And speaking of electronics, i think an end must be put to self ballasted CFLs where you have to throw away the entire lamp instead of replacing just the tube

They are throw-away, as the electronic lifetime is due to operating conditions about the same (or sometimes even shorter) then the life of the tube, so dead device mean both parts are nearly spent...

If you make the adapter to last longer, it will last longer

I have magnetic PL adapters that i useinstead of CFLs, i bet they will last "forever"

I have some that have been on essentially 24/7/365 for the last 15 years or so. A little brown on the top, but otherwise they are fine. I replaced them with Phillips Tornado 20W lamps, so far they have done over a year of operation running base down.

@SeanB: The lifetime strongly depend on the temperature. And the temperature depend (beside other aspects) on the burning position, while the base down is the coldest running one, I would guess the only one, when the ballast have a chance to last longer then multiple tubes. But when used base up, you would be happy, then the electronic endure something comparable to the tube (regardless of the category, as cheap, inferior tubes are usually mated with cheap, inferior ballast components).

@Ash: In the base up you won't go above ~10..15khours with the best components available with the reasonable ballast size. It run simply in too hot area...

Add thermal separation between the tube and ballast to reduce heat transfer

substitute the electrolytic capacitor with polyester + boost converter to use higher duty cycle of the AC

Add thermal separation between the tube and ballast to reduce heat transfer

Quite difficult, when the hot air, rising from around the tube, is all around the ballast...

substitute the electrolytic capacitor with polyester

The size would be huge and heavy, so the PCB would be even less reliable, supporting all the mass. And there woulsd be even less air circulation inside the ballast...
And it is not only the capacitor, but the PCB as well - the solder (mainly the lead free one) and mainly intermetallic compounds (formed on the border between solder and the base material) are weak and tend to develop metal fatigue cracks, what is accelerated by the high temperature and thermal cycling...

boost converter to use higher duty cycle of the AC

This is exactly, what an active PFC does... But result is more complex circuit, so more tight assembly and more parts and joints to fail...

Quite difficult, when the hot air, rising from around the tube, is all around the ballast...

Add an empty tube of transparent plastic (to not block light) that goes down near the CFL tubes and ends lower then them, so that this tube acts as chimney to suck cool air from below the CFL into the ballast

Separate the ballast enclosure to 2 parts, one with only the lamp socket (which get hot), which is fully enclosed and suspended few mm below another part with the ballast, so that there is free air between them

The size would be huge and heavy, so the PCB would be even less reliable, supporting all the mass. And there woulsd be even less air circulation inside the ballast...
And it is not only the capacitor, but the PCB as well - the solder (mainly the lead free one) and mainly intermetallic compounds (formed on the border between solder and the base material) are weak and tend to develop metal fatigue cracks, what is accelerated by the high temperature and thermal cycling...

With the PFC used, you can use capacitor of smaller value. And you can connect the capacitor alone thorough fly-wires (long leads)

Arrange with the most sensitive components around the cooling tube. If you dont eliminate the capacitor, stick it right into the cooling tube to get the coolest air possible

Leave longer leads to the components and bend them down, so the leads act as springs

Use copper rivets (as used in CRT monitors for heavy components, tube base etc) to improve soldered joints

Cut the PCB to make thin "bridges" to components, so that whatever little flexibility is in the PCB itself can help

Connect the larger components with Wrep

Separate the ballast enclosure to 2 parts, one with only the lamp socket (which get hot), which is fully enclosed and suspended few mm below another part with the ballast, so that there is free air between them

This is already done on many higher power lamps, it is only able to bring their life closer to the low wattage CFL's...

With the PFC used, you can use capacitor of smaller value. And you can connect the capacitor alone thorough fly-wires (long leads)

Of course, but still that would be too huge in the film form. To work without the PFC, the capacitor won't fit into most CFL bases at all...
Long leads are already used, in order to push out the capacitor to the thread area (the coldest place in the ballast). But still the life is rather limited...
And the life of film capacitors is limited at high temperature as well, so the improvement would not be as large. Moreover the electrolyte capas offer 100% self-recovery mechanism (the electrolysis really reconstruct the dielectric, not only separate the broken down section as film capacitors do), so they may be operated at the order of the actual breakdown (and in fact are; due to dielectric degradation they maintain the dielectric capable to withstand just the voltage applied on them during most of their operation). So all the elco's life problem is limited to the seal capable to hold the water and it's vapor inside the can. If this seal is working, the capacitor work well...
The film capacitors degrade by applied voltage on any temperature, so the only way is to have high margin til the breakdown, what ask for huge components...

Arrange with the most sensitive components around the cooling tube. If you dont eliminate the capacitor, stick it right into the cooling tube to get the coolest air possible

Such tube would work as a heat exchanger and have limited airflow...
By the way, it is sometimes used, but not much success...

Leave longer leads to the components and bend them down, so the leads act as springs
Use copper rivets (as used in CRT monitors for heavy components, tube base etc) to improve soldered joints

You have to fix the components first, what ask for them to be larger. And in that way they won't fit into the socket area...

Cut the PCB to make thin "bridges" to components, so that whatever little flexibility is in the PCB itself can help

That would limit the usable area, so the design won't fit into the base. And on the PCB you are limited by the formation of electromagnetic parasitics, what increase the stress on components, so make them less reliable...

Connect the larger components with Wrep

What is it? A copper foil?


Many techniques to reinforce the ballast mechanical structure are known, but all of them ask for larger space, so they are not suitable to fit into the small base of a lamp. The same is for heat management: There are many of known ways, but again all of them ask for space and/or mass.
Even the problematic electrolytic capacitor may be made more robust, but the seal would occupy way larger portion of the can, yielding it bigger...
All of that together is usable in the quite empty box of the CRT TV receiver, but not in the tight space of the selfballasted lamp base...

So with a bit larger and heavier base the lamp can be made, and used in places that allow for the larger size. Why not make it then ?

And how about magnetic ?

Then they won't fit...
If a special lantern have to be designed, what about using real decent fluorescent lamp and put the ballast aside from the heat?

I mean as retrofit for incandescent fixtures. If so large CFL is needed it means that the fixture was for high power and very hot incandescen in the 1st place, so there must be lots of space in it for the adapter

For some years Panasonic made in Japan the lamp you need!  It was a 4' T8 warm white tube sealed within a 4'T12 cool white tube.  The colour temperature of the light could be varied between daylight and tungsten colour by a fairly simple electronic controller.  I don't think they are produced any more though.

Also many years ago while visiting Philips Eindhoven I saw some special variable colour PL-L lamps in a trial public installation.  These contained two different phosphor types and could also be varied in colour temperature from 2500K to about 4000K with a special ballast, but they did not go into production.

These exotic ideas are however not necessary when you consider the much simpler option of just having two different colour temperature tubes in a single luminaire, equipped with two dimming ballasts, and where one lamp is dimmed while the other is increased in output.  I believe the famous German photographic lighting company Dedo Weigert Film is offering something like this using either the KinoFlo tubes or Sylvania CineLynx CFLs with colour points tuned to match photographic film stock.  In HID he certainly has similar solutions.

Going back to the original idea, the phosphor in any tube is exposed to the discharge direcly, I have thought of this to use modern tri-phosphors in mercury lamps, but it was pointed out that the light from a mercury lamp comes from the brightness of the discharge tube.
Using tri-phosphor wouldn`t result in a MV lamp with a tri-phosphor tube colour and brightness, just a poor colour, and a phosphor coat that would be distroyed with the intense heat.

Regardless of the feasibility of this device, we must think of all situations in which the UV emitter will be left bare, without fluorescent cover. I'm met broken mercury lamps a few times: the internal quartz bulb was running bare and glowing with a nice deep blue tint, and people were walking or chatting around them; I assume their eyes were damaged by the UV light.

However this concept of external fluorescent layer exists in LED lighting: blue led + yellow cover (while regular white leds have both blue die and yellow phosphor layer enclosed in the same component). For example Thorn makes this kind of ceiling lamp; I use the Philips AmbientLED bulb that has blue leds within a yellow bulb (when lit: blue led + pinkish phosphor = white).

The fixture can be made with an interlock that requires the cover tube to be installed. For example the cover tube end caps can be made into adapters that are required to make contact between the actual lamp and the fixture sockets

The fixture sockets and adapter endcaps can be made springy, so that if there is attempt to fit just the end caps (from a broken phosphor tube) and lamp into the fixture, the springs will force the end caps to be at the wrong distance from each other, and preent some contact (either between lamp and adapter or adapter and fixture) from reaching and making contact

And besides, instead of all this, each component should just come with a warning label in a visible place (not on the box but directly on the lamp, on the phosphor tube, and in the fixture in a place that'll be directly viewed when replacing the lamp). If the warnings are ignored, thats their problem