Interesting. I didn't even thought about that. Finally the writers on article about lighting health concerns, writes good article with health issue that can be proved by some means (Except mentioning the cancer, which is one of the most fearing Illness, as it is incurable, and there are only few proved causes to it occurance [Smoking]).

Another obvious danjer is if one malfunctions and pours out failed capacitor smoke.

To be fair, the "type A" capacitors used with magnetic HID and Fluorescent let out smoke and catch fire very well.....

Some HIDs don't need caps to function properly and there are always preheat ballast chokes. Plus the fluorescent caps are in a metal ballast case so that will extinguish the flame pretty quick…   Plus, LEDs are known for their suckiness here on LG.

Over here (240V) pretty much none of the low wattage HIDs (up to 400W at least) dont need the capacitor, and none of the fluorescent except some types of 8ft dont need it either

However, all magnetic ballasted lighting without capacitor is NPF. When you put more then a few NPF lights on a circuit, you have a problem : The current draw becomes significant, and with the more lights you add, you can easy reach or exceed the current capacity of the circuit....

So the capacitor is added straight between the line Hot and Neutral in every lantern, to correct the power factor and take only the current that is really needed

I'll have to check, but I'm pretty sure that my HPS 70w security light doesn't have a cap. I may add one to raise the power factor. And with the caps for the 8ft fluorescents, I saw something on here where the cap was used to raise the 80w ballast capacity to 125 to provide more of an inductive kick. Is what I'm saying correct?

It has nothing to do with the kick, but with the headroom for lamp arc voltage. The British 100W and 125W 8ft lamps use the same ballast, but the 125 have higher arc voltage - one that is marginally too high to work with a choke on 240V. The capacitor in series allow to push the limit a bit more and still use a choke

Ok I understand now, thank you. 

LED's do not need complicated ballasts, some companies offer a transformer/inductor with just a rectifier.

I took apart an LED lamp from eBay and it had a transformer as small as one you'd see in an iPhone charjer. It was a decent lamp, too bad it died.

I took apart an LED lamp from eBay and it had a transformer as small as one you'd see in an iPhone charjer. It was a decent lamp, too bad it died.

The isolated LED ballasts are essentially the same circuits as the phone chargers, even the wattage is about the same (3..10W), only the voltage uses to be higher (8..12V for the "3W" or "5W", 11..16V for the "4W" or some "8W",...). The control is a bit different, because the main operating mode for the LED's is in the constant current regime, reaching the voltage limit means an "open load" fault and with many designs it shuts down the ballast for some second or till the input power is removed (the chargers operate in both constant current during fast charge and then constant voltage for the top off part).
And many LED's are even simplified: No input filter (mainly for the dimmer compatible versions), less of filtering on the secondary (no wiring acting as an antenna to radiate the residual HF), just an inductor instead of the transformer (if the isolation from the mains is not necessary, when the LED's are covered), etc.

Many really simple low power (up to 3W) designs use just series capacitor and a bridge rectifier (plus some kind of filter capacitor parallel to the LED's, but that is not that much effective to supress any current ripple) feeding a series string of the LED's, for those low power levels it was the cheapest way to power the LED's. But even that range gets replaced by the DCDC converter based concepts, with the competition of the modern IC's the "simple circuit" becomes too expensive due to the rather high cost and large size of the ballasting capacitor (e.g. with the "filament LED" designs over 2W you would not be able to fit the required series capacitor just into the socket space...).

Noth only that, the advantage of the isoalted designs is safety, as with them the exposed SMD LEDs on the faces of the lamp (as most Ebay LED lamps are) are not live at mains voltage

But i managed to blow 2 of those in a row at a friend's home, within seconds of each otehr....

He was bragging about how cheap and good the Ebay LED lamps are..... I decided to check what will happen if i hold the switch of the lamp in the center position, so it is arcing in series with the lamp. After 2 seconds of arcing the input rectifier n the LED lamp blew up, emitting a puff of smoke from a hole in the front of the lamp

Then he was like "doesnt matter i got 5 of them for the proce of one". Unplugged the desk lamp, screwed another one in, plugged it back in. The switch was left in the arcing position..... The new one blew up with a puff of smoke after 2 seconds as well

The one I got was a color changing spotlight of I think four watts. It was like $15. To be fair, I kind of cooked it on an IS ballast once it fialed.

@Ash, I think your friend has a case of type 1 LEDisease. Symptoms include:
Bragging about LED
Defending the failure of them
Declaring how cheap they are
Being a bit snobby about the failure. Ex: doesn't matter, I got 5 for the price of 1, etc.)

Noth only that, the advantage of the isoalted designs is safety, as with them the exposed SMD LEDs on the faces of the lamp (as most Ebay LED lamps are) are not live at mains voltage

It is of course necessary when the LED's are accessible, but in such designs is usually quite a lot of space.

But if you take the design of the "filament" LED's, there is does not make much sense (the incandescents do not have any either and yet are not assumed as dangerous in that respect).
The thing is, the nonisolated buck allows to make the design way more robust and way less demanding on the individual components - it allows the input voltage to really reach above 750V without causing any overstress yet (so quite sufficient room for a VDR to really do it's job); with flyback and all isolated designs that limit is barely 500V, what is way too low for a 230V mains (the VDR's rated for 230VAC at the full rated pulse current could reach usually 700V peak voltage, so does not protect the isolated convertor with the most common 750V rated switching elements).

Plus it does not suffer from safety issues like some part usually being isolated, so treated like that (= touched carelessly, because "it is isolated") becomes alive when some fault happens. When something really looks like alive (the internals of a filament LED look the same as the ordinary bulb even when smashed, so make people cautious upfront)...
This psychological safety aspect I see as extremely strong (observing how people react on faults), while still completely neglected by all standards (it goes slightly against the direction of "cover everything live")...