Four years ago I wrote a post comparing the Sunwayfoto Fill Light FL96 and the Aputure Amaran AL-M9 mini LED light. They are both still available new. Sunwayfoto has released a few variations on the original model, the FL96 C with batteries with increased capacity and faster charging through USB-C, and in the case of the FL120, slightly higher light output and power-bank function.
A few days ago I bought an Elgato Key Light Mini, and decided to run a new and more thorough comparison between these the three commercial LED light sources that I own. In part because I am now able to compute CRI and CCT values from the measured spectra and because the Elgato Key Light Mini has much improved diffusion. I also add to the comparison some LEDs bought as electronic modules.
The three LED lamps differ in size and weight, however their maximum light output and battery capacity is not proportional to their size or weight.
| Feature | AL-M9 | FL-96 | Key Light Mini |
| Power (W) | 3 (?) | 8 | 15 |
|
Lumens |
180 (?) | 600 | 800 |
|
Lux at 0.6 m |
(300) | 1120 | no spec. |
| Diffuser (cm2) | 34 | 50 | 100 |
|
Illumination |
behind diffuser | behind diffuser | diffuser edge |
|
Weight (g) |
70 | 132 | 300 |
| Battery (mAh) | 1800 | 2800 | 4000 |
| Control | Buttons | Wheel | Buttons + Wifi |
| Display | No | OLED | No |
|
External power |
charge | charge | charge + use |
I measured spectral irradiance at 160 mm distance from the three LED lights in a darkened room. I repeated the measurement on the Key Light Mini as it turn out to be way out specifications. I tested over the whole range of dimming and of CCT values of each light, but not every possible combination. Of special interest is the change in light output vs. CCT because the variation in CCT is implemented by mixing light from two types of LEDs.
We first look at the output expressed as energy irradiance, which is most likely to remain constant with CCT change if the sum of the electrical power driving both types of LEDs remains constant. The AL-M9 has nine identical LEDs and is rated at a CCT of 5500 K, and serves as reference. The FL-96 and the Key Light Mini each is based on LEDs with emission at two different CCT values. In the FL-96 the change in CCT is implemented so that adjusting CCT does not change significantly the total light output, while in the Key Light Mini, full output is obtained only at CCT values close to 4000K and at extreme values of CCT the light output is only only half the maximum. The FL-96 is far superior in this respect as its full rated power is available at all CCT settings. In the FL-96 adjustment of dimming and CCT are independent of each other (orthogonal controls) which is also easier and more intuitive to use.
Surprisingly, the much smaller FL96 outperformed the Key Light Mini in light output at the very short measurement distance 16 cm I used for most tests. I expected that farther away, given the larger emitting area, the Elgato LED light would catchup with the Sunwayfoto, but a measurement at 64 cm showed the FL-96 continued outperforming the Key Light Mini. This is surprising as the Key Light Mini according to specifications emits 33% more light than the FL-96, 800 lumen vs. 600 lumen.
Output from LED lights is normally described in specifications in lumens and lux, which are technical quantities designed to approximate the spectral response of human daytime vision. The spectral response of camera sensors is different, and to an extent dependent on the camera model. We can convert the energy irradiances in the figure above into illuminance expressed in lux (= lumen × m-2). Given that the eye is most sensitive to green light, the shape of the curves very slightly changes (not shown).
At 0.6 m with both lights set at 4000 K and 100% output, I measured an illuminance of 825 lx for the FL-96 and 730 lx for the Key Light Mini, the value for the FL-96 is about 13% higher. The rated illuminance output of the FL-96 is approx.1100 lx at this distance (from a total rated output of 600 lumens). For the Key Light Mini I have not found any illuminance specifications, while the luminance rating is 800 lumens, or 33% higher than for the FL-96. I measured the output in surroundings with low reflectivity and the FL-96 has been in use for some years. The Key Light Mini has better diffusion and larger emissive surface, thus a broader light beam that may explain the lower illuminance. If we look at the relative change in illuminance with measured CCT, then it is obvious that the FL-96 is much more versatile if one needs illumination at different colour temperatures, in spite of its narrower nominal CCT adjustment range. At 5500 K or 3000 K, illuminance obtained with the FL-96 is more than twice as strong as with the Key Light Mini.
Next we explore how well the nominal CCT reported by the LED light sources matches the CCT computed from the measured spectra. A LED lamp with nominal and measured CCT in agreement is described by the black dashed line. We see that the nominally much broader range of colour temperatures in the specifications of the Elgato LED lamp (2900 K to 7000K) compared to the FL-96 (3000 K to 5500 K) does not match the measured values, approaching those of the Sunwayfoto LED light which performs very close to the reported nominal values.
The three light sources differ in how dimming is implemented. As shown in the post of four years ago, dimming of the Amaran AL-M9 is done by pulsing (PWM at 40 kHz) while that in the Sunwayfoto FL96 is done by constant current (CC). The dimming in the Elgato Key Light Mini is done by PWM at 146 kHz. The dimming of the Amaran AL-M9 is done in 10 steps, while in the other two LED lights the setting is described as a percentage with a resolution of 1% for the setting.
In principle constant current dimming is preferable to PWM, as with PWM light output pulses between maximum output and off, with varying duty cycle, rather than being constant in time. When the frequency is very high, the effect is unlikely to be reflected in the images we take. In the case of 40 kHz, one might start seeing inconsistent exposure on successive image frames with shutter speeds close to the maximum available in digital cameras, around 1/16000 of a second. PWM at 146 kHz is unlikely to affect any normal photographic camera. Many modern digital cameras can scan the light to determine the frequency and automatically adjust the shutter speed and metering system to ensure consistent exposure from frame to frame. Let’s not forget that illumination at many home and businesses flickers at 100 Hz or 120 Hz, because of the frequency of AC mains power. However, only in the case of LEDs the pulsing is on/off. With incandescent lamps and fluorescent tubes the fluctuation is between two levels of brightness. Most problematic is when different light sources, possibly with different emission spectra, are pulsed at different frequencies. This can result in very clear colour banding in photographs, especially when using electronic shutter in cameras in which rows of pixels in the sensor are read sequentially.
In the FL-96 the real dimming is less than the nominal one, while in the Elgato LED light, the real and nominal dimming match very well. It is remarkable that the Elgato LED light can be dimmed down stably to less than 5%.
All three LED lamps are described as having very high values for the colour reproduction index (CRI). The figure shows how the measured CRI varies with measured CCT. All light perform reasonably well with the FL-96 and the AL-M9 overall performing better than specified and the Key Light Mini, tends to perform below specifications, with specifications that are lower than for the other two LED lamps. A difference at 5500 K between 97.5 and 93 may for subjects of specific colours and cameras.
Probably in actual use, one of the most critical aspects of performance is that dimming should not affect the colour temperature. This is likely to be most difficult to achieve at intermediate CCT values when the light is coming from two different types of LEDs. I tested this at 4000 K and as shown in the plot below, there is an important change in colour temperature when light output is strongly dimmed.
The change in the spectrum with dimming also leads to a small decrease in CRI.
The light diffusion is much better, given its larger size and diffusion by illumination from the edges of the panel rather than a simple plastic diffuser on top of the LEDs. How much difference this makes depends on the situation. Another advantage is the remote control through Wifi. This is indeed very handy, but is marred, because as we saw above by settings do not agree with the real CCT values. I also find rather problematic that the light output changes drastically when changing the CCT setting away from intermediate values.
A question I did not address in my previous post is if the light sources matched or not the specifications given by the manufacturers. Here is where I was most surprised. While the AL-M9 and the FL-96 were very close in CCT to the setting at 5500 K (at the “cool” end), the Key Light Mini was more than 1000 K warmer than settings (less than 6000 K at the 7000 K setting) This is a huge difference! At the “warm” end differences between settings and actual performance both for the Sunwayfoto and Elgato were rather similar in magnitude and smaller. For the FL-96 the difference between CCT measurement and settings was largest at the middle of the CCT range.
Specifications for CCT in the case of the FL-96 are plus or minus 200 K. For the Key Light Mini Elgato makes no promises other than a CCT range of 2900 K to 7000 K.
Given the large difference in CCT, I contacted Elgato support. The initial answer was fast and once the problem was confirmed the support staff consulted the technical staff. The last information receive was that the question had been passed along to the manufacturer. This was more than three weeks ago. The suggestion from support was for me to return the Key Light Mini to the local retailer.
This article is based on a single unit of each LED light. I do have two FL-96 and two AL-M9, and there seems not to be any differences within these pairs, but I haven’t done measurements to compare them. Anyway, in both cases they I bought them at the same time, and thus can be expected to be from the same production batch. I have a single Key Light Mini, and it is impossible to know if the CCT difference affects only my unit, a whole production batch or all units sold. That Elgato did not offer to replace it, suggests that possibly all recently manufactured units similarly. If you are reading this and have access to a Key Light Mini, please illuminate a white balance card with it with no other light present and check with a camera the report white balance colour temperature.
Take home message. Did I return the Key Light Mini? No. Would I have bought it, knowing what I know now. Yes, as alternatives do not have built-in batteries. I mostly use 4000 K, the CCT at which light output is reasonably strong. What I find most annoying is the mismatch between the settings in the App and the actual CCT. The actual range of CCT even if not as broad as promised is enough for me. For the intended use it is still the best of the three lights, because its large surface and better diffusion makes it much more comfortable on the eyes of the subject. When illuminating shiny objects at close range it is also better than the FL-96 and AL-M9, which can produce multiple bright spots that look very unnatural. Overall, I think the Key Light Mini is better than the other two LED lights when used as main light. The FL-96 is better as a fill light (and doubles as a pocket-sized bright flashlight). So each one performs best for the purpose they were designed for (the FL in FL-96 is for “fill light”). The AL-M9 is small and light weight, but in practice I have very rarely used them after buying the two FL-96. There is room for improvement in the Key Light Mini, and hopefully an upgraded version will be produced in the future. However, I do find it disturbing that specifications are not fulfilled by a product from a well known brand. This raises the question of what is the real CCT range of LED lights from other brands which are rated for 3000 K to 9000 K.