LED Bulb Reviews, Evaluations and Teardowns

LED Bulb Reviews

[ElectronUpdate] has posted many great reviews of commercial LED bulbs that one can purchase to replace standard E26 incandescent light bulbs. In his reviews he evaluates the light emitting performance and does a thorough and detailed teardown, evaluating and understanding the circuit technologies used. For the light emission evaluation he uses a light meter and some homemade graph paper to plot the lumens at different angles. Flicker is easily evaluated using a solar panel from a discarded solar path light connected to his oscilloscope. Any flicker will show up quite nicely and can be measured. Of course a kill-a-watt meter makes an appearance in most reviews to read watts and power factor.

Recently [ElectronUpdate] wanted to understand the meaning of CRI which is advertised on many of these commercial LED packages. CRI stands for color rendering index and deals with how colors appear when compared to a natural light source. After doing some researching he found that a CRI over 80 is probably good for LED lighting. The next dilemma was how to measure CRI without expensive scientific equipment. He found a website that we have featured before with free software and instructions on how to build a spectrometer. The web instructions include building a meter box from paper but he found it was much more reliable if built out of wood. We’ll let you follow [ElectronUpdate’s] recommended build if you like, but you’ll need a few items which he does detail.

After a short calibration procedure the final rig will measure power spectral line densities of your light source. [ElectronUpdate] is promising more details on how the colorful measurement data can be related to CRI ratings, but you can get a jump on the details at Full Spectrum Solutions. We also recommend you browse through all of [ElectronUpdate’s] LED bulb reviews on YouTube if the progressing performance and innards of LED bulbs fascinates you as much as it does us.

24 thoughts on “LED Bulb Reviews, Evaluations and Teardowns

  1. Interesting set of articles he’s been doing. I wish, though, that there was more text instead of a quick text lead-in and then a video link. Also, a chart of some kind that would allow for objective comparison of the various evaluated factors would be useful. I hope he continues to build out his reviews and his site!

    1. Thanks for letting us know. I’ve changed the link to “blogspot”…. I’ve never seen a sneak attack quite like this one before. The page loaded just fine at first. There must have been a traffic threshold at which it stopped mirroring the proper page and started serving up ad-spam.

  2. I tried hooking up a LED light bulb to a Kill-a-watt, and it didn’t even register as a load. I couldn’t even measure it’s power usage.

    I love LED lighting.

  3. In most of his videos [ElectronUpdate] points out that there is an RLC circuit on each bulb’s power supply that prevents too much EMI going back into the power line. But he doesn’t do any measurement of that interference. Can one put a number on EMI by analyzing mains voltage with an oscilloscope?

    1. Short answer yes, but not really. if substantial noise is generated it’s possible. However, interference takes multiple forms, including harmonic distortion common to SMPS (with high switching frequencies) and radiated noise. Measuring EM noise requires some pretty specialized equipment or some pretty specialized skills to use ordinary equipment. For energy star certification LED lamps are to be certified to FCC part 17, which is more stringent than CFLs that must comply with FCC part 15. If an LED lamp complies with FCC part 17, it (should) have a high power factor and limited EMI.

      1. I add the qualifier (should) because tolerancing in LED manufacturing remains immature when compared with other commercial electronics. Measuring a sample of a 100 units from ‘even’ name brand manufacturers can yield unacceptably high variances from specifications. As an example, I sampled 100 LED fixtures and measured the total harmonic distortion which was not to exceed 20%. The THD in fact averaged 27% with more than 3/4s of the units above 20% and half the units above 25%. Install 1,000 of these in a building and you’ve created power quality problems.

        1. Well, Part 15 says everything electronic is an antenna, and Part 17 you should put a light on it, then I guess a LED bulb should pass.

  4. The webcam is a pretty poor way to measure spectra — you’re combining spectral dispersion of the grating, plus the color filters on the camera, a very noisy imager, and the interpolation and color correction done by the camera software. All it can really show you is gross features of the light spectra. I’m not even sure you could calibrate around those issues.

    How to get a better spectrum for the bulbs: get a ColorMunki Photo (around $500, good spectrophotometer, lousy software) plus Argyll open source color management software. Argyll’s command line tools can read the raw data from the ColorMunki in reflective or ambient/emissive mode (and run the built-in calibration). Then just dump the spectra into your favorite spreadsheet and normalize the intensities (since different distances from the meter, plus different wavelength characteristics will make the peak value vary).

    Regarding CRI – it’s a horrible measure of how well the bulbs will render color. Everyone in the color science industry knows this, but there are entrenched interests that do not wish to change official measures because their own products really aren’t that great at rendering color. Your eye may see a light as white, but that doesn’t mean that food will look good under it…. Anyway, there are several proposed replacement measures, but they’re all bogged down in committee discussions.

    The recent Cree based WarmWhite bulbs are a pretty good tungsten match. Their daylight bulbs are better than anything else out there, but still not great (not enough cyan, green, or red).

    Yes, I dabble in color science. And no, I don’t work for any of the companies mentioned.

    1. I’ve used ocean optics spectrometers when working in labs but really tried to get around spending $4k+ when I was paying for test equipment. I’ve achieved some surprisingly good measurements using things like display colorimeters but when all is said and done, the amount of time spent benchmarking and calibrating such home built equipment is not economical. When UPerTek released the MK350, with a street price of $1700, it’s not a hard purchasing decision if you have to measure color as part of your job.
      As far as CRI, I agree, it sucks and everyone knows it. Using Ono and Davis’ CQS version 9, is the only sensible way to evaluate color quality. I know it’s been in NIST committees for years, and educating the masses is hard.
      At a client meeting the other day I help up two LED lamps. Both had specified CRIs of 90, and CCTs of 2700. I measured both with the MK350 and both registered 90 and 27K. However, one of them made apples look like potatoes and the other did not bleach out saturated reds. I advised this client that prospective suppliers should furnish samples or SPDs so that I could calculate color quality using CQS and that this should be adopted company wide. You make an apple look like a potato and people will jump on the CQS wagon.

      1. I’ve been using the Ocean Optics units for years. But yeah, their $2k – $4k entry prices aren’t exactly friendly (but still better than finding a working PR650). The ColorMunki is a decent spectrophotometer at $500, even if you have to get additional open source software to do much more than desktop calibration with it.

        There are some other, debatably better, light quality measures floating around the CIE – but, again, stuck in committees (or this week, discussed over drinks at CIC). Sadly, my political skills are not up to the task of moving them along.

  5. Oh, and Cree’s new TR lamp, available through Home Depot,is the first and only lamp I know of that meets California’s “VOLUNTARY CALIFORNIA QUALITY LIGHT‐EMITTING DIODE (LED) LAMP SPECIFICATION”. By the numbers it’s pretty great. But, it’s also 50% more expensive than the regular Cree–Home Depot lamps and I’m not sure if anyone/anything other than super color sensitive tasks could even discern the difference.

    1. The Cree WarmWhite is very good. The daylight, not quite as good.

      But yes, they are still better than anything else I’ve seen sold to the public. And the prices have been coming down…

  6. Even the shape of a light bulb is ugly. But LED’s in that form is just dumb. Concentrated light is known to be uncomfortable, that’s what diffusers and worse yet shades are for. LED’s can be spread about and aimed, then the heat will not be a problem as well. These bulbs will have thermal problems with everything crammed in to a small space and getting cooked on the top when on the ceiling. Enclosed fixtures will kill. With LED’s and panel lighting we will have to leave the oil lamp idiom behind.

  7. Last year I had to measure CRI’s of various light sources, however I didn’t have access to a spectrometer. Instead, I printed out a color palette sheet I found online, and then took photos of it with different light sources: outside, at noon under the sun, and then 1m away from each lightbulb above. Next, I counted how many colors I can tell apart from each photo, and calculated the percentage based on how many colors I can see on the control. I got some very interesting values, they looked very close to the average values you’d find for the bulbs I tested. Also, I needed to measure the color temperature, and I used my camera’s white balance to figure it out. The values were also quite close, I was surprised at how well my ‘homebrew’ techniques worked.

    1. That works decently, but you were limited to simple combinations of the inks from your printer. Using a MacBeth ColorChecker (original, DC, or SG) might have improved the repeatability and reliability of your measurements. Using those, you can quickly identify the deficiencies in the cheaper white LEDs even without a spectrum. But measuring the spectrum shows you why they are so bad (most are just blue or violet and yellow, with little to no green, cyan, orange, or red).

  8. At the end of the day does it really matter? The general adoption of LED light bulbs has arguably already reached the critical mass that then drives down costs and increases performance (as happened with all previous technologies based on semi-conductors).

    Given the cost of ownership between incandescent or even CFL lamps compared with LED it’s something of a no-brainer, no? I have a house full of LED and I could care less about lumens and CRI – I just made sure to get bulbs I personally liked and enjoy the much reduced electricity consumption.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s