I think you will find that most LED lights run the chips well above the maximum efficiency point, in an effort to reach a good lumen/$ point.
It's a rough race out there. Confusing marketing wording is used to cover up flaws frequently.
Deficiencies in standards can produce paradoxical effects.
Premium T8 lamps and LPS have a shallow cookie sheet curve. The initial dip in output is small enough to be ignored in design consideration. Using fresh out of package lumens per watt performance that is favored by LED practitioners will gain a few percent of output on paper.
Fluorescent lamps have a design lumen which is based on about half way down the relamp cycle (40% of rated life... you'd lose half of lamps if you let them run to rated life, so in practice, they're replaced at 3/4 of down the rated life in well planned group relamping). Due to LED's propensity to degrade significantly through their lifetime, if maintained output was used as the basis of design, it has more than a few percent to lose.
Premium T8 lamps maintain about 95% of initial output during their re-lamp cycle, but this could go down to 91-92% for the super extended life lamps that use two sets of cathodes (making them essentially like installing two lamps in parallel). T8 degradation slope is shallow enough to not warrant compensatory circuits.
LEDs tend to behave similarly to classic mercury vapor HID lamps. Very very long life before the lamp element goes poof, but steadily losing output.
http://energy.gov/sites/prod/files/2016/04/f30/mccullough_tleds_lightfair2016.pdf
The data presented on page 4 shows adopting design practice similar to the one used for HIDs, which starts off significantly overlit to maintain acceptable light level until the scheduled maintenance time.
Some extremely expensive LED fixtures utilize managed over-provisioning such as Lithonia's N80. N80 luminaires are open loop and it is over-provisioned by 25% above the maintained design lighting level and utilize lookup table built into the ballast calibration to raise up amount of power fed into LEDs based on accumulated run-time to counteract LED degradation. These have delayed, planned demand growth as opposed to starting off overlit at rated kW demand. Starting off at 80% throttle slows down LED degradation and it does extend the time before it can no longer maintain design lumen. This kind of degradation compensation can not be added without integrating all the hardware necessary to make it a dimming system.
GE breifly offered expensive electronic HID ballasts that applied a similar type of degradation compensation with metal halide lamps, so it wasn't a commercial success.
Degradation counteraction principle benefits LEDs and obsolete legacy lighting products alike. Take F96T12/110W HO for example. Though legacy magnetic ballast can operate from 250 to 305v, they experienced power drift roughly proportional to input voltage. A constant voltage regulator feeding lighting circuit with voltage set near the lower end and a mechanical timer that is used to raise the target voltage with accumulation of lamp hours would do the same thing as N80. Properly calibrated, illumination level starts off right at design light level and power consumption will continue to increase in proportion to lamp decay until power is raised to permissible maximum level and once maximum power is reached, lamps are allowed to depreciate.
Take a look at the Cree Xlamp datasheet.
Luminous flux at 25C is 10% higher than at 85C.
Luminous flux at 375ma is about 60% of nominal, at 3000ma it is about 325% of nominal (8x current and 5x output).
Voltage at 3000ma is about 10% higher than 375ma.
But the chips are expensive. Run the chip at higher power and while it is less efficient you get more lumens for each $ worth of chips.
-Jon
Exactly. That's the trickery used by the $50 PR stunt light bulb. That thing used 18 expensive high power chips (which has lower degradation than medium/low power plastic SMDs that are more prevalent today)
AC 60Hz is first converted into DC by the front-end and converted into high frequency AC. A detectable amount of line frequency ripples pass through to the output, but a high quality electronically ballasted fluorescent systems have less flicker than incandescent lamps.
When you observe it on an oscilloscope, a good ballast will have a straight line with almost no artifacts while the DC rail on cheap LED ballasts will look like an egg crate which causes high flicker index.
Sometimes, cost can be shaved by compromising on performance parameters not specifically required by various standards. Lumens, watts, kelvin and CRI are common requirements and power factor rating increasingly so. Cost engineers have figured out that flicker index and flicker percentage have not been explicitly dictated and they've figured out a way to cut corners on LED ballast to raise power factor to meet regulations while maintaining lumens/watt, but many of these products have flicker index on par with or an order of magnitude worse than magnetically ballasted fluorescent lamps and induce headaches. LED elements are expensive and leaves less budget available for LED ballast.
A slightest 60/120Hz component in output causes disproportionate fluctuation in instantaneous power fed to LEDs due to natural characteristics of LEDs. This will result in flicker.
http://www.amperor.com/products/led/constant_voltage_constant_current_led_driver.html
Some of the well designed fluorescent ballasts have power factor in excess of 0.97, 90% efficiency, <10% THD, flicker index below incandescent with fantastic transient response that lamps are well sprung and rides well. So good that you could power it the same outlet as a large compressor and you're not going to be able to "see" when the compressor starts up.
el-cheapo LED ballast in my sample of glass bulbed Cree 60W bulb has significant flicker index and will make sure you'll know about every pebble you ride over despite meeting government and voluntary standards to bear Energy Star label. For example, the inductive kick back from a bathroom fan shutting off can confuse this dimmable LED ballast and make it flash. Well, this kind of adverse reaction is not specifically prohibited, so who cares?
I haven't had a chance to test the newer generation, 4 Flow LED bulb that sports steeper LED degradation slope as well as inferior initial LPW despite having a lower unsubsidized purchase cost.
