 wires coming out, and the capacitor is directly across 2 pins of the socket. I in theory could convert it to true preheat by just removing the capacitor, and putting a glow bottle in its place.

If I was to convert this desk lamp to preheat, I would rather use a proper full power F20T12 preheat fluorescent tube ballast and FS2 starter.

I tried a starter out of a broken PL-S lamp and it just sat there and preheated endlessly. Would an electronic starter like a Pulsestarter work?

Electronic starter won't work. They rely on the rather low mains frequency to operate. Plus they can not tolerate even a small capacitor - it kills the special thyristor in them (don't ask me how I learned this... :-D ).

Taking starter out of a "broken tube": I would expect this to be already bad, as EOL lamps tend to destroy the glowbottle starter (it tends to overheating even before it starts flashing permanently), you would need some fresh known good starter.