I am thinking and have been at it to build a fully dimmable fluorescent ballast system that uses a conventional dimmer and existing magnetic ballast. The only extra wiring is one active live from the supply terminal of the dimmer the rest is wired through the system's input/output which is basically Active Live, From Dimmer, Ballast, Neutral and 4 outpus to the fluorescent lamp. The problem would be the size of the unit to fit into a batten fitting or concealed lighting. But I think it would be more viable if made for PL down lights to be dimmable by one wall dimmer since everything is above ceiling. Let me know what you think...

With most newer lamps (all, except for circlines and small 4..8W T5) you would have troubles with ignition: With the dimmer you can not use the starter (the tube would have unusually high  reignition spikes, what would trigger the starter), so you need the RS circuit to be able to ignite the lamps. And on 230V mains the OCV of the RS circuit would be only 230V, so not enough for most lamps (except those listed above).
You may get a bit help from the small capacitor parallel to the lamp (no more then 47nF, otherwise it would damage the lamp, rate it to at least 1kV; at 50% dimmer setting it would generate pulses up to twice the mains peak voltage), but I doubt it would be still enough for PL...

What if you add PTC in series with te starter (with the glowbottle only, not the capacitor), or use a thermal relay instead of the starter (similar to PerfektStart or 4pin starters) ?

Reason I posted this is that I think noone has made an affordable dimmer for residential use this dimmer if sucessfull would cost me less than 30 to build I'd say, maybe it's cheaper in the US to buy one? But is it because dimming of flourescents is not big in residential installations I think  but Anyways...The dimmer is connected to an 'optical variable resistor'. This variable resistor controls a dc voltage circuit which will run a dc ballast and a low voltage sensor circuit will controll the relay to activate the filament heaters and dc ballast at a set point when the circuit is 'dimmed'. So in theory...the tube will still light even when the dimmer is at lowest level because the dc ballast is high tension and tube life will not be reduced very much because there is filament heating provided. It will also be able to run at full 50Hz current without a starter because of the high tension provided earlier by the dc ballast. But I haven't done anything yet it's an idea based on some circuits I have... Bigger problem would be to fit it in a single tube batten fitting for a 36w one it would barely fit because it would need one halogen transfomer for dc power.

What if you add PTC in series with te starter (with the glowbottle only, not the capacitor), or use a thermal relay instead of the starter (similar to PerfektStart or 4pin starters) ?

There would be not enough current to keep the thermal relay open.
The starter does not need much current to trigger, so the PTC would stay cold, but the lamp wold be flashing. The missing capacitor would not help, it would make things only worse (it generate the voltage kick, what then aid the reignition)

The only thing what may help, would be an autotransformer boosting the OCV up. But then you would need special ballasting coil designed for such elevated voltage.


I think better idea would be to connect two 1200V IGBT's back-to-back with emitters and gates together, collector of each of them connected instead of the starter, 1kV diode parallel to each transistor (to provide reverse current path) and steer the gate from a "555" circuit with variable duty ratio at about 1kHz. Connect diodes paralel to filaments, anode to the starter side. Varying the duty from 0 to 100% you would vary the time the ballast current flow trough the switch (0..100%), so trough the arc (100..0%), so vary the brightness. This circuit would provide the extra heating necessary in dimmed state...
The the lamp have to be started with the dimming setting slowly from minimum (= transistors 100% ON) and slow sweep to maximum (o% transistors ON) in order to preheat electrodes and create pulses for the first ignition.
The 1kHz is used to avoid the need to synchronize with the mains


@imj: I think then such creation would be way more complex, way less efficient and way less reliable then "regular" HF dimming ballast... The simplest specialized dimming ballast controller IC's do not cost more then 3Euros, the rest of specialized components (choke, capacitors,...) you may extract from any Cheepee electronic ballasts for given and/or similar lamp.
Moreover the electronic dimming ballast could be designed so, it is able to extract the power from the "dimmed" line (an active PFC circuit does the job well till very low dimmer setting), the overall complexity would not be worse...

Those HF also don;t last very long I worked with them before and they have RC. They start to flicker and can nol onger be dimmed. I know about the chip but I do not have access to it but even so Y is there no demand for such a system for residential use because maybe they changed it but there needs to be dedicated wiring for dimmable ballast like the 1 to 10v? There may be dimmable CFLs in the US but so far the only dimmable CFLs I've seen are from Philips and they cost $101 for a pair with RC dimmer. Or maybe people are too silly to wire one up..XD.

@imj: What did you mean by the "RC"?
I have quite good experience with IRS2530, it s available trough e.g. Farnell - it form simple, yet closed loop control system, well work from ~15%, after some fine-tuning (first step would be using elevated DC bus voltage, or low arc voltage lamp) it should work well below the advertized 10%.

With fluorescent you should always provide separately enough power for the system and separately the required dimming level information, that's why the 1..10V systems are so popular - they are simple and cheap.
But it is possible to extract both from the common line from the phase-cut dimmer: The active PFC is able to bost the voltage up to the required 400V DC on the bus trough whole dimming range. The control voltage signal you then extract by sensing the pulse width of the voltage after the input rectifier, so you simply clamp it (two resistors and a diode towards the local 15V supply line - so when the triac is ON, there would be 15V, otherwise 0) and then using simple low pass filter (I would recommend 2'nd order to have fast response, but low output ripple) convert it to the DC voltage necessary to steer the ballast controller. Good thing is to add an offset to this signal, so the DIM signal shut the lamp down earlier, then the remaining small pulses from the dimmer would not suffice to power the whole ballast.
To avoid triac switching problems, the capacitances on the mains input should be limited and yoou should use damping of the input filter (C of the same value as other filter capacitances in the circuit in series with the R=sqrt(Lemcinductor/Cemccapacitors)).
So compare to 1..10V ballast the phase-cut controlled version cost about 5 components extra, while the control signal isolation circuits could be omitted, so bottom-line the whole ballast complexity remain the same...

RC I meant remote controlled dimming with a transmitter unit. Sounds like alot of knowledge of electronics is required which I am not trained but trained in electrical... And even if one attempts to make one using the dimming IC there's still the inverter transformer to run the tube that would be commercial standard.

RC I meant remote controlled dimming with a transmitter unit. Sounds like alot of knowledge of electronics is required which I am not trained but trained in electrical... And even if one attempts to make one using the dimming IC there's still the inverter transformer to run the tube that would be commercial standard.

On www.irf.com webpage you may find "ballast designer" software, that generate all required components of the basic ballast circuit, include the design of the chokes (both ballast, as well as the PFC).
Chokes I usually calculate myself in order to utilize components (bobbin and the core) from salvaged CFL's...

That ballast designer software looks neat! I just downloaded it and will play with it when I get a chance.

Yeah but I'm refering to residential lighting..noone has made any for that except dimmable CFLs which are very rare and difficult to dim. You can't expect a layman to go buy those parts right? Anyways I will go ahead with my simpler design and post it...

Yeah but I'm refering to residential lighting..noone has made any for that except dimmable CFLs which are very rare and difficult to dim. You can't expect a layman to go buy those parts right? Anyways I will go ahead with my simpler design and post it...

Present systems designed for incandescents already include E27 sockets, so CFL format is the only non-incandescent option.
And if somebody build new installation basef on fluorescents, it mean new wiring anyway, so it is of no problem to lay the control wires together with the power ones, so there is no need to deal with troubles related to phase-cut dimmers (cost, disturbance, reliability, dimming range, dimming setting synchronization between multiple fixtures even at low levels,...), as e.g. "1..10V" system could be used.

Cases, when someone want to use present wiring, but upgrade to fluorescent fixture are quite rare, what mean no maker was interested to design special ballasts for it - they would not be profitable.
So what emerge were sets, what generate the 1..10V signal using some sort of wireless control. As these do not deal with any power, they are very simple, universal for all light sources (as "1..10V" dimmers are available for all of them), so may be made in reasonable volume for reasonable prices.

And there is one exception: Megaman's "DoRS" (they sell not only CFL's, but "mature" fluorescent ballasts with that system as well), that use power cycling by ordinary switch to select between few (usually four) intensity levels. Bad is, then there is no other means to resynchronize multiple ballasts on the same switch, then keep hem OFF for longer then the time-out period, so such multiple ballasts connection is rather problematic.

So if you want a bit more comfort (e.g. not to cycle trough all levels) and do not want the wireless (easy to be disturbed), you have to make some custom ballast system (e.g. something like this ). By the way I'm convinced, then any 1..10V ballast would be possible to convert to the "polarity" controlled 3-way one...