I recently wondered, after replacing my second one, how the internals of compact fluorescent (CFL) bulbs compare and contrast to an LED light bulb? I anticipated I’d find a plethora of passive analog and power components (along with a comparative dearth of digital components) versus their LCD counterparts but there’s only one way to find out, right?

My various LED bulb teardowns over the past several years have consistently been among my most popular. We’ve covered standard, Zigbee-controlled, Wi-Fi augmented, and Bluetooth-enhanced versions, so we have plenty of information for comparison.

Today’s victims are both 13W (60W incandescent-equivalent) T2 form factor CFL bulbs, with claimed 825 lumen output, “generic” (i.e., not brand-name), and in equally-generic packaging:

Here are some overview shots for the first one:

And here’s the second of the two CFL bulbs, accompanied by a 0.75″ (19.1 mm) diameter U.S. penny for size comparison purposes:

Here’s a series of closeups of the product markings:

And of the passive airflow vents at both ends:

Time to dive in. Although the two halves of the base are presumably pressed together, judging from the seam, their union is likely augmented by adhesive. The glass tube attached on the end makes twisting them apart even more complicated from a potential-injury perspective; I therefore initiated the process with a shallow hacksaw cut:

Subsequently inserting (and twisting) a wide flat-head screwdriver completed the deed:

In that last image you can see the two wires that connect the PCB to the electrical contacts in the base. They seemed to be sturdily soldered at both ends so I just snipped ‘em leaving the PCB topside (i.e. the “ballast”) inside fully exposed to view:

Here are three straight-on shots of various orientations:

Obvious constituent components (even to my binary IC-biased eyes) include the toroidal inductor and transformer, along with the discrete transistors and several different-sized capacitors and other inductors (what more/else of note do you notice, my more analog- and power-attuned readers?). Note, too, the two wire pairs, presumably headed to both ends of the coiled fluorescent tube. Let’s find out:

One of the four wires was pretty short, so I left it connected:

The PCB backside is pretty unremarkable, unless solder points and a mix of standard and bidirectional diodes are your thing:

Now let’s look at the second CFL bulb:

Here’s another series of shots of the ballast:

This time I got all four wires headed to the fluorescent tube snipped off (although I snipped the top off one tube end in the process):

That lead to a clearer straight-on shot of the PCB backside:

So, circling back to this writeup’s title, what caused these CLF bulbs to fail? The potential-cause list is long, and these bulbs have been sitting on my teardown pile for a while now, so my memory’s a bit fuzzy. But I don’t recall a pop, puff of smoke, or other evidence of ballast-component failure, nor do I see any bulging capacitors, singed circuitry, a discolored base, or the like. Instead, take a look back at the overview photos (specifically the blackened segments of the fluorescent tubes in the base’s proximity, and you’ll likely conclude (as I have) that the root cause in both cases is more traditional in nature: degradation of the electrode filaments.

With that, I’ll turn the fluorescent bulb’s illumination in your direction for your thoughts in the comments!

—Brian Dipert is Editor-in-Chief of the Embedded Vision Alliance, and a Senior Analyst at BDTI and Editor-in-Chief of InsideDSP, the company’s online newsletter.

Would you dare connect the PC board to 110V and O-Scope the leads to the flourescent tubes? This might prove the tubes failed. Hmmm Thanks for the tear down. Pete Rathfelder

I wasn’t very impressed by the wire wrap (WW) work connecting the lamps filaments to the PCB. Traditionally, WW uses silver plated wire and square pins with sharp corners. The goal is to form air-tight flex resistant connections between the wires and the pins. The corners of the pins are supposed to “cut though” the wires silver plating to the copper underneath. Making sure the 3-4 bare wire turns on the pin are touching is part of that. Some standards also wrap 1-2 turns of insulated wire before the bare wire is wrapped onto the pin for flex/vibration resistance. I saw very little of any of that… Some of those wires were badly corroded. I expect there’s really no effective cooling inside, so I’d look at the electrolytic (what’s the measured ESR?), and semiconductors (open/shorted diodes?). I also saw something marked as a fuse.

In the mid-90’s I worked for a small company that made electronic electronic ballasts (separate from the lamp). My comments on bulbs and ballasts are directed toward them as separate components, as I worked on ballast designs for 4-pin bulbs without internal starter or ballast components. I use the terms lamp and bulb inconsistently – they are interchangeable to consumers, but have specific meaning to a lighting person. My apologies for inconsistency.

So, I, too, feel driven to open CFL’s that fail prematurely. I must admit I have never been able to keep track of how long they operate until they fail and have encountered far fewer failed CFL’s that were ‘name-brand’, ‘Energy-Star certified’, etc. It’s too appealing to buy six (or more, at clearance prices!) generic bulbs for the cost of a premium one. I think I can only remember sequestering one ‘quality’ CFL for surgery and it is easy to lose track of how old they are. I started marking replacements with a pencil but wasn’t very organized about it.

Supposedly the term ‘ballast’ came from the analogy of a ship’s ballast stabilizing it’s orientation (top side up, anything that helped reduce seasickness). A fluorescent bulb operates as a gas discharge device, with a conductive plasma inside once the arc is struck. An inert gas (typically argon for linear tube bulbs, and a blend or argon & neon for smaller folded & spiral CFL) supports establishing and maintaining the arc more easily and a small amount of mercury provides a spectral line that is mainly in the UV wavelength range. The internal coating (typically called ‘phosphor’, but containing a variable mix of ‘rare earth’ elements) is excited by the UV energy to produce as-designed visible wavelengths (consider “Tri-phosphor” bulbs that produce a mix of three wavelengths, depending on the target color temperature, and the human brain accepts this as almost full-spectrum).

A resistor will work, and was used decades ago, as a ballast, but is terribly inefficient and also caused one end of the bulb to darken with usage more than the other end.

(Electro)magnetic ballasts in simplest form use an inductor in series with the bulb’s effective plasma resistance to form a voltage divider to result in the proper voltage, current and power in the bulb. That is why ballasts are labeled so specifically about which lamps they are intended for. 48″ T-8 and T-12 have significantly different operating voltages & currents despite both being fittable in some fixtures.

More sophisticated ‘magnetic ballasts often contained transformers rather than inductors, to better control the voltage on the bulb filaments, depending whether they were designed for pre-heat, rapid-start, instant-start, etc.

The ballast designer has to contend with a higher voltage required to ignite the arc, then running at a lower voltage to operate the bulb at an operating point conducive to long, or at least acceptable life span. A refrigerator bulb experiences many ‘starts’ and short duration ‘on’ periods’. Office lighting require(d) few starts and extended on time.

Simplest electronic ballasts used a self-resonant half-bridge inverter, with selection of winding inductance and series capacitance for each bulb (or narrow family) operating characteristics…the voltage would rise high enough to arc the lamp then the voltage would divide across the L, C & lamp effective resistance. The filaments were usually in series. Aging thru use resulted in end-of-life properties that are detected in newer bulb-ballast combinations where failure might be less graceful.

Fancier ballasts, including dimming ones, might control the inverter frequency instead of relying on fixed components.

Instant start, in the simplest form, simply shorts the two filament leads at each end of the bulb, and ignition is by brute-force electric field (similar to CCFL’s). In rapid and pre-heat circuits, the filament is heated, producing some space charge electron cloud effects that make the gas mixture easier to strike the arc across, compared to field emission.

The CFL with series filaments may be a hybrid of sorts, kind of between instant start and pre-heat…instant for start and where-things-fall voltage division for running.

CFL’s are recyclable (used to be disposable) but the least costly, and the ballast is replaced with the bulb. For cost reasons, the designs tend toward the primitive rather than sophisticated.

The bulbs are incredibly durable if operated correctly. Rated lifetime is usually based on a model of start, run time and temperature range assumptions. Conventional (non-CCFL) fluorescent failure is defined as the point at which 50% of a group of them will not light. (in contrast with LED lamp failure definition of lumen reduction to 70%). That assumes it’s time to replace them. No one changes the ballast characteristics for aged lamps, but in my experience the tired old bulbs usually haven’t physically failed…they just don’t light in the standard operating circuit. There’s an occasional filament fracture…from handling or manufacturing statistics. I’ve seen some linear tubes that failed after many (20+) years because they were not installed properly in hanging fixtures fro ma ladder and only one (of two) pin at one or both ends were in each socket. On those, the glass overheated near one side of a filament and a pinhole to the outside world resulted…exhausting the gas. By that point the socket contacts were also damaged from overheating and a replacement bulb was omitted from the fixture (ballast accommodated 3 instead of 4 bulbs).

Finally, back to the semi-forensic CFL investigations…

As is common with many applications, something else fails in circuits managing significant heat, voltage, etc.

I have never found an open (failed) CFL filament yet. I have found many failed electrolytic capacitors (visibly deformed or not), some found to be open-circuit or extremely low capacitance. Next most frequent has been overheated or punctured epoxy-dipped film capacitors. I assume the green ones in the pictured dis-assemblies are the polyester/MKT type, and the reddish-brown one is polypropylene/MKP in the high frequency (15-45 kHz) circuitry. I have occasionally found a failed open resistor in the ‘fuse’ location, probably because something else failed…I assume one of the TO-92 MJE13002 NPN transistors. I haven’t bothered to check those…I was initially shocked to discover them, having used TO-220 MOSFET’s in a comparatively larger external ballast.

Sometimes I find a mix of bright green and dark, almost black, green capacitors. When they are unevenly colored or there is a rupture in the epoxy coating, I assume those failed due to running out of self-healing abilities. I have never verified what dielectric is in each color capacitor…just assume based on convention. If they measure close to marked value.

I have seen MKT capacitors (attempted) in fixed 2-level dimming circuits. Kind of an eye-opener to witness catastrophic capacitor failure at 60 Hz due to dissipation factor/losses. AC ratings of capacitors are not a simple single number.

I have a bigger pile of CFL tubes removed from the assemblies to someday try them as voltage regulators (like neon bulbs). Just because. CFL’s are probably in the 35-50 V range, but unfortunately like to run at 140-180 mA, and are impractically bright…and hard to mount. No telling how poorly or well they work at low current…unless I try.

LED lamp failure experiences for me have all been in the IC’s that endure all the abuse of conversion down to LED voltage level and apparently insufficient heat dissipation. The LED’s (smaller statistical sample group as they are much harder to disassemble than CFL’s) have all worked, at least with a quick power supply check – I gave them to someone else who hadn’t lost interest in them.

Oh, last thought…on the wire-wrap observations, the old backplane wire-wrap technology I associate with comparatively low-voltage logic. Silver-plated copper wire wrapped around gold-plated brass pins for a gas-tight connection sounds good for low voltages. It is very reliable when done properly. IIRC, the turn of insulation around the post was a ‘modified wrap’? Maybe a vibration solution.

The wire coming out of a CFL bulb’s filament is probably some alloy that is TCOE-compatible with the needs of glass-metal seals…old technology dating back to the vacuum tube era…Kovar, FerNiCo, etc. It’s pretty stiff wire…trying to unwrap it to save the tubes for…re-purposing…I often break a lead right at the glass-metal seal…but they weren’t meant to be unwound and reused.

You must Sign in or Register to post a comment.