Yup, that's marketing.
And then, these bulbs don't last long, either. They work fine for about 3 months, but then they start randomly getting dim for a minute or two, then getting bright again for a minute or two, over and over.
Eventually they just stay dim and then go out a month later.
Can someone tell me what the problem is and how to fix it?
They are working exactly as they are meant to.
- Build uber-cheap bulb that barely clears UL listing (UL lists for safety, not quality).
- Consumer buys bulbs.
- Profit! On government or utility subsidy program.
- Bulb has very short life. As intended.
- Consumer does not return bulbs.
- Profit! Because consumer did not return bulbs.
- Consumer buys more of the same bulb because consumer likes "cheap".
- Or consumer rejects this brand, no problem we have other brands of the same thing.
- Rinse, wash, repeat.
"You not returning them" is an essential part of the plan here. The vast majority of consumers do not, and that's how they make money. (the same is true of the sellers of awful or even fake products on Amazon, like the "1TB" USB sticks that are actually 16GB repeated, and will fail when it becomes full enough).
As you can see, this lackadaisical attitude toward being ripped off has created a perverse incentive. All of these companies do a "number-crunch" on the percentage of expected returns, and figure out how to price so they are profitable nonetheless. If I had all 6 infinity stones, I would snap my fingers and cause all Americans to be of a mind to diligently return all such products. Home Depot, CostCo and Lowes would file bankruptcy a month later.
Since you asked, my "never-buy" brands are any "house brands" of big-box stores, and Utilitech, Lights of America, Feit Electric, and Commercial Electric. My go-to brands are GE, Cree, and (surprisingly) Ikea.
Overdriving bad. Derating good.
You are absolutely correct that the LED emitters in those bulbs would outlive all of us, if the drivers and build quality were up to snuff, and the LEDs hadn't been overdriven to destruction. Because it is possible to drive LEDs so hard they burn out, and the purpose of the current limiting IC on Big Clive's video is to peg them right at the cusp of burn-out.
What Clive is doing with modifying the resistors is effectively turning the cheapo LED into a "Dubai LED", which (by order of the Emir) must hit a 200 lumen/watt target. They use a larger number of LED emitters per lumen, and don't so severely over-drive the LEDs.
You can hit mouser.com and browse their LED emitters and see the data sheets of any of them, but they typically state "spec" performance at a low current, and a "max" performance at a much higher current.
Take this one. The spec is based on a forward current of 350ma, even though the device can run 2000ma (almost 6 times higher). But let's look at what happens there.
- At 350ma, you get 100% of spec lumens. Voltage at 350ma is 2.69V. (page 18) So total power is 0.941 watts.
- At 2000ma (571% of current), you get 420% of spec lumens. Voltage at 2000ma is 3.07V. So total power is 6.14 watts.
So for 652% of the power consumption, you only get 420% the raw lumens, but we're not done. The higher ampacity will require running quite a bit hotter, and on page 16-17 we have a thermal derate of 10-15% for the higher temperature. So 380% of lumens may be more realistic (for 652% the power). So you can see it's not terribly hard to double efficiency simply by using better LEDs and not over-driving them. And this will vastly increase service life as well.
So this is the secret of Dubai LEDs.
Now if your cheapo LEDs are built like Clive's many samples, you can identify and pulverize one of the two resistors with a pliers, to radically change the power setting (and somewhat less radically change the brightness, since you are climbing the efficiency curve considerably). However, that will only tell you if the rapid burnout is due to overdriving the LEDs, or due to a plethora of other design shortcomings typical in el-cheapo LEDs.