Repairing CFLs 101!

[Vince]

As you may know, CFLs are usually not the highest quality electronic goods around. While a majority of CFLs burn for their entire rated life and burn out without any problem, a good, non-neglectable part of them die prematurely because one single component in the ballast fails. In many cases it's always the very same component, and this thread will show you how to find that failed component!

All you need is some electronics basics, a soldering gun, a desoldering pump, some solder and a multimeter (with capacitance setting if possible).

There are tons of CFL diagrams on the net, and I have the diagram for the GE helical 26W model if someone is interested to have it. The principle is always the same, few components change from a model to another. It is always based on a HF push-pull 2 transistors oscillator.



The diagram shown here is for a 230V CFL, but 120V circuits have few differences.

Here's how I proceed from the harvest at the recycling centre to free, working lamps:

- First, look for CFLs with non-blackened ends, this may be a sign of a bad ballast that can be easily repaired. Feel free take a couple of burnt out CFLs to get you a good supply of spare components.
- Open the CFL with a flat screwdriver. Be careful during this step! The base may be glued, if it doesn't seem to open at all, forget it. But most CFLs from major brands are just clipped and can easily be opened.
- Then take your multimeter to the Ohm setting and first check the fuse. If it's open, the bridge rectifier of filter capacitors may be shorted. In that case check both RF and voltage regulating caps, if one is shorted, replace it.
- Check the lamp electrodes. No need to troubleshoot any further if one electrode is open. Although you may want to check the ballast with a working tube if you have one in hand.
- If both the fuse and electrodes are good, set your multimeter to the capacitance setting and check the capacitor between lamp electrodes. This is the cap that receives the highest voltage peak at startup, and usually the one that fails. Your multimeter should normally find a value anywhere from 1 to 4nF. If it doesn't move, the cap is shorted. Replace it and in most cases the CFL will then work fine!
- If that cap is good, check the cap in series with the lamp (C6), that cap being shorted may only cause the CFL to flicker and won't usually prevent it from starting, although it may be open, which is rare but possible.
- In the case the two previous caps are good, check all the other caps in the circuit. In the diagram above, If C1 or C2 is shorted, it may prevent the oscillator circuit from starting. Otherwise the problem could be related to the HF transformer of the other oscillator control components, these are harder to troubleshoot but are fortunately rare. Such problems may cause the transistors to blow, or not to start at all.

Most prematurely failed CFLs are fortunately easy to repair, and with a little work, like 20 minutes at the most per CFL, you may get a virtually unlimited supply of CFLs! The only downside is the kinda unpredictable lifespan of the repaired lamps. But they're free anyway LOL.

[Form109]

Very Nice Vince!

[Medved]

The only failing components I've seen:
Result of lamp EOL and/or dry joint: Fuse, transistors with R1, R2, R3, R5, sometimes C3
C4 EOL: fuse, sometimes lamp and consequently the above.
If only the lamp filaments glow white (discharge is present), on ly C3 fail by itself and it's replacement would fix the problem.
If lamp filaments only glow yellow (only incandescent glow, discharge not present), again the C3 is failed, but due to lamp EOL. Sometimes the lamp is damaged as consequence of above, but such damage is rare, as most likely the filament would physically break.

Generally repairing the ballast make sense, when you want to use it with fresh lamps, as most of the faults are the consequence of the lamp loosing the emission electrode coat, so even if the lamp appear good, mostly it will fail very soon again.
Whole circuit is working well as ballast for PL-S, even for 2-pin ones (function of C3 is then taken by the capacitor connected parallel to the starter).
But here I would recommend to reconnect the cathodes of D2 and D3 to the base of the respective transistor, as their top killer is the excessive EB (reverse) voltage during the ignition/cold cathode phase. Such reconnection would clamp this voltage and transistors become way more robust in surviving lamp EOL.

[GEsoftwhite100watts]

can you send me the diagram for the GE helical 26w? By the way, are all GE helical wattages similar?

[SeanB~1]

The electronics inside a CFL are pretty much all the same, as you can not really put much into the space. There would be a difference between the common lamp for 230V and one for 115V operation, and a different one for a 3 way lamp. The basic system will be the same, just differences in the input side.  There are IC's for lamps, but they are not often seen in CFL bodies, as they cost more than the 2 transistor system, and would need the 2 transistors in most cases anyway.  They will be seen in dimmable lamps, as this needs a more complex circuit.

Your 26W lamp will be basically the same as the illustrated in many respects, just a different inductor, and most likely nothing else.

[Medved]

There are little variants in the output stage too, mainly when dealing with medium wattage tubes on the 120V mains as an attempt to go around the problematic doubler on the input (the weak output compromise the filament fusing functionality as an EOL protection).

The 3-way lamps are technically dimmable ballasts, so you need a bit more complex controller then the simple 2-transistor power oscillator.

[Vince]

can you send me the diagram for the GE helical 26w? By the way, are all GE helical wattages similar?

You can download the PDF document right here!

Just make sure you are familiar with electronics and diagrams before getting started!  

========

Now more details about tests of the most commonly failed components:

Capacitors

Capacitors are probably the #1 most common cause of ballast failure. Two capacitor types are widely used in CFLs, the electrolytic and the polyester film.

Electrolytic caps are known for their large values, mostly in the uF range. In CFLs they are used to filter the rectified alternative current to get a DC current with a higher mean voltage, which will be more efficient. In other words instead of "bumps", the current form will look somewhat like blunt sawtooth. Those filtering caps are exposed to a peak voltage varying from 170V to over 300V. Despite being kind of low, the cap is exposed to more or less intense heat, conditions in which they don't age too well.

Most capacitors of this kind will fail shorted. It is easy to check that with an ohmmeter. A capacitance meter is even better, since a shorted cap gives 0uF.

Electrolytic film capacitors can be made with very high breakdown voltages and are usually in the nF range. Don't be surprised to see 1.5kV rated caps in a CFL ballast! But even if they are rated that high, capacitors and high voltages don't make the best combination. The capacitor connected in series with the tube filaments (C3 in the diagram above) is exposed to hundreds of volts each time you turn on the CFL. If that capacitor ever shorts, the tube will never start.

Here again you can check that type of high voltage caps (removed from the board please! D: ) with a capacitance meter, or an ohmmeter if you have nothing else. A reading of 0nF of 0 Ohm indicates a shorted capacitor.

[Medved]

The capa-meter reading of the shorted out cap strongly depend, how the meter exactly work. Normally it should read "overflow" (as infinite capacitance yield zero impedance), but many instruments go out of correct operating point, if the terminals are DC shorted and then read the zero.

For electrolytic is very useful the ESR meter. It is not as common tool, but the only one able to show you the soon-to-fail electrolytic capacitor (the failed seal, so water deficiency in the electrolyte, demonstrate itself by rising ESR, while the capacitance is still unchanged).


@Vince: In the 4th section you have a typo: The word "electrolytic" should not be there...

[SeanB~1]

ESR meter is more important than capacitance meter on high value electrolytics, as many will show high ESR well before any signifigant capacitance loss. On high voltage low value electrolytics you need to measure both, as the impedance is becoming an important part of the measurement.

On motor capacitors and PFC capacitors I measure both capacitance and insulation resistance, as these capacitors tend to suffer from breaking down under running conditions.