Went to Melbourne last week and managed to pick up some Osram Endura control gear, but unfortunately, no lamps as yet.
I procured enough units to strip some and extract the circuit. They consist of a two stage R.F.I. filter, followed by a bridge rectifier and a 470nF H.F. bypass cap. The next stage is a continuous-mode, boost architecture, crest factor corrector based around ST's L6561 chip and an Infeion 11C60S5, 600v, 11A M.O.S.F.E.T. The filter capacitors are two 150uF 250v units in series with discharge resistors. At the cold end there is a 1 watt, 100 ohm resistor to limit the inrush. When the boost converter starts, rectified A.C. from an overwind on the boost inductor switches on a Fairchild 12N60C3, 600v, 24A I.G.B.T. to short out the resistor.
All this guf drives the oscillator, a totem-pole, steering transformer driven design using two more 12N60C3s. The design is
very similar to that used in a myriad of cheap
Like in a compact fluoro, the oscillator is started by a diac connected to the gate of the lower 12N60C3. P.N.P. transistors fitted between the steering transformer secondries and 12C60C3 gates chop off the negative going excursions from the secondries.
An SCR, also hanging from the gate of the lower 12C60C3 to deck is triggered if the lamp is broken or removed and the series LC circuit on the output approaches resonance. (Note that a series LC circuit at resonance is a dead short and the voltage on the node between the L and C becomes excessively great, the loading of the lamp across the C element maintains a low "Q" and prevents it running away). Some of the H.F. from the output terminal is "sniffed" off between the two series connected ceramic caps in the feedback path and clipped by two series, back-to-back 33v zeners loading a 470 ohm resistor. The voltage across this 470 ohm resistor is capacitively coupled to a "radio-ham's" rectifier (one inverse diode to deck, one foward diode out to the load), by a 1nF ceramic cap. This is fed through 500K of resistors to charge up 1.2uF of capacitors. The voltage on these caps triggers a second diac, (20v) going into the gate of the SCR, so the response of the SCR is quite slow.
A 1 megohm resistor extends from the cold end of the 150nF output coupling capacitor up to the 350v D.C. rail. The small D.C. current from this goes straight through the litz choke and the lamp primaries maintaining no DC component at the output. The gate of another MOSFET, a small 1A drain current one, is coupled to this same node by 500K of series resistance and another "ham" rectifier, but not capacitively coupled. Two 1n2 capacitors, before and after the "ham" rectifier act as hash filters. The square wave output is rectified by this "ham" rectifier and turns the MOSFET on, its drain is connected to the oscillator starting diac and its RC circuit, discharging C and preventing the diac firing after the oscillator starts. If the lamp is removed the windings driving it go open circuit and the current from the 1Meg resistor is now free to keep that MOSFET on, preventing the oscillator from being started after the SCR has killed it.
Considering that the Endura lamp itself is designed to run fo between 5 and 800,000 hours, seeing this gear is an eye-opener as to an electrical engineer's challenge to design a circuit to last this long. The inrush 100 ohm being shorted by an IGBT is novel and reduces stress on all upwind components exposed to this surge. The bridge rectifier diodes are quite solid for a 150 watt unit and a single 275v varistor has been retrofitted under one of the line filter bucking inductors. Thw oscillator transistors are really solid overkill, 600v Vcgo, 24A collector current, when compared to the MJE13007s often seen in larger compact fluoros, (3A, 500v). All the housekeeping stuff, extra small MOSFET and SCR, is something that would definitely mot be considered in a compact fluoro.
The thing did have a weakness though, the RoHS, lead-free solder used. This stuff is almost pure Tin, with a little copper dissolved in it. It was noticeably "crystalline", sort of in the way molten Bismuth freezes. In the centres of vias one could see the crystal grains and flexing the board bought fourth a lit of little "creaks and squeaks" so common of metals like indium, tin and bismuth. I swear this is why it had failed, something had come adrift when the solder cracked! Note that tin has two common allotropes, white, or metallic tin, stable above 16.5*C, and grey tin, a grey, less metallic form which slowly forms if tin is kept at below 16.5*C.
I have seen shots taken in Mawson's Hut, in Antarctica. The food cans are still stacked there 70 years later, but they are all grey and dusty looking from the tin plating being so cold for so long! I'm sure the small copper content alters this process, bit the RoHS solder has a long way to go before it attains the reliability of the old 60-40, Tin-Lead eutectic grade.
I have taken photos, created the circuit and written a prasie of how it works, so can somome tell me how to upload them so everybody can see, in detail, how it works.
I have also got some electronic gear to run a HQI-TS 70W, but I have not drawn its circuit yet.
P.S. Anyone want to see photographs, at close range of an XBO-4000 (200A D.C. at 20v arc drop), super high pressure xenon lamp in operation!?