Comparing light patterns for Cree, Philips LED bulbs

Incandescent light bulbs produce light omni-directionally so that the light falls evenly through a spherical pattern. LED’s, however, are inherently directional: They perform like tiny spotlights. Designing an LED bulb that mimics the hemispherical light pattern of an incandescent is one of the biggest challenges for bulb designers.

Here’s a simple but practical way to compare the light output patterns of several LED 60W-replacement bulbs, an incandescent bulb, and a CFL bulb. Rather than just take photos of bulbs against a white background, this test set-up uses a white paper ring that angles out around the light source at 45 degrees, bouncing the bulbs light toward the camera which is set at a 1/60 shutter speed and an F21 stop. (Many thanks to my husband Doug Conner and his Canon Rebel.)

Here’s the test set-up:

Light Pattern Test setup off

…and here’s the same view with the room light off and the light-under-test turned on:

Light Pattern Test setup on

This is a head-on photo of how the light looks from an incandescent bulb, which I’ll use as the baseline standard.

Incandescent bulb light pattern

The light is evenly bright around the ring, with just a small shadow underneath.

The CFL is similar, with just a slightly larger shadow.

Light Pattern CFL bulb

Impressively, the Philips L-Prize LED bulb has an even smaller shadowed area than the incandescent:

Philips LED L Prize bulb light pattern

You may recall that the L-Prize bulb uses a remote phosphor (that’s what makes the bulb cover that unusual yellow color) and effectively disperses its LEDs’ light in a near-perfect hemisphere. It also makes for a more expensive design: The bulb costs $25. It’s likely that changing to the cheaper sno-cone design, as well as eliminating the dimming circuit, were at least part of the reason Philips is able to sell its new 60W-equivalent non-dimmable bulb for $15. 

Both Philips bulbs are 60W-equivalents: If you don’t care about dimming, will the cheaper bulb work suffice?

Philips 10W LED non-dim bulb

It depends on how you’ll use it. As you can see, the light is mostly projected out into the top half in a semi-hemispherical pattern. If you expect an omni-directional, even light, the $15 bulb may not suit your needs.

Cree’s new 60W replacement bulb does an admirable job of filling out the hemisphere, with the shadow at the base smaller than the CFL and the same or less than the incandescent. The Cree mounts its LEDs internally on a pedestal with the bulb, and the bulb seems to have a very faint reduction in brightness at the top of the ring. You’d have to have an integrating sphere to measure this, and I believe the difference in brightness is minimal. The Cree bulb, which is dimmable, sells for $13.

Cree 9_5W dimmable LED bulb

(Here’s a detailed tear down of the Cree bulb.)

 

Comments

  1. Nice comparasion. I learned a lot. There are some omni-directional LED bulbs in market such as GE lighting’s and LSGC’s. It would be better for consumer if you could collect and add more bulbs for a comparasion. We also would like to know who is the winner.

  2. Any reason you didn’t look at the updated Philips Omni-directional? It’s a more direct comparison to the Cree bulb due to it’s omnidirectional nature, cost (now 14.99), dimmability, and relatively low CRI (80-ish) in comparison to the L-Prize. The Home Depot part number is 424382.

  3. You must have a HUGE dog, Margery….

    Nice analysis – as usual

  4. One problem with Phillips LED bulbs is that they can get quite hot. I haven’t measured the actual temperature, but they’re definitely too hot to touch. This can’t be good. Did you notice Cree bulbs heating up?

  5. Margery, your method is flawed.

    The Cree bulb, from independent tests has more that a minimal reduction in output upwards. The loss is > 25% if not higher compared to the sideways emission. It is enough that they would not pass the new tests.

    One issue is your use of a standard camera (and monitor) with associated non-linear transfer functions so that images appear more as your eye would perceive them.

    The second issue will be reflections off the diffuse paper onto other sections of paper that will smooth out any bumps.

    Last, you would not use an integrating sphere but a goniophotometer to measure this.

  6. The Philips 420240 is cheap, and not really worth it. These seem okay out of the box, but the capacitors go bad very quickly. They flicker noticeably after a few months, and some just outright fail. The 424382 is a few dollars more, but a much better value.

  7. Boris – There are some thermal measurements for the Philips and Cree bulbs at ledbenchmark.com. The Cree bulbs do get quite hot as well.

  8. Jack – Yes, this is a much cruder way and inexact to indicate light patterns. But it does show at a glance the difference in the light output among the different bulb designs, even the unevenness of the Cree’s top light, and has the advantage of using commonly available equipment such as a digital camera and paper.

  9. Take a look at Home Depot’s most expensive 60W replacement bulb – the GE – with your dog collar, Margery. I think we’d all be interested in your findings.

  10. Thanks a lot for this link Adrian!
    It says Cree gets up to 89°C (192°F).
    That’s a lot hotter than what I experienced. After a couple of hours my house, a Cree lamp was very warm to the touch, but not unbearably hot, like a Phillips. No way it was above 60°C.
    Still, I guess enclosed fixtures are a no-no. This makes switching LED bulbs an expensive (and labor-intensive) preposition in my house.

  11. Christopher Yochum says:

    Hi Margery,

    Your demonstration is a singularly useful way to evaluate one of the biggest problems with LEDs. Inadvertently, you have touched on a pet peeve of mine: The paucity of LED-based globe bulbs that look good and work well. I’ve found, oh, two that look like standard incandescent globes without an overdone heat sink structure or weird shape. I can get a Cree bulb for under a lampshade, but I’m not going to put something like that in our bathroom fixtures. Perhaps in the future, you might do your demo more times, expanding on the bulbs, or types of bulbs you compare. By the way, the two I’ve found that look “pretty” are the GE 76464 and the Samsung 2323. Your demo has whetted my appetite for a comparison of light-distribution patterns for globe bulbs. Even if you don’t, thanks for the current demo; it’s very clever and the only one I’ve ever seen that demonstrates empirically what these bulbs do. Nice job!