We occasionally receive emails from users who say that their lights are measuring too warm or too cool on their Lumu devices. Our experience shows that when we send a replacement light that we verify to be in spec, the measurement difference remains consistent on the user side.
Your MediaLight is designed to give you years of accurate backlight and dim surround. They are designed to be accurate out of the box, and to remain accurate (to within 1 MacAdam ellipse for the first 20,000 hours and within 2 ellipses for 50,000 hours) through the life of your MediaLight.
Our lights are tested multiple times during production. First, in the selection/binning process, next, during the design and testing phases and, finally, by independent third parties, such as the Imaging Science Foundation.
We maintain our own barium sulfate coated integrating sphere at our home base in Kinnelon, NJ. Our Lisun LMS-7000 spectroradiometer has a spectral resolution of ±.2nm and is accurate to within ±.003 of chromaticity coordinates (Δx,Δy). During testing periods, our spectroradiometer is tested for drift weekly against and calibrated to a reference (tungsten) lamp. Completed strips are also spot-checked with our Sekonic C-7000, verified on a Colorimetry Research CR-300 by Flanders Scientific, and certified for color accuracy by the Imaging Science Foundation.
Generally speaking, users with Lumu (a popular crowdsourced metering dongle for mobile devices) devices are reporting warmer measurements. Our own research confirms these findings -- with readings coming in about 200K warmer or cooler respectively.
So, what's going on here?
Are the lights wrong? Are the meters inaccurate?
The most likely answer is neither, but there are some points worth detailing.
Is the MediaLight only designed for consumer use, and not accurate enough for professionals? What are our tolerances?
The vast majority of MediaLight customers are professionals. They purchase both our standard and Pro lines and both are extremely accurate. The binning tolerance for the MediaLight standard line is +/- 75K and the binning for the Pro line is +/- 50K. For reference, the Ideal-Lume and Ideal Lume Pro bias lights report accuracy on their independent lab results within +/-200K. The Ideal Lume Pro actual measurements are about 6250K out of the box (about 100K higher if you remove the diffuser), so if your meter is measuring it as 6500K out of the box, this might also explain some of the measurement error with our lights. Ideal Lume doesn't measure exactly 6500K out of the box.
The MediaLight Mk2 is designed with CRI ≥ 98 chips and the MediaLight Pro is designed with CRI 99 chips. During binning at the SMD (LED) factory, the LED chips are graded in a temperature and humidity-stable integrating sphere with a spectrally neutral and high-reflective coating (barium sulfate). The sphere combines the light from all viewing angles before a separate measurement is taken for each SMD chip.
Still, we do occasionally receive emails, chats and calls about users measuring different values, and there can be multiple reasons for this:
1) Measuring device. The differences might be within the spec of the measuring device. For example, a popular Kickstarter device called Lumu claims accuracy within 300K for lower CCT and within 3000K for higher CCT of ~10,000K. It's very possible that two lights with different spectral power distribution but similar CCT can measure differently on this and other meters. Again, two light sources can have the same CCT with different spectral power distribution properties and these different SPDs can contribute to different measurements on some devices.
In other words, the performance of some meters can be more precise with certain light sources, wavelengths and within certain color temperature ranges. CMOS sensors, in particular, are usually more accurate at measuring lower color temperatures and, in fact, there is more of a visible difference between 3000K and 3100K than between 6500K and 6600K. Small differences are more noticeable at lower CCT. In fact, the JND (just-noticeable difference) at lower CCTs like 3000 is 100K, while the JND at 6500K is closer to 500k. This impacts the tolerances of CMOS devices.
2) Environment. The reflecting surface might be impacting measurements or measurements might be taken directly from the lights without being reflected first. It's not uncommon to see a brand new gray card present results that can vary by hundreds of degrees from an older gray card. From our experience, older gray cards tend to drift towards warmer CCT as they fade, yellow and/or oxidize.
3) Angle. If you measure the blue-emitter based MediaLight (standard version) directly, the blue LED engine could impact the measurements and result in cooler color temperatures. Measuring reflected light on a spectrally flat gray surface may result in warmer measurements than you'd see with direct measurements. Reflected light is what we see when we look at bias lights, and this is the measurement that we use in our testing process.
4) Calibration. The meter might need to be calibrated, or may be operating well within the spec of the meter manufacturer. Simply put, if the measurement is off by 200K or even 500K, this might be within the intended spec of the meter manufacturer. Lower resolution meters make assumptions about the parts of the SPD that they don't detect. This is partly why two different light sources with the same chromaticity coordinates can measure differently on different devices.
5) Diffusers. Using a diffuser on either your MediaLight or a reference light might impact the color temperature of your measurements. From our experience, even diffusers touted as ND (neutral density) often result in a slightly warmer color temperature (about 200K). In fact, while designing our MediaLight Beam bulbs, we increased the CCT of our COB slightly to offset the filtering of the bulb's housing.
6) Reference source. You might be measuring a reference source with a baseline of 6500K, which has drifted from, or is still within manufacturer tolerances of the target CCT. For example, Ideal Lume LED lights can measure ~250K lower than 6500K out of the box and still be in spec. If your meter measures it as 6500K, there is likely some error in the measurement.
We reiterate again that different meters can return different results.
If you've read this far, we don't mean to discount the possibility of a defective MediaLight, as everything is within the realm of possibility. We take accuracy extremely seriously and work with users to identify any causes of any discrepancy and resolve them. However, when we do encounter defects, they have historically been related to wiring, switches, dimmers or other items that don't impact the color temperature of the lights. If we send a replacement unit, which we've tested on our calibrated equipment and you find that the measurements are still off, the possibility of device error should at least be considered.
In the cases of very large discrepancies, the LED binning process prevents the inclusion of LEDs with vastly different color temperatures (we aren't buying any LEDs outside of our required spec and we don't make other products where we might use the wrong chips for a given production run), so the likelihood of there being a strip where all of the lights are 5500K, for example, is extremely small.
On lower quality LED strips with wider binning tolerances, you can sometimes see a visible difference between the LEDs when the lights are closer together. If you look at your MediaLight, you will see that the LEDs all look identical to each other.
If you've read this page and don't feel that any of these points apply in your case, please feel free to reach out to us. Every meter returns slightly different results, but your measurements should be reasonably close to our factory and lab measurements. And don't forget, if you simply don't like the lights, you can return them in any condition within 45 days for a full refund.