@WWHID
So how do you account for all of the LED lanterns with their diodes turning purple, they're not very robust are they.? 

I believe that the purplish color might be due to the fact that such LED diodes use phosphors and that as the phosphors degrade with age, more purple light gets emitted, but I don’t know for sure. I have known that some early LED lamps such as the Philips AmbientLED from 2011 have yellow colored phosphors to convert the light from blue LED diodes into a soft pleasant warm white glow.

They still failed though.

Certainly an indicator of poor phosphor quality.

But you can't replace just the phosphor, it has to be the whole lot that's replaced. 
Very green, all of that ewaste heading to the landfill. 

Integrated LED fixtures are the worst in my opinion. Better LED fixtures should either have replaceable lamps or replaceable drivers rather than a fixture that has no replaceable parts.

In addition, the manufacturing of LED fixtures is highly resource intensive because they require lots of rare earth metals to be mined out of the earth and even some precious metals such as gold to be mined.

To make an incandescent lamp, you don’t need to mine as many resources because they are simpler in design.

Better yet, stick with discharge lighting, that can last for years with only an occasional relamping. 

The manufacture of discharge lamps is also less resource intensive than LED lighting as well from what I understand.

Well we've got enough of a stock of 'proper' lamps to last the next thirty years, and failing those we do have all my oily lampses, with those you get the benefit of heat as well as light 

It really saddens me to see 60+ year old fixtures that were BUILT TO LAST get scrapped for crappy E-Waste. That’s the worst part for me. Instead of being relamped, they’re sent to the landfill.

I wish GE/grouse hinds would offer a LED "refractor" kit for there existing luminaires as I imagine the COSTS savings for the utilities doing the mass LED swap out AND even better for smaller ones running a "replace as fail" upgrade and for them it would be a drop in repair kit for otherwise good lanterns and also KEEP the street scape 

Yeah phosphors can wear out, resulting in dimming, colour shifts to purple, etc. Pretty much all LEDs have the phosphors bonded to the diodes so they're not replaceable although I suppose the old Philips AmbientLED lamps with remote phosphors can have the phosphors replaced by replacing the yellow phosphor covers on the lamp. Aside from the driver and diodes, there are other parts that can fail such as the diffuser sheets on a LED panel, resulting in uneven illumination on the panel. Technically not a failure in any electrical parts but due to the nature of the LED panels, people would scrap and replace them.

Yeah I don't think there's a proper LED retrofit refractor kit. Some decorative luminaires have a LED retrofit kit that is basically a replacement refractor with LEDs but they're model specific. Would cost more than a corncob but a properly heatsinked design with a remote driver should last longer than a corncob. You can get remote driver retrofit products for fluorescent retrofits though, but they're not that common.

As far as I understand, if you do not take LED drivers into account, LED diodes by themselves last a very long time. The electronics that power the LED fixtures themselves are oretty much the only point of premature failure.

Not really. The thing is, in a LED lamp product there are 100's LED dies and each has to have two connections. So we are talking about 200 bond wires, all high temperature exposed and when one single fails, the whole thing fails.
These numbers make the LED array alone less reliable than ballasts.
Of course, the major factor is the general quality and design of all those components, so you may have a piece of c**p for either of them, as well as quality long lasting design. But I would not expect if the quality level and design and production care is equal, I would not expect any difference between LED array itself vs its ballast.

I wish GE/grouse hinds would offer a LED "refractor" kit for there existing luminaires

I just wish they'd stop with the LED onslaught altogether. Nobody ever asked me if I wanted them.

I agree that I wish they left good enough alone and let consumers buy what they WANT and as far as I am concerned the energy saving do NOT offset the extra energy involved in mining / making LED units / drivers ETC VS a "simple" HID and choke and some of the MODERN HID lamps in Europe are quite GOOD for light output / life / efficiency compared to 20 years ago 

LED emitters themselves can be made to last a very long time and even with near to 100% lumen maintenance and without colour shift.  Indeed many of the early white LEDs from the golden era of the mid-2000s meet that claim.  The peak of lifetime reliability was reached with outstanding packages like Lumileds’ Luxeon-A in 2010, Cree XTE, Nichia 219 series etc.  Many packages even eliminated the gold wirebonds by shifting to flipchip designs with AuZn eutectic braze.

Everything changed with Seoul Semiconductor’s invention of the #5630 midpower package design in 2009.  That was originally developed for LCD backlights for the television industry but It was soon discovered that when driven harder, it delivered comparable initial performance to high power LEDs for lighting applications, for a fraction of the cost. 

Naturally, its lifetime suffered hugely.  But as mentioned above, the vast majority of customers are more interested in initial purchase cost than lifetime performance or future replacement costs.  The result was a mass storm of the low end of the lighting industry to midpower package designs.  I remember many of the Lighting-class LED manufacturers at that time all trying to dissuade their customers from going in the midpower direction because of the massive decrease in reliability.  For years they even refused to make such types themselves.  Of course, few listened.  The end consumers created a massive pull for the new low cost LED products, and it was impossible for even the most reputable manufacturers of luminaires, lamps and LED emitters to ignore that trend because it would have led to their own obsolescence.  The result is that the useful service life and reliability of the LED emitters used in general lighting products has been continually decreasing during the past decade, and huge efforts are still invested in cost savings that further reduce reliability.  Even the ‘gold’ wirebonds are usually now cheap gold plated silver wire.

But the average customer is happy because costs are decreasing and the energy savings that these cheap LED emitters deliver are easier to afford.  Government regulators are also happy because they see energy usage decreasing.  Manufacturers meanwhile are generally not at all happy because the emphasis has shifted mainly to lowest costs rather than quality, and this is what has cause many of them to exit the business or fall into bankruptcy.

Only in very recent times are regulators beginning to wake up, and realise that the story they were sold about LED lighting benefits may well have been achieved in terms of energy reduction, but was most definitely not achieved in terms of product life.  Its interesting to remember the studies made by the regulators in the mid 2000s which regularly spoke of LED product lifetimes being 20-50 years.  The reality is that few products even pass 5 years! 

This has not gone unnoticed at political levels, because the quantity of E-waste they have to deal with has skyrocketed during the past 5 years - mainly due to the failures of the LED lighting industry.  It’s interesting to see that the next round of EU lighting legislation has given up its former strategy of requiring additional energysavings.  The focus now is on the circular economy and repairability.  This now gives manufacturers some light at the end of the tunnel, by enabling a return to the considerably more profitable replacements industry.  Back to the “Fixture+Lamp” model that worked so well for over a century.  With any modern consumer electronics business, the profits are earned not so much on the initial products themselves, but on the spare parts and accessories (eg printers+inks, projectors+lamps etc).  I do not believe that the reliability of the LED emitters themselves will ever increase again because that would mean huge cost increases.  But perhaps the model of installing a new LED lamp/driver into a luminaire chassis every few years will result in better system reliability for the end users, jobs for the maintenance electricians, profitability for the manufacturers, and reduced E-waste for the regulators.