i have a ballast with ONLY 3 output plots/wires
it goes not have the case/housing with wiring diagram.
it's a half bridge-1st output goes thru the inductor to the power transistors.
2nd output plot it's connected on the pcb thru one 1nF/2kV cap to the first output plot.
3rd output plot goes ,of course,to the capacitor divider of the half bridge.
physically,the pcb it's 24cm long,4 cm wide;made in 2004 year,nec power transistors,rubycon,nichicon,
15 other low power transistors,active pfc (no china).

that is the question:how it can be wired? 3 wires output!

Do you have a pic of the ballast?

Waiting for the picture

Preliminary guess : The HF output is symmetrical positive/negative relative to the center point of the 2 capacitor divider. In this case, the wire that comes there might be common for the AC input as well as being the Neutral on the HF side

no color code on the plots.just plain white plastic.
i think it's an "instant start" ballast-works with tubes "single pin" like philips TL-S series or ordinary linear with shorted out filaments.
my best guess,based on a old wiring schematic:two tubes (with shorted out filaments) SERIES connected between plot1 and plot3;the middle of the circuit goes to plot2,the one connected internally,on the pcb,thru 1nF/2kV to plot1.i'll try with 2*20w for starters...
the ballast was "declared" dead by someone because of a hairline crack/exfoliation at one INPUT plot (at 240v). exfoliation because the damned plot jiggled and ruptured the copper track-planned obsolescence/bad parenthood?!
the risk of experimenting?o,some UNOBTANIUM nec transistors....

The description is troublesome without a picture....

i have a ballast with ONLY 3 output plots/wires
it goes not have the case/housing with wiring diagram.
it's a half bridge-1st output goes thru the inductor to the power transistors.
2nd output plot it's connected on the pcb thru one 1nF/2kV cap to the first output plot.
3rd output plot goes ,of course,to the capacitor divider of the half bridge.
physically,the pcb it's 24cm long,4 cm wide;made in 2004 year,nec power transistors,rubycon,nichicon,
15 other low power transistors,active pfc (no china).

that is the question:how it can be wired? 3 wires output!

It indeed looks like a (slimline) instant start ballast with lamp presence locking:
The 1'st wire goes to one lamp end, it is the "hot" output.
The 2'nd and 3'rd are during normal operation connected together by the other lamp end (either a socket contact or a lamp filament).
The circuit then completes the resonant circuit only when at least the cold end of the lamp is inserted. That means either the lamp is connected normally, so it lights up, or there is nothing in the hot end.

This locking is meant as a protection against an electric shock, when someone would insert just the hot end of the lamp and touch the still bare cold end lamp contacts. Without such lock the ballast would generates high enough voltage, the lamp would ignite, so become conductive, so the still exposed pins on the cold end would become energized. With that lock the ballast just does not generate sufficient voltage to ignite the lamp, so the exposed pins can not become energized.

@medved:interesting ideea but i neglected to say that between pin1 & pin3 there is still another capacitor on the pcb:4.7nF/1000V

there are diagrams (on internet) for 2 instant start tubes on a ballast but
there are marked 2 "blue" wires-on equal footing,equivalents,for each lamp
and a single "red" common for both other end of the tubes.

here there are no 2 electrically IDENTICAL wires.

For ě lamps you would need either the two ballasting inductors (for a parallel circuit), so indeed twice the same set of components, or the common wire would go to the middle of two series capacitors (for a series circuit).

The capacitors you mentioned are really connected exactly like you have said? Isn't in series with one of the 1nF e.g. the primary winding of the feedback transformer (few turns of wire on a ferrite ring core; for the main functionality may be neglected)? In that case the missing cold side connection breaks the ballast oscillator feedback loop, so the ballast stops completely.
Or are not to that wire connected some other components (beside of the 2nF)? It could be some MOhm resistors or so... In that case the open circuit there is detected by the control part and the ballast controller is allowed to start the ballast only when the connection is reestablished...

May you post some photos? Maybe someone will recognize that ballast type and find the wiring diagram...

this is it.i'll scan both sides later.
just talked about the external connection and the 2 caps:1n little blue,4.7n big blue.
of course,from pin1 the circuit goes with a smd cap,2 series resitors to the control
section (15 other transistors on the back side),the big inductor (output one)has a tap that goes to control.

i was hopping that someone knew about such 3 wire thing.
i do not want to reverse engineer the gear or blow it in black magic smoke at the first try.

one post got accidentally deleted:
between pin1 and pin 3 there is a cap,4.7nF/1kV-the big blue.
 
IT WORKS!!!!!!!!!!!!!!!!

2 tubes TL-D 18W philips,series connected inter pin1 & pin3,the middle to pin 2-the one with 1n cap.
why pin2?it provides a mean to strike the first tube(inter 2 and 3)and only after that the second tube WILL strike (inter 1 and 2).
otherwise no one can strike 2 series tubes at the same time.
going in reverse,when you series connect capacitors to attain higher voltage you must place balancing resistors across every one of the caps.

QUESTION:how much can in TAKE?like 2x58W??!!!!!!
i mean active pfc for a total of 36W is just overkill,that's what china tells us.never mind "electronic transformers" for 12V/150W halogen without any proper mains filter,without capacitors,inductance or even transistors-the silicon chip inside china heads is so tiny that a 5-6 TIMES
DE-RATING (amperes) is useless because they cannot take THE VOLTAGE.

about bad ballasts-the end result is premature ejaculation of Hg-MERCURY into our lungs-about 1000 tonnes yearly.

 i think is an philips oem from e-kyoto series.
at least 2x36W,maybe 2x58W.
there is/was e-kyoto II series -sequel?
 e-kyoto stands for:ecology-kyoto 1997 protocol where the whole planet swore to cut off carbs-the result is that U.S.A grew it's carbon intake
FURTHER that the rest of the world (the second world-china,third world-europe,trailling,fourth world,etc).they are gulping like 29% of world's C,belching relentesly CO2,they
are putting a filthy show with venezuela(usa's 33% oil comes from da communistas of chavez extraction),etc,etc,etc.

maybe philips understood that nec,rubycon,nichicon are greener...