Intersting
Could you write the accurate informations about the paramters of a components in the ignition system?

Maybe I will do similar ignition system to my lamps

I see it is the "SN57/58" pulser part, only inductors replaced by capacitors. But the concept of letting the resonance effects build up the higher voltage across the capacitor to start the lamp, is the same. Only here is no transforming up (compare to the HID ignitor).

And I would guess it is more the RS starting concept, than the preheat - the high voltage is applied at the same time as electrode heat up...

I think that it's older and easier construction as HID ignitor (SN58). Do you know how parameters of the components it have? Especially about the ignitor between electrodes. I will build the same ignition system but I must know how voltage (and other parameters) it have.

The triac is virtually any 800V triac, the capaitor is few 10's of nF and the resistor is set so, it would trigger just around (or a bit later after) the mains maximum and do not trigger when the lamp is lit (so have below 100V). Otherwise it act as a regular dimer, only set to fixed position.
The coil should have high or no saturation current limit, so an "air-core" coil would be the best.
The PTC serve there only as a failure-to-strike protection.

How big are the coils in this ignition system? Where can I find it?

I built this ignition system in two versions:

1) the same version as in photo but without coil ahead the power capacitor (I think that it's unnecessary). I noticed one problem and I don't know how can I eliminate it. The 15W T8 tube is no starting in all turning on. Sometimes I must turn on the lamp one more time and the tube starting. I think that the voltage impulse is smaller than the ignition voltage of this tube. I tested some coils (26-75uH) but the effect is the same... Who can help me The tube is starting hot (about 0,5 sec to heating the electrodes) - the same as vladi described in different photo about start this lamp.

2) version with exchanges the elements - classic magnetic ballast as current limiter and the capacitor in the ignition part of this system (exchange the coil with capacitor). In this version tube is beautiful starting immediately and in all turning on

In this circuit is no means to boost the voltage above the mains, maybe except tghe doubler effect (charging the capacitor in one polarity makes the charged voltage to sum up with the mains in the next half wave). The triac thing on the right only preheats the electrodes and maybe forms the doubler with the capacitor.
The fact the lamp started only when you have fiddled with it means your hand have formed an external electrode, lowering the ignition voltage so, the mains become sufficient.
The inductors there are to limit the current slopes (di/dt) in the circuit, so protecting the triac. When you remove them, the circuit will still work, but the triac will be overloaded (exceeding the maximum di/dt) and so tend to fail. This overstress does not generate any excessive heat on the whole triac, so you will not observe anything like higher operating temperature or so, it is just very local overheating on the semiconductor.

When you replace inductors with capacitors (your second experiment, as I understood it) means once the triac current dropped to it's holding current, the triac switches OFF. That means there is steep current ramp, forming a fast, higher voltage spike igniting the lamp quite easily. That spike could be easily above 1kV and it's generation uses to be killing for regular triacs, making troubles in inductive circuits with triacs (so snubbers are necessary to protect the triacs there). There are special triacs on the market designed for the snubber less inductive switching, but these then won't work as the lamp ignitors (the principal difference towards standard triacs is really significant reduction of the holding current from 30 to 100mA range of standard triacs to sub-mA range for the "snubber-less" types, so these triacs do not generate those high voltage pulses in the first place)

I used also the PTC in parallel to all inition system, so it realise hot start in magnetic ballast. I think that it make less voltage impulse when lamp starting. And also I use the PTC in series with the triac / capacitor ignition circuit, similar as in the circuit from Circolux. The PTC in series with ignition circuit protect the triac when the lamp is worn out. In small tubes (about 8W) the voltage impulse is usually smaller than 1kV. I will do the tests with longer tubes.

If the lamp with capacitor as current limiter (for example this Circolux) haven't external electrode, why it starting in all turning on Vladi doesn't writing any about problems with starting in his Circolux. How can I eliminate this problem in the same technical construction as in Circolux?

The lamp internal design is a strong factor for the starting voltage. At first, it is the gas fill. It could be some form of a Penning mixture (in mix of gasses the breakdown voltage uses to be lower than with any of the components in the pure form at the same pressure and temperature; the ratio with the lowest breakdown field is then called "Penning mixture"). Then the pressure of the fill plays it's role as well.
Then some lamp designs use a conductive coating inside of the bulb (e.g. the US F32T8).
Then the exact electrode shape plays it's role as well: Straight single coil filament, bended into an arc, triple coiled,... That influences, how much such electrode generates a corona discharge,...

So no wonder otherwise very similar tubes (length) may have different striking voltages, some needing external electrode, some sufficing without. And then the only thing you may do is use the same lamp, as it could well be, the F15T8 will just not work with this ballast type...
Don't forget the circular lamps are more likely expected to work without anything metallic around, while the typical F15T8 fixtures have at least the bunch of wires close by, these could have been factors taken into account when designing those lamps...


And don't forget nobody said the Circolux haven't used specially modified lamp - it is a CFL, so it may rely on special tube design...

Plus when you have added the parallel PTC, it may discharge the capacitor before the next halfwave, so lower the voltage available for the start, so not reaching the lamp breakdown voltage.



And for the voltages seen by the triac and the ignition spikes: The voltage may have been below 1kV, but triacs are generally rated lower and do not like being exposed to voltages above that limit. Mainly when the current is insufficient to trigger them properly (their exposed junctions stay in the avalanche breakdown mode, so with high voltage drop, so generating locally a lot of heat; when the high voltage is capable to feed higher current, the triac then turn ON to the low voltage drop mode, so the damage of the semiconductor is way less severe...). Because with the inductive load this pulse happens when the current drops below the holding current, the spike can never provide more, so the triac will always stay in the most damaging avalanche breakdown mode until the pulse disappears completely. I would not be surprised the observed 1kV would be just the voltage of the triac avalanche breakdown...
But the 1kV is usually sufficient for most fluorescents, even the long ones...

I used PTC to heating electrodes in magnetic version of ballast, not in the capacitor version (but I used it as safety equipment, the same as Circolux schemat in both versions).

My triacs are from 800V but I think that some longer tubes (20W and longer) want just about 600V impulse to starting. I will do more experiments and if the tirac will be breakdown, I will go to electronic shop and buy next one It's the same danger as when you used a Graetz in classic magnetic ballast, between ballast and tube with neon starter. The diodes must be in higher voltage than is generated in the ballast when the lamp is starting.

I know that some tubes have the special ignition construction to less voltage, for example with starting electrode inside (for example Philips TL-X) and some tubes starting just in 230V after heating electrodes.

The triacs are quite sensitive components, mainly compare to thyristors and diodes, so the diodes can handle the low current breakdown way better than the thyristors and mainly the triacs. So the diodes would most likely handle the spikes without any problems, if the overall power dissipation wouldn't be too high to really overheat them.
Plus even the cheapest 1N4007 are already rated for at least 1kV (many makers state 1.5kV for that type)...

There are special thyristors, specifically designed for the fluorescent starters (makers call them "fluoractor"), these have gate controlled holding current, their design controls their avalanche breakdown (usually 1500V), so it is not damaging for them. These thyristors are then the basis for the "electronic starter" designs, e.g. the Philips S10e or the "famous" Arlen "Pulsestarter".
The drawback is, these components are extremely sensitive for di/dt and peak current, in other words they can not tolerate any parallel capacitor at all (few 100's pF was already killing the S10e, tried...), so are usable only with the inductive ballast (could be lead type, but the inductor, limiting the di/dt and pulse current is there very essential for the thyristor to survive).

I built succesfully some examples of this "Circolux" ignition system - the capacitor + resistor as current limiter and triac with other components in starting part of it. I think that in last version (which starting not always), the triac could be damaged or had incorrect paramethers.

Are you sure that the coil in starting part of it (next to triac) was not a resonance with the capacitor in ballast In your first post in this page, you wrote that the coil could do the resonance effect to built higher voltage in capacitor, and later you wrote that it only protect the triac.

I built it in version for 15W T8 tube and it starting beautiful The same when I built it for 10W T8 tube and for T5 4W and 6W tubes. But when I try to use in it the 8W T5 tube, it was not starting and only the cathodes were lighted too strongly. Why the 8W tubes are not starting in this ignition system? It depends on ignition voltage of this lamp, or what When I used classic neon starter, the 8W tubes are starting without problems also in capacitor ballast.

There are resistors on the input and when you calculate it, you get Q factor of way less than 1 ((SQRT(47uH/5.3uF)/54Ohm) << 1), that means no resonance effect at all.
Initially I have neglected them, but later I made that calculation...

A series LC is used in SN57 like HID ignitor (balast inductance plus capacitor in the ignitor; but there the resistance is very low, so the Q factor high), it made me thinking of a similar functionality here. But the resistors on the input here make the story way different...

For thhe 8W tube it is strange: These lamps even coldstart just on the 230V OCV (and series inductor, but during start that plays no role, as there is not yet any current).
All the lamps you have mentionned suffice with 230V for hot electrode ignition (some even for a cold electrode start), so in fact need no boosting at all.
But the lamps may need an external electrode, practically in the form of a grounded metal fixture body or reflector in close proximity along the lamp.