Seems like 95% of the LED retrofit lamps that go bad are caused by failures in the driver electronics in the base of the lamp, not the diodes themselves failing.

I wish the manufacturers would separate both parts. They could sell universal screw in driver pieces, and have the top half of the lamp with the diodes screw in separately.

That way you don't have to replace the diodes if the electronics fail, and you can change the color temperature of the diodes without replacing the electronics.

Ah well you see that would require mixing money with common sense. LED retrofit companies lack the ability to use one or both.

Interesting concept  , and that would have benefited the very early (expensive) lamps, but looking at today’s micro-drivers that are fitted to the latest filament lamps, it’s not worth a 2 part lamp, also given now how the Chinese have brought manufacturing prices down to acceptable levels and lamp life is also good, I suppose it’s not an issue anymore?

Interesting concept  , and that would have benefited the very early (expensive) lamps, but looking at today’s micro-drivers that are fitted to the latest filament lamps, it’s not worth a 2 part lamp, also given now how the Chinese have brought manufacturing prices down to acceptable levels and lamp life is also good, I suppose it’s not an issue anymore?

Yeah with the prices that current domestic LED lamps go for, it's not worth it IMO. However, the concept might be worth it for higher power lamps intended for HID replacement which cost more, either a lamp with a replaceable driver or a lamp with a external driver. In the case of HID retrofits, the external driver would be installed where the ballast would have went.

One of the most frequent causes of premature failure are failed bondwire welds on the LED die metallization. So really a LED fault (or better to say thermal mangement fault as these welds do not handle high temperatures well, but resulting into LED failure; the nature of the thermal managementfaults are often improper "thermal grease" application, incorrectly tightened heatsinks, leak or generally bad gas fill in case of the filament designs). To me these tend tohappen at similar rates as the ballst failures (related mainly to their complexity and sensitivity to a single point failures). Making the components separated may seem to help (like on the circle retrofit boards containing a bunch of LEDs and their ballst, there you may easily fix whatever breaks), but mainly with the modern linear JFET based constant current regulator the ballasts became so simple they do not fail that often anymore.