2024-01-27
The existence of the Moonleds MN-S3535 in its high CRI version recently appeared by chance in the BLF. A special feature of this LED is the extremely high CRI of typ. 97 and the likewise very high R9 and R12 of min. 90. However, the very high thermal resistance of 8.0 °C/W could significantly limit the electrical performance and therefore the maximum luminous flux.
This test clarifies exactly how. These emitters were provided by BLF user @INeedMoreLumens. Many thanks for that!
The emitter tested here was delivered to me from a sample order in January 2024 from the USA.
Tj 25 °C, If 700 mA
- Type: single die, lateral
- Bin: 190-220 lm
- Color group: 6000-6500 K
- CRI: typ. 97 (R9/R12 min. 90)
- Rated voltage: 2.8-3.2 V
- Max. Forward current: 1,000 mA
- Max. Peak current: — mA
- Viewing angle: 120 °
- Thermal resistance: typ. 8.0 K/W
- Max. Temperature Tj: max. 125°C
Datasheet can be downloaded here: Moonleds MN-S3535 (PDF, may require login)
The maximum operating current is contradictory in the data sheet. There is talk of both 1000 (page 1) and 1200 mA (on page 2). In the interests of a conservative operating point, I use 1000 mA here as the basis for the maximum current, although the maximum power of 3 W clearly indicates a maximum operating current of 1000 mA.
At first glance, the Moonleds MN-S3535 looks very much like an XP-E2. The dark gray substrate is covered with a shimmering silver coating, on which the LED chip and the silicone dome sit.
The application of the orange-yellow phosphor is striking. This is not applied exactly onto the LED chip, but rather distributed in a circular pattern. This may be to avoid bluish shimmering at the sides. This sometimes occurs with LEDs in the classic lateral design.
The anode is marked with a clearly recognizable plus marking, which greatly simplifies handling, especially for manual reflowing. In the test, there were no problems with excessive sensitivity of the dome or substrate.
The footprint is standard 3535, so there are no surprises. The thermal pad is electrically insulated so that DTP boards can be used without any problems. The cathode is marked with a notch in the thermal pad.
Such significant reductions in the heat dissipation surfaces should be viewed critically; they further reduce heat dissipation, which is already severely limited, especially in the compact 3535 format. A marking on the top or a beveled corner of the anode connection is more advisable here.
The luminous area is 2.2 mm² in size. The LED chip is similar to the classic design of Cree XP-E2 and -G2 and some older Chinese LEDs. Some light shimmers through at the sides or is reflected by the phosphor protruding from the edges, which has a warmer CCT and therefore makes it much more difficult to determine the area of LES.
In the die shot, it can be seen that a chip with a classic rectangular luminous surface is used rather than a round LED chip as with newer Chinese LEDs. This has cut-outs at two corners for connecting the two bonding wires. This means that some of the luminous surface is lost; this is one of the reasons why the flip-chip design is being used more and more frequently.
Within official parameters, as far as known:
- at 1,000 mA (official maximum current): 268 lm @ 3.04 V
- Power at official maximum: 3.04 W
- Efficiency at 1,000 mA: 88.4 lm/W
- Maximum reached at 1.8 A, at this point 356 lm @ 3.23 V
- Power at maximum 5.8 W
- Efficiency at maximum 61.1 lm/W
The Moonleds MN-S3535 has a very modest performance. The efficiency is generally low; at 1 A it achieves just 268 lm. This is low even for LEDs with a small illuminated area similar to an XP-E2.
Two things are particularly critical here: the very low Tj of just 125°C and the very high thermal resistance. In combination, these two factors ensure very low overcurrent capability. From just 1.5 A, the luminous flux begins to drop by a few percent within a few seconds, which indicates a lack of heat dissipation from the chip to the substrate. This is surprising, as other manufacturers such as Luminus/Sanan or Cree enable significantly higher operating currents with similarly sized LED chips. However, this seems to be a problem for Moonleds with many LEDs; the Moonleds royal blue 3535 LED (P/N MN-S3535B450340) also has a very low maximum current of 700 mA.
It is not known which LED chips are used here. However, the chips used for the MN-S3535 series appear to be more sensitive to heat, as can be seen from the unusually low Tj. The lower the Tj (temperature junction, maximum temperature of the semiconductor), the shorter the service life at high ambient temperatures. In relation to these LEDs, this means that they should either be well cooled or operated at a very low current in order to achieve the longest possible service life. Operation above the official maximum can significantly shorten the service life. In addition, a low Tj also means a lower maximum performance - the general rule for white LEDs (with a royal blue LED chip) is that with an identical operating current and a 50 °C increase in Tj, the luminous flux drops by around 20 %. This is the main reason why the LED luminous flux measurements in LED tests have this typical curve shape.
Ten minutes of operation at 1.6 A does not damage the emitter. However, continuous operation at these current ranges is not recommended, especially as heat-related damage cannot be ruled out due to the low Tj and the high thermal resistance.
Data for 25 °C Tsp (at 85 °C the luminance values are around 13 % lower).
The luminance is not as poor as might be expected. Due to the relatively small LES, a high luminance with low operating current is possible compared to other 3535 LEDs despite the very low efficiency. As a result, this LED can be used in small lamps with small reflectors or lenses; a discipline in which this LED is best suited anyway due to its low performance and heat sensitivity.
The beam is perfect and corresponds to that of Nichia 519A or older Cree LEDs with lateral chip (e.g. XP-L HI 1st gen).A Convoy S2+ with 10509 Carclo Triple Lens is shown here, but the light image looks very similar when reflectors are used.There is no disturbing tintshift or color distortion.
The following information relates exclusively to the 6000 K version.
The spectrum corresponds virtually 1:1 to the standard illuminant D65. This represents a standard radiation distribution of a light source at 6504 K, which corresponds almost exactly to the spectrum of a cloudy sky at midday.
In addition to efficiency and “throw” (luminance), factors such as CCT, duv and CRI have recently come to play an important role in the flashlight scene. The Nichia 519A is currently (as of early 2024) the measure of all things and has established itself in lamps for users with high color rendering requirements. 519A generally have a duv of either 0 or below 0 (negative range). This means that the light has a higher red component. This ensures a subjectively very pleasant color rendering, which is also nominally reflected in the CRI standard. The desire for emitters with a negative duv is one of the reasons for the great success of the 519A.
In 2008, the ANSI/NEMA C78 377A standard was introduced for the binning of semiconductor light sources such as LEDs. Virtually all LED manufacturers use this standard to sort LEDs into color groupings and sell them accordingly.
In this ANSI standard, the center of the 3- and 5-step binnings from 5000K no longer runs exactly along the BBL (black body locus), but offset above it. This results in a duv of 0.002 to 0.0035. This is clearly recognizable in almost every data sheet for LEDs with neutral and cool white binning. The higher the target color temperature, the further away the center of color localization is from the BBL. Nominally, this means a higher proportion of green; however, it should be noted that a duv of 0 does not necessarily mean a “truer” light. Depending on the surroundings, degree of cloud cover and time of day, daylight may well have a significant green component. A duv of 0.002 or greater is typical for daylight under a cloudy sky.
The D65 standard illuminant reproduces precisely these color characteristics. As D65 comes closest to daylight without direct sunlight, a light source with a similar or identical spectrum has the more “natural” light.
This becomes clear in the spectrum above: the blue line in the spectrum shown is D65; the Moonleds MN-S3535 offers an extreme similarity here, except for the blue peak and the generally known cyan hole, which still affects virtually every commercially available white LED on the market.
The color rendering value and R1-R14 are among the highest I have ever measured for a white LED. Unfortunately, the value of at least 90 specified by the manufacturer is not achieved for R12 (82); however, R9 and Ra are extremely high and close to what is currently technically possible.
Subjectively, the light from the MN-S3535 tested here appears perfect. However, it also appears different from the light of a 519A with 5700 K (sm573 R9080). The following beam shot shows this impressively (fixed white balance 5000 K).
The 519A sm573 R9080 oversaturates red and brown colors. This is particularly visible in the beam shot with the couch and the red stool.
In general, everything appears more reddish. The light from the Moonleds appears more balanced; the actually grey-blue flower pot looks less artificial, the overall color impression is simply more “realistic”, more like daylight. Colors such as beige or light brown appear very different in places, which can be problematic with the light of the 519A sm573. Of course, the different CCT plays a certain role here, whereby the difference is significant, especially when walking in the forest and generally on surfaces in the red-brown spectrum, and is also clearly perceptible to the eye.
The Moonleds MN-S3535 tested here basically offer a D65 standard illuminant and are therefore perfectly suited as a (portable) reference light source for simulating daylight. The 519A does not do this - it offers a subjectively soft, very pleasant light, but is not comparable to daylight, as it does not reproduce the colors in the same way as daylight due to the strong saturation in the red range.
The supposedly high duv of 0.0026 does not actually play a major role here and corresponds relatively accurately to daylight. The duv of the D65 standard illuminant is 0.0032. It does not interfere at any time and, thanks to the high CCT and the special spectrum, actually makes the light appear more real without losing too much of the pleasant tint at the same time.
- Ra: 98
- R9: 97
- CCT: 7034 K
- duv: 0.0026
I have rarely tested such an ambivalent LED before. Joy and sorrow lie close together: as low as the maximum performance and the very low maximum luminous flux per emitter are, the quality of light is simply impressive.
The spectrum corresponds almost exactly to the standard light source D65 (daylight under a cloudy sky). The Ra and R1-R14 are extremely high. The beam with reflectors and TIR lenses is almost perfect, there are no disturbing color artifacts or other anomalies.
Pro
- Impressive light quality, extremely high color rendering
- Spectrum and color impression almost like standard illuminant D65 (daylight)
- Very good light image in secondary optics
Neutral
- duv above 0 (corresponds to daylight/D65)
Contra
- Very low overcurrent capability
- Very low luminous flux possible in the safe operating range
- strong drop in luminous flux already at over 1.5 A (heat sensitivity)
- no official source of supply known
Thank you for reading the test. 
