2024-01-20
Chinese LEDs are still a great mystery. Often, neither the manufacturer nor other detailed information is available. In this case, however, things are different. Here I am testing an LED that is similar in design and application areas to the Nichia 519A and Samsung LH351D and offers a lot of potential.
The LED tested here was provided to me free of charge by Sofirn along with further information and the data sheet. There was no influence on the part of Sofirn. I have received permission to make the data sheet available here in the test.
Most of the information in this test applies to all variants. If there are differences between high CRI and low CRI types (primarily in terms of power and luminance), these will be mentioned separately in the test.
Tj 85 °C, If 1,050 mA
- Type: single die
- Bin: TB (min. 370 lm / max. 390 lm)
- Color group: 3B (typ 5000 K)
- CRI: min 90
- Rated voltage: typ 2.8 (min 2.7 / max 2.9) V
- Max. Forward current: 3,000 mA
- Max. Peak current: — mA
- Viewing angle: 120 °
- Thermal resistance: typ 3.0 K/W
- Max. Temperature Tj: max. 150°C
Datasheet can be downloaded here: Datasheet (newest version, LatticePower)
The LatticePower TN3535 is an LED with the familiar XP footprint. With its design of white silicone around the yellow luminous surface and the dark gray substrate, it is relatively similar to the Lumileds HL2X, but has a significantly larger luminous surface.
An interesting feature is the shiny glow above the phosphor (which makes the yellow layer almost completely disappear) when the LED is viewed from the side. This is not very common for LEDs of this type and is probably intended to prevent a color shift with increasing beam angle, which occurs more frequently with this type of LED. Presumably two different types of silicone with different refractive indices were applied over the phosphor to achieve this effect.
Thanks to the symmetrical design and the standard footprint, accessories for XP or 3535 (3.45 x 3.45 mm) can be used freely. Centering aids produced with lathes can also be used.
The footprint offers no surprises. The thermal pad is electrically insulated, which means that DTP boards with direct heat transfer can be used without any problems.
The LES is very large and appears almost round thanks to the silicone dome. As with most current emitters, it is connected via the underside (flip chip design).
The small speckles on the phosphor correspond to those of the HL2X or the Nichia 519A and should not affect the beam. In the center of the LES, the speckles on the phosphor appear blurred, as if the silicone is slightly deformed.
The luminous area is 8.4 mm² in size. There are no significant differences between the variants. As some light is also emitted from the side, it is difficult to determine the illuminated area using the surface alone.
Dedoming is not easily possible. In one case (LED shown in the previous picture), the dome could be separated from the substrate extremely easily, which happened unintentionally during normal handling. In this case the phosphor layer will be ripped off from the LED chip. Caution is therefore required here; this effect appears to be even more pronounced when heat is applied externally or immediately after high-current operation and at a corresponding temperature inside the LED.
Within official parameters, as far as these are known (90 CRI / 70 CRI 1A / 70 CRI 1B) :
- at 3000 mA (official maximum current): 1050 lm @ 3.19 V / 1391 lm @ 3.18 V / 1433 lm @ 3.24 V
- Power at official maximum: 9.56 W / 9.53 W / 9.71 W
- Efficiency at 3000 mA: 110 lm/W / 146 lm/W / 148 lm/W
Overcurrent (90 CRI / 70 CRI 1A / 70 CRI 1B):
- Maximum reached at 9.4 A, at this point 1971 lm @ 3.82 V (90 CRI)
- Maximum reached at 11.4 A, at this point 3084 lm @ 4.11 V (70 CRI 1A)
- Maximum reached at 11.0 A, at this point 3174 lm @ 4.21 V (70 CRI 1B)
- Power at maximum 35.9 W / 46.8 W / 46.3 W
- Efficiency at maximum 54.8 lm/W / 65.9 lm/W / 68.6 lm/W
The maximum achievable performance is very good for an LED in XP format (3535 or XP footprint). Above a certain level, the limited heat dissipation surface of the XP footprint is limiting. The wide spread of the maximum possible luminous flux is striking. The 90 CRI variant performs worse in this discipline than the low CRI variants. Due to the limited area of the thermal pad in the 3535 footprint, more than 11 A can hardly be achieved - experience has shown that the power limit here is 50 W ± 5 W.
This is exactly what the two LEDs with 70 CRI achieve. Not only do they achieve a maximum current of around 11 A, they are also the first 3535 LEDs with an LED chip that I have tested to achieve more than 3000 lumens! Very impressive, a few years ago this would have been unthinkable in this small footprint. The high achievable current indicates a lower thermal resistance than specified.
Unfortunately, the Vf varies significantly between the tested variants. While the high CRI variant has a reasonably normal Vf, the low CRI variants are different: a Vf of over 4 V at maximum current is no longer up to date. The different characteristic curves of the two variants are also striking. The high CRI variant has a significantly slower increasing Vf. Either there are significant differences in the yield or in the LED chips used or the emitters tested here have simply been binned differently.
Operation in direct-drive (DD) mode, where the LED is connected directly to the power source such as a Li battery, is therefore only not recommended for the high CRI variant, at least for 21700s with a good voltage range, which can reach up to 4 V under load at 10 A. The variants with low color rendering, on the other hand, can be operated in direct drive without any problems if required, provided that the Vf is checked beforehand at high operating current.
A comparison with similar LEDs in a similar performance and color rendering class shows the good performance of the TN3535. It is more efficient than the new XP-G4 in 90 CRI and clearly outperforms the Lumileds HL2X. The latter has a lower luminous flux anyway due to the smaller illuminated area, and the Vf and thermal resistance are also significantly higher. The 519A is less efficient, but has an extremely low Vf. All in all, the performance of the TN3535 is good and definitely competitive with established manufacturers. The 70 CRI variants naturally have a massively higher efficiency, so a comparison with 90 CRI emitters is hardly meaningful here.
Values at 25 °C Tsp, at 85 °C Tsp values are 13 % lower
The luminance is low due to the large illuminated area. This applies to all variants. These LEDs are not suitable for throwers; they are more suitable for use in lamps with a flood beam characteristic. The luminance corresponds approximately to that of the 519A or XP-G3/G4.
The light pattern on whitewall is quite usable, although the colored rings around the spot are visible at least on the whitewall and can be quite annoying. This effect is less pronounced with OP reflectors, and in nature this color inconsistency is no longer noticeable. With a suitable focus in the reflector and appropriate calculation of the reflector, it should be possible to significantly reduce this effect. The TN3535 is best used in textured (beaded) TIR lenses which ensures a fully mixed light.
5000 K 90 CRI
6500 K 70 CRI (color kit 1A)
6500 K 70 CRI (color kit 1B)
The samples tested here are green. Although they are not as green as most other China LEDs, they are clearly above duv, which in turn affects the subjective perception of color saturation and color rendering.
This was already to be feared due to the order code with color grouping 3B, and is only slightly better than the tint of the previously tested Lumileds HL2X. Should a color grouping in the negative duv be available in larger quantities, these LEDs could be very good and represent an alternative to the Nichia 519A, which is sometimes difficult to obtain, although the latter still offers a much nicer light image and currently represents the ultimate in light image and light quality.
The 70 CRI variants are also both above the BBL. Tint 1A (position below BBL according to datasheet) is quite usable here, whereby the CCT is also above the specified value and the duv is well above 0. Due to the higher red component, 1A performs slightly better here in the CRI/Ra and R9 range. The variant in tint 1B, on the other hand, is well above BBL and appears slightly greenish. This effect decreases slightly with increasing current, although the green always remains recognizable.
(90 CRI / 70 CRI color 1A / 70 CRI color 1 B)
- Ra: 94 / 73 / 72
- R9: 71 / -28 / -37
- CCT: 6194 K / 7169 K / 7057 K
- duv: 0.0055 / 0.0014 / 0.0042
The LatticePower TN3535 is fundamentally similar to the Nichia 519A and Samsung LH351D. The color rendering corresponds to the manufacturer’s specifications, the efficiency of the 90 CRI variant is very good even with increasing operating current and even surpasses the new Cree XP-G4. The 70 CRI version even achieves over 3000 lm in the 3535 footprint.
The Vf fluctuates significantly. It is ok for 90 CRI, but not for 70 CRI. Over 4 V at max. current is no longer up to date. The light image in reflectors could be better, the clearly visible green cast is still a problem, except in color group 1A. At least if the latter point is also remedied in the 90 CRI variant by selecting suitable color groupings and special attention is paid to a homogeneous beam / light pattern on the whitewall, these LEDs are very suitable for use in flashlights.
Pro
- very high efficiency (70 CRI)
- very high maximum current possible
- very high light flux in 3535 footprint (70 CRI)
- variant with high color rendition available
Neutral
- low luminance, not suitable for throwy lights
- light pattern could be better
- Vf could be lower (70 CRI)
- fluctuating electrical characteristics between both variants
Contra
- greenish tint
- no official source known
Thank you for reading the test. 
