Some questions on hacking CFL bulbs.

[Evil Lurker]

Ok so I am curious to know more about how these things work, and while I basically understand that the majority of the cheap bulbs sold in stores operate on self oscillating resonant half bridges and I'm a bit clueless as to how the values for specific components are determined, specifically the small bobbin wound choke/inductor on the primary side. I'm a bit short on time at the moment and about to have to leave the house so I can't go very deep into the finer details of the circuits but for the moment I'd like to know what the purpose of the inductor on the primary is and how it effects the circuit as a whole. Typically the values I have seen are 200uH to 1mH, they are quite small being around the diameter of the TO-92 13003 switching transistors, and are connected in between neutral and one half of the 1n4007 diodes arranged as a bridge rectifier. I've also seen  much larger type 26 iron powder toroid inductors placed in more or less the same location on offline SMPS (not common mode chokes wound on toroid cores).

Any insight would be helpful.

[mariush]

Start by reading this: http://www.pavouk.org/hw/lamp/en_index.html

[Vincent]

Yup, this website has lots of informations about CFLs.

Philips Semiconductors released a whole series of application notes for CFL design at some point, I would take a look at them as well. They usually produce excellent application notes from my experience. In my case at least they helped me a lot understanding the workings of those CFL electronic ballasts.

[vikasbly44]

NOTE: This message has been deleted by the forum moderator Simon for being against the forum rules and/or at the discretion of the moderator as being in the best interests of the forum community and the nature of the thread.
If you believe this to be in error, please contact the moderator involved.
An optional additional explanation is:

[Evil Lurker]

Well I checked out a couple of those application notes about the self oscillating resonant half bridge designs and just as I suspected the math involved makes me think the greek alphabet exploded in an algebra equation.

 Anyways for anyone wondering I was looking to make some standalone drivers for the occasional tv back light test, a UV light source for making my own PCBs, and who knows maybe even for some DIY fixtures using components salvaged from off the shelf bulbs. Technically it is possible to remove the ballasts themselves from bulbs fairly easily and use them as-is but they are a bit impracticable to mount and the PCBs are every bit as fragile as cheap Chinese bakelite perfboard when it comes to lifting traces. In fact I think the biggest flaw with CFLs is in the way the ballasts are designed into the base of the bulbs. Essentially you have two heated filaments and two switching transistors contained in what is essentially a non-ventilated plastic shell which are absolutely hellish conditions for the main electrolytic capacitor, typically some Chinese off-brand which tend not to do very well with heat to begin with. I have taken working bulbs apart that have had lot of hours on them and the insides were not pretty... lots of signs of heat degradation of the plastics, including the non-polarized caps, and even blackening/melting of the insulation of the lead wires from the base of the bulb to the PCB. I have literally lost track of the number of prematurely failed bulbs where the ballast kicks the bucket long before the tube has reached the point where the lamp itself still works but the lumen output has dropped to the point where it should be replaced. And for what it is worth, I don't think I have seen maybe one or two instances where the fusible resistor actually opened up. The most ironic thing about CFLs though is that the same plastic shell that cooks the bulb to death is the same thing that keeps CFLs from being such a fire hazard when they die in the first place.

Anyhoos, for what I am looking to do it seemed to me that the best way to get around the limitations was to design my own PCB and simply transfer over the components from an off the shelf CFL. There isn't any real need for me to completely reinvent the wheel, simply optimize the basic circuit for my own purposes while eliminating the shortcomings of designs that are already on the market. The basic plan is to make a "universal" PCB capable of accommodating different lead spacings of various components such as the film capacitors and the large inductor which would allow for output scaling. That there is more or less why I'm looking to figure out how component values affect the circuit as a whole. In addition I'm looking to eliminate the 4 diodes arranged as a bridge rectifier and replace them with a single discreet component, put in an actual fuse, and maybe some surge protection along with some socket connectors for lead wires. Later on down the road I may add on some additional things like a triac based light or dark sensor.

One other idea that I have had is to come up with a fault sensing circuit that provides some warning that the ballast or lamp may be about to fail. I'm thinking that perhaps some sort of reactive capacitance powered op-amp or comparator could be used to sense the health of the circuit. If I'm thinking correctly if say a capacitor was beginning to die the resonant frequency of the ballast would change and should the switching frequency go too far from some nominal value it would trigger an LED on the side of the ballast fixture letting me know that something is up. Of course from a practical commercial standpoint such a thing would probably be worthless but it would be nice to know when something is wrong to save having to acquire some new switching transistors should the end up letting out the magic smoke. Heck if I go that far I might as well go all out and put in a microcontroller, temperature sensing circuit, and hour meter to let me know when it is time to replace the lamp.

Anyways, enough rambling for now, I have a VFD to blow up.

[SeanB]

If you want to do it just use real power transistors instead of the TO92 ones they use, a TO220 package will do fine as the heat is low enough for the package alone to dissipate it. There are controllers for CFL use that actually incorporate all you want, but they cost $1 each in bulk, so are not used. If you look at Linear Technology you will find a lot of CFL drivers and appnotes which are very good reference designs, and which also have info on component selection as well. Using 4 diodes instead of a bridge is not really any issue, so long as you run them within thermal limits and current ratings. A fuse and a thermal cutout will be fine as well, along with some inrush limiting and probably better capacitors with slightly higher value, so they last longer.