Cathodes in Compact Fluorescent Lamps

[klugesmith]

One barrier to recycling dead CFL bulbs is that they invite experiments.
Reverse engineering the HF drive?  Energizing the tube directly with NST etc.?

It looks like the coated-tungsten filaments are designed for electric preheating.  Is that only for 120-volt applications, and unnecessary in 240-volt land?
Typically the filaments at both ends of tube are connected together, in series with a capacitor.


Here's what one tube looks like with filaments directly in series, with DC power of 12 V, 0.35 A.


So how do they get powered (for preheat) in normal operation of the lamp? 
Easier to ask on forum than to trace the circuit or search on Internet.

[HiVi]

Preheating is always needed, not matter what voltage land you are in.
It is needed for mercury to vaporize so the right mixture of gasses is in tube.
Inductive ballast takes care of current regulation.
In startup operation, usually, there is thermal switch, that in series connects the filaments, so you heat them with resistive heat generation. When heated enough, the switch opens.
In normal operation, after preheating was done, the filaments are getting heated by electrons flowing though gas hitting the filament.

Hope I covered the basics of what you have asked.

[klugesmith]

Thanks, but it's not what I was asking.  Plenty of fluorescent lamp ballasts do no preheating (see Rapid Start and Instant Start).
In fact there are fixtures and lamps with only one terminal on each end, not to be confused with cold cathode systems.


Electric preheating 1) reduces the arc-striking voltage and 2) greatly reduces the wear on cathodes in the first second of operation, if they are forced to emit while cold (a major factor in lifetime of lamps).   Hg concentration builds up as the whole tube gets up to operating temperature, and normally reaches about 1 atom of mercury vapor per 1000 atoms of argon.

Was just curious about the preheat circuit details, in the cheap ballast circuits built into screw-base CFL units.

[klugesmith]

Had to play instead of throw away.   Lit up a few CFL tubes using very simple bench setup:
Wall plug - variac - MOT - curly lamp.  With AC voltmeter or ammeter, not at the same time.
This depends on the ballasting behavior of MOT because of its core shunts.


Starting from 0 on the variac, stepped up voltage reached 350 to 450 volts before the tube lit up.
With lit tube, the measured U immediately dropped to order of 100 volts.
Crept up in the course of a minute or two, as tube warmed up and got brighter.
Turning up variac, to increase current, made arc voltage lower (e.g. 85 V at 0.45 A). Soon the tube was much hotter than in normal operation.
Turning down variac, to reduce current, made arc voltage higher (e.g. 115 V at 0.08 A).

How about measuring the cathode filament temperature at different arc currents,
by DC resistance measurement at the same time we have AC current shared by the two leads at each end of tube?

[davekni]

I've noticed the negative-resistance of fluorescent tubes too, mostly ~20 years ago playing with DIY battery-powered fluorescent lights.  My low-pressure sodium vapor tube is also negative-resistance like that.  I think it's common for almost all gas-discharge tubes, from small neon bulbs to xenon arc lamps.

[T3sl4co1l]

The driver is a current mode self-excited oscillator, half bridge.  Typically a small toroid transformer has a few turns carrying the load current, and a few more turns for each transistor base.  This drives the transistors at a roughly fixed hFE, because the transformer sets the ratio of base current to load current.  Startup is provided by a DIAC, which is in turn powered by a cute little missing-pulse detector.

Load coupling is with a series resonant tank.  This draws high load current at resonance, which the oscillator tracks of course; how high depends on the Q factor, which in turn depends on the load resistance in series with it (or in parallel with one of the elements).  Specifically, there's an inductor in series with the inverter, then one filament, then the capacitor, then the other filament.

Hmm, I forget offhand how the load's ground return is wired, if there's a larger value coupling capacitor or it just goes to ground and the DC bias is fine somehow.  (Checking typical circuits, it appears there is a coupling capacitor.)

The advantage of this arrangement is that, when the tube is cold and deionized, current flows through the filaments, which are cold, so the Q is high.  This stresses the transistors and resonant capacitor, but doesn't last long.  (Indeed, because these capacitors are under so much cost pressure, they are typically specified in terms of capacitance versus number of start cycles -- they are intended to self-heal in normal startup conditions!)  The high resonant voltage also helps spark plasma in the tube.

As the filaments heat up, their resistance rises and the resonant voltage falls; as electron flow begins, sooner or later the plasma discharge ignites.  This severely reduces the Q (the tube acts in parallel with the capacitor), dropping power output to nominal levels and turning the circuit into more of a multivibrator (half bridge equivalent Royer oscillator perhaps) with a series inductance limiting current (so, it still has a ballast inductor, of sorts, but it serves dual purpose -- cool, huh?).

Probably the coupling capacitor serves some function as well, as the plasma has a low dynamic impedance so doesn't behave quite like a damping resistance, but also acts to short out the main resonant cap; the operating frequency likely drops from cold startup to operation, as this capacitance becomes significant.

Typical failures are the resonant cap dying, electrolytics dying of heat and age (especially in inverted vertical fixtures, especially especially in closed fixtures), filaments burning out, or transistors dying of various fault conditions (often brought on by other failing components).

Heh, personally I've had incredible luck with CFLs; I've purchased only a handful in the last decade, some of which are still in service.  Mind, I use most of the installed lights in this apartment only intermittently; my workbench is lit with LED fixtures instead (which are something of a special case, as I built them myself).  Most of the failures have been filament related.  LEDs weren't even an option last time I was shopping for bulbs, but they've utterly surpassed other lamps now.  Crazy how things advance, eh?

Tim

[klugesmith]

Thanks for the details about CFL circuits.

This seems like an OK thread, certainly in the right subforum, to carry on about some germicidal lamp observations.

In Covid Lockdown thread I posted a picture of a disinfecting robot being put into service at my workplace.
On closer inspection, the robot head has 8 lamps positioned around an 8-sided prism made of mirror-finish sheet metal.
Apparently T5 lamps inside clear plastic tubes, maybe for thermal insulation as well as protection. [edit] Maybe the outer tubes are glass or fused quartz. [\edit]

Each lamp is endowed with a pair of really big coiled cathodes.

and takes a few minutes to warm up to full brightness (visual brightness).   
Don't know if that's because of current being ramped up, or just increase in the partial pressure of Hg atoms.

I'm mystified by tubular lamps (even curly CFL's) that have different brightness in different places, as they warm up.
Would expect that diffusion rates in low pressure gas would make the Hg pressure practically constant in the whole length of tube,
matching the vapor pressure Hg at the coldest place in tube envelope.
As mentioned before, argon pressure has to be at least a few torr (mbar), and Hg pressure has to be below 0.01 torr (mbar).

[klugesmith]

Who else has been amused to see that curly CFL lamp tubes are configured as non-inductive bifilar windings?
Obviously to simplify the glass-bending process (with both electrodes ending up at same end of coil) instead of for electrical reasons.
I just measured one lamp.  Figured that a plain helix of same dimensions and turn spacing would be about 1/2 microhenry.
Was reminded of excellent and detailed circuit description by T3sl4co1l in this thread.

In another thread, following Alan's plasma toroids and low pressure sodium lamps, we saw answer to my question here about non-uniform brightness.   The partial pressure of metal atoms is extremely small. Nonuniformity can persist for a long time because diffusion along the arc tube is slow.