Test results for LED, CFL, and incandescent lamp evaluation
by Doug Conner
This evaluation project was undertaken to quantify some of the characteristics of lamps (light bulbs) affecting user experience. Tests were carried out to measure LED, CFL, and incandescent lamps for turn-on time, warmup time, and dimming performance. During the testing I also added 120 Hz luminosity fluctuations (ripple) to the testing because I found a large variation.
The first section will provide graphs giving comparisons of the lamps tested. At the bottom of the page are links to pages containing plots of the individual lamp data. You can also see the test setup used.
Initial turn-on time
Incandescent and LED lamps normally rise rapidly to full luminosity. Although LEDs are capable of switching on faster than incandescent lamps, the time required for the power supply to ramp up limits the LED lamps to about the same or slower speed than the incandescent lamps.
CFL lamps (compact fluorescent lamps) typically have a stepped start-up with an initial time delay and no light output, then a rapid rise to some low level output, followed by a gradual increase to full power. The following plot gives a general comparison of LED versus CFL initial turn-on time and brightness. LEDs don’t have a warm-up time.
The following plot shows the time and % of maximum lumen level reached after the the initial turn-on delay for all the lamps tested.
Note that the incandescent and LED lamps all have a turn-on time of less than 200 msec with the exception of the Philips 12.5w LED which has a turn-on time of 460 msec. All of the LED and incandescent lamps reach 80% or more of their maximum lumens after this relatively short turn-on time.
The situation for CFL lamp turn-on time is more varied and generally lower performance. Only one CFL managed a turn-on time of less than 200 msec and initial light output was about 45%.
The two flood lamps tested both had very poor turn-on characteristics which may be a characteristic of CFL floodlight design. Initial turn on times of about 1 second before any light output coupled with very low light levels (around 10% of maximum lumens) means you may actually have to wait for the light before entering a truly dark area. In defense of the Feit flood I will note that this CFL is rated for outdoor performance (something all the others are not) which may result in necessary compromises to turn-on performance.
CFL warmup
The long warmup time required by CFL lamps to reach 80% of their maximum lumens is shown on this chart. Ranging from 25 to 90 seconds, these times are quite long and one of the significant differences when compared with LEDs.
At low temperatures these warmup times are even longer. Testing the 23w Feit flood after cold-soaking it at 31 degF, the warmup time increased to 242 seconds. The 4 minute wait makes this lamp one that requires planning ahead when you need light. It’s not a light you want to flip on to see who just arrived at your front door.
Dimming performance
All three of the LED lamps tested performed about he same (only one shown above) with a linear relationship of luminosity to input power. the CFLs performed nearly the same for power levels above 60% and then deteriorate to lower lighting efficiency at lower power levels. The incandescent lamps are the most power inefficient when dimming, requiring about 80 power to achieve 50% light output. All the dimmable lamps tested would dim below 20% of maximum lumens without flicker or other problems. The dimming switch sometimes makes irritating humming or buzzing noise when dimming.
120 Hz luminosity ripple
Whether a lamp uses the 60 hz 120 Vac directly, as in the case of incandescent lamps, or converts it for LED or CFL use, some level of AC voltage fluctuations usually ends up in the light. Incandescent lamps experience a small heating and cooling of the filament that results in 120 Hz luminosity ripple. LED lamps use switching power supplies that operate at much higher frequency than 60 Hz although they must first work with the rectified 60 Hz input. If not filtered the 60 Hz input appears in the light output.
Not every Lamp was tested for luminosity ripple, just those that appear in the chart. There were two noteworthy results. The 12.5w Philips LED had no measurable ripple in my tests. This leaves me wondering whether the relatively long turn-on time of this lamp (460 msec) was the caused by large filter capacitors after rectification and the result is the low ripple. The contrasting LED was the Samsung 10w that had by far the largest ripple level of 19%. I also notice the Samsung lamp tied for fastest turn-on with the incandescent 38w lamp. I have not studied luminosity ripple sufficiently to determine at what level (if any) it causes problems.
The true 120Hz luminosity ripple is probably higher than I measured because my readings use a 13 msec integration time which would provide some filtering of the true ripple level.
How I measured 120 Hz luminosity ripple
The above plot shows the 5% luminosity fluctuations in the lamp’s output. The 1.2hz frequency is actually the beat frequency of the 33 msec sample rate and the 120 hz luminosity fluctuations. The 33 msec sample rate (30.3 hz frequency) has a 121.5 hz harmonic that mixes with the 120 hz frequency, resulting in a 1.2 hz output.
Some of the turn-on time tests (on the individual test result pages) show 5 Hz fluctuation because those samples were taken at 40 msec, causing the 125 hz harmonic to mix with the 120 hz.
Plots of individual lamp data
LED lamps
CFL lamps
Incandescent lamps