Convoy S2+ 365 nm LED torch: a cheap alternative to MTE UV torches?  

For a few years, the MTE 301 LED torch has been the reference for 365 nm LED torches. Among other uses, it has been regularly employed for UV imaging by several photographers specializing in this field. This model was replaced by the MTE 303, which uses a 3W Nichia LED with a 50% bandwidth of approximately 10 nm and a peak around 365 nm (according to Nichia specifications). Both torch models are reported to work very well. Their main problem, however, is a high price.

While writing the original version of this page, I was mistakenly under the impression that the MTE 301 uses a Nichia LED of similar or slightly lower power as the MTE 303 (according to users' impressions). However, the MTE 301 web page (near its bottom) does report that the MTE 301 uses a LED with a typical emission of only 450 mW. The MTE 303 is rated at 1030 mW. This explains the initial mystery of my finding that the Convoy S2+ 365 nm emits quite more UV power than the MTE 301. In view of this fact, my findings are completely as expected, and my measuring instruments reliable.

A thread in a forum discussed the positive experience of a user with a torch of similar specifications, branded Convoy and with an S2+ model denomination, apparently made in China and available through several online retailers, including eBay sellers. Be careful you order the right Convoy S2+: the Convoy S2+ is also sold as an ordinary white LED torch. What you want (since you are reading this page), is instead the Convoy S2+ with 365 nm Nichia LED. In the rest of this page, I refer to the 365 nm model simply as Convoy S2+.

The price of these torches is roughly one-tenth of the normal retail prices of the MTE 301 and MTE 303. An obvious inconsistency is that the Nichia NVSU233A U365 LED, used in the MTE 303, and initially assumed to be used also in the Convoy torches (more on this below) usually retails in the EU for about three times the price of one of these Convoy torches. The question immediately arises whether the Convoy torches really perform like an MTE 301 or 303. If the answer is negative, this helps to explain the difference in price, and requires me to debunk the claims of the sellers. If the answer is positive, on the other hand, this means that the prices we have been paying for the MTE torches and Nichia LEDs are grossly overinflated (at least today, although not necessarily 3-4 years ago), compared to the normal profit margins of Chinese online sellers. A meaningful discussion also requires a head-to-head, quantitative comparison of the MTE and Convoy torches in conditions that reflect their practical use.

Physical appearance

MTE 301 and Convoy S2+ with 365 nm Nichia LED
Leftmost: MTE 301. Remaining three torches: Convoy S2+ with 365 nm Nichia LED.

The MTE 301 has a wider head, and therefore a larger reflecting cup and front window. The MTE also has a hexagonal anti-roll ring and three protruding machined rings at the base of the head. Both features increase the weight of the barrel and its efficiency as a heat sink, which facilitates cooling of the 3W LED. The Convoy has no such features.

The MTE 301 is also physically larger, longer and heavier than the Convoy S2+. The appearance of both models is typical of 3W torches, although the Convoy S2+ is at the low end of the range of these torches in terms of size of the reflector. In fact, the same torch barrels are or were used for a number of other LED torch models besides these UV models.

The Nichia NVSU233A U365 LED of the MTE 301 uses a ceramic chip carrier with a flat glass window. The LED in the Convoy S2+ is instead mounted on a smaller ceramic base and embedded in a roughly hemispheric plastic lens. I cannot see any model number on this LED, but the appearance corresponds to the NCSU276A U365 model, rated by Nichia at 780 mW of emitted power. Nichia makes other models rated at higher emission power in similar packages, but their illustrations show a black ceramic base, not white like the NCSU276A U365 illustrations and the Convoy S2+ LEDs. The size and orientation of the LED die on the base in the C2+ also conform to NCSU276A U365 illustrations.

The NCSU276A U365 retails from some European mail order sources at a higher price than the Convoy S2+ 365 nm torch including shipment from China.


Both torches use a single 18650 battery. For the present tests, I used two new-out-of-the-package, identical Panasonic CGR18650E (same production batch) simultaneously charged in the same charger, then let to cool down overnight. I did one set of tests, then let the torches and batteries cool down for a few hours, swapped the batteries, and repeated the temperature and current tests. Results remained the same, showing that the two batteries perform in the same way.


To test whether normal use results in a higher operating temperature of the Convoy torches, both models were kept at room temperature, and the external temperature of the torch barrel was measured with an IR thermometer as close as possible to the internal position of the LED. Both torches were placed vertically on a bench, with the head uppermost. They were turned on for 5 minutes, then their temperature was measured again in the same places. The external temperature is not a foolproof indication, since a lower temperature may indicate any combination of the following causes:

  • The torch may have a poorer thermal conductivity between LED and barrel exterior, which in turn means that the LED inside is actually hotter.
  • The torch may drive the LED with a weaker current, so the power used by the LED is lower.
  • The torch is better at dissipating the LED heat. This is of course the most desirable cause.

While the second of the above points is easily tested (see below), the remaining two are not. In particular, I was unable to measure the temperature of the LED chip carrier in the Convoy torch (see also below).

At the end of the test, I measured the temperature by pointing the thermometer against the LED through the torch window. The thermometer used for this test does not allow a pinpoint measurement, so this temperature likely is an average of the LED and surrounding area of the reflector.


  • MTE 301 temperature before startup: 25 °C, 5 minutes after startup: 29 °C. Temperature through the window: 30 °C.
  • Convoy S2+ temperature before startup: 25 °C, 5 minutes after startup: 35 °C. Temperature through the window: 36 °C.
  • Qualitatively, the barrel of the Convoy S2+ felt slightly warm at the end of the test, while the MTE 301 did not feel warmer to the hand than at the beginning of the test.
  • The barrel of either torch did not become warmer in correspondence of the battery compartment. Battery overheating is therefore not a problem in normal operation.

Current and voltage

I measured the current on torches at ambient temperature, by removing the battery compartment cap and shorting the rear of the battery to the torch barrel with a digital multimeter in mA DC mode. The constant-current drive circuits in both torches are expected to compensate for the voltage drop caused by the internal resistance of the current meter. I also measured the voltage at the battery poles with open circuit and with LED turned on.

Since I obtained meaningless readings in an earlier run of these tests, I recharged the batteries before testing.


  • Open circuit battery voltage: 4.23 V
  • Battery voltage and current with LED on: 4.06V 0.69A

Assuming the LED driver uses a negligible power, this means the LED is fed with 2.8W.

The NCSU276A 365 nm is rated at a typical voltage of 3.8 V (max 4.4 V) at a 0.5 A current (max 0.7 A), which means a total power of 1.9 W (typical) to 3 W (max). The Nichia specifications don't explicitly say so, but I assume that from this power we must subtract the energy radiated as UV (0.78 W) to obtain the energy dissipation that heats up the LED. The LED in the Convoy S2+ is therefore, apparently, driven relatively close to its maximum power rating, but not above it. It also needs to dissipate about 2 W of heat.

Illumination intensity

In addition to the LED chip emission intensity, there are additional factors that affect illumination intensity, mainly:

  • Efficiency of the torch reflector - I do not expect substantial differences in this respect, since an aluminized or chrome-plated surface (unless coated with UV-absorbing film) reflects UVA with a high efficiency. However, I have no way to measure the actual UV reflectance of the torch reflector.
  • Geometric radiation pattern - This is affected by both LED and reflector. For LEDs, see for example Moreno & Sun, 2008: Modeling the radiation pattern of LEDs. Optics Express 16 (3). The reflector of the MTE 301 can be entirely removed, leaving the LED chip carrier naked. This could arguably be a better way to compare the LEDs, but unscrewing the head of the Convoy S2+ disconnects the LED from the torch barrel (the LED PC board is mounted in the head with a retaining ring). A second drawback of comparing naked LEDs is that the chip carrier in the MTE 301 has a flat and thin glass window, while the Convoy S2+ has a thick plastic convex lens molded around the LED die. As a result, the two LEDs most likely have different geometric radiation patterns.
  • UV transparency of torch window - While the window of the MTE 301 is easily removed by unscrewing its retaining ring from the front, the window of the Convoy S2+ cannot be removed without some additional disassembly, for the same reason explained at the preceding point. I just assume that the latter torch has a UV-transparent window. It does look transparent when imaged through an Asahi Spectra ZRR0340 UV-pass filter.
  • Distance from illuminated surface, and measurement position in the illuminated area.

A meaningful comparison of the illumination intensity of two torches, in the context of close-range illumination for digital imaging, requires that the intensity be compared between two illuminated areas of the same diameter. For an initial qualitative test, I laid both torches in front of a white A3 paper sheet (which strongly fluoresces in blue when illuminated with UV) and adjusted the distance of each torch from the target to obtain an illuminated circle with a diameter of 20 cm. Assuming that the illumination wavelengths are similar between the two torches (this was verified, in a separate test, see below), the fluorescence strength is assumed to be proportional to the illumination strength. I imaged both illuminated areas simultaneously with an ordinary digital camera and lens.

The MTE 301 torch used for these tests had its transparent window replaced with a U-340 filter to cut the VIS emission of the LED, which may alter the results when imaging the results of UV-stimulated fluorescence. For this test, I removed the filter.

MTE 301 and Convoy S2+ with 365 nm Nichia LED, illumination circles
Illumination circles with diameters of 20 cm. Left: MTE 301. Right: Convoy S2+ with 365 nm Nichia LED.

Both torches produce a central hotspot surrounded by a more evenly illuminated circle. It is visually evident that the Convoy S2+ produces a higher fluorescence intensity both in the central hotspot and the surrounding area. The central hotspot in the Convoy S2+ appears larger in the above picture, but this may be a combination of larger size and higher intensity. The hotspots of both torches are overexposed in the picture, in order to show the rest of the illumination circles.

Tenmars TM-213
Tenmars TM-213 UVA and UVB meter.

I then used a Tenmars TM-213 UV meter to take quantitative readings in the illumination circles. This meter has a flat, circular plastic diffuser embossed with an irregular pattern (functional as cosine corrector, to reduce biases in readings caused by oblique illumination) with a diameter of 20 mm covering the sensor. A 10 by 10 mm square light blue filter is visible under the diffuser (most likely a NIR-cut filter). There has to be also a UV-pass, VIS-cut filter out of sight, covering the sensor. According to specifications, the peak sensitivity of this meter is at 365 nm. This suggests that sensitivity is not completely flat across the UV, at least in part because of the VIS-cut filter. This meter is quite insensitive to VIS illumination (except daylight and certain fluorescent sources, which contain also UV).

Both torches overloaded the meter (its max reading is 20 mW/cm2), so I repeated the test after moving the torches farther from the target to obtain illumination circles with a diameter of 40 cm. I took measurements at the center of each illumination circle, 1/4, 1/2 and 3/4 along the radius, and as closest as possible to the edge of the circle but still illuminating the whole sensor diffusing window. The three tested Convoy S2+ torches gave very similar readings. The table below displays results for only one torch.

Results for the MTE 301 are shown both with and without the U-340 window.

UV meter readings:

Torch center 1/4 way out 1/2 way out 3/4 way out near edge
MTE 301 without window 9.96 mW/cm2 0.18 mW/cm2 0.13 mW/cm2 0.10 mW/cm2 0.08 mW/cm2
MTE 301 with U-340 window 8.38 mW/cm2 0.15 mW/cm2 0.09 mW/cm2 0.08 mW/cm2 0.07 mW/cm2
Convoy S2+ 365 nm 21.61 mW/cm2 5.02 mW/cm2 0.24 mW/cm2 0.19 mW/cm2 0.10 mW/cm2

The same data in graphic format:

Distribution of power along diameter of illuminated circle, linear (left) and log (right) diagram.
green: MTE 301 with U-340 window
red: MTE 301 without window
blue: Convoy S2+ 365 nm

The Convoy S2+ clearly emits more radiation than the MTE 301. The difference is one stop or more, except near the edge of the illumination circle, which is more diffuse in the Convoy S2+ than in the MTE 301.

Geometry and uniformity of the field of illumination:

Both torches display an evident central hotspot.

A uniform field of illumination without hotspots is better for imaging small subjects placed at a short distance from the torch. It is possible to homogenize the field of illumination by placing a UV-transparent diffuser between torch and subject. Naturally, with these torches, a diffuser substantially reduces the UV illumination intensity at the center of the illuminated circle. Distributing the central hot spot over a larger area should not be regarded as a loss, since this results in the same energy being distributed across a larger area. However, diffusers also introduce real losses of energy, e.g., a part of the energy is absorbed, for instance by being reflected multiple times between diffuser and reflector, or within the diffuser itself. Even the best diffuser introduces a small loss at each refraction or reflection. A microlens array limits the number of refractions and reflections, and is generally more efficient as a diffuser than the traditional sanded or opal diffusers. Both traditional diffusers introduce significant overall losses.

Emission spectrum

The peak wavelength of the LED emission spectrum is the most critical specification of these torches. Most LED manufacturers sort their products into bins, based primarily on the emitted wavelength, and often also on emission intensity. Nichia NVSU 233A LEDs are sold with different specified wavelengths (365, 375, 385, 395 and 405 nm), which may represent different sorting bins and/or different manufacturing processes. LEDs from the 365 nm ±4 nm bin, of course, are the best and sell at a premium price. LEDs from the other bins are sold at lower prices. This results in a high incentive for torch manufacturers to use low-cost, out-of-specifications LEDs. Many Chinese sellers on eBay advertise "UV torches" that actually emit at 400 or 405 nm, i.e. in the VIS range instead of UV. Some torches, advertised on eBay as 365 nm, may actually emit at 380 to 390 nm. This makes a substantial difference when using these LEDs as illumination sources in digital UV imaging.

I recorded the emission spectrum of these torches with a Lasertack SKU PD-01094 USB spectrometer, with a rated range of 200 to 1,200 nm and a rated precision of 2 nm. This spectrometer is capable of sub-nm actual resolution, but the data was reduced to 1 nm resolution by interpolation, which causes a little aliasing (waviness) in the charts. Data was collected by the ASEQ Instruments 1.7 spectrometer software and normalized and displayed in Excel.

The peak emission wavelength of LEDs, including UV LEDs, is known to increase as a function of LED die temperature (e.g., Reynolds et al., 1991: Temperature dependence of LED and its theoretical effect on pulse oximetry. British Journal of Anaesthesia 67 (5): 638-643). A temperature increase of 50°C may result in a 5-7 nm increase in wavelength of the emission peak of VIS LEDs. I have no data for UV LEDs, but considering that in power LEDs the operating junction temperature might exceed 100°C, there might be a "red shift" of about 10 nm once the equilibrium temperature is reached. The spectra shown below (normalized for intensity) were recorded within seconds of turning on the torches.


Emission spectra, cropped to show only an interval between 350 and 400 nm
Series 1: MTE 301 with U-340 window
Series 2: MTE 301 without window
Series 3, 4, 5: Convoy S2+ 365 nm

The emission spectra are shown above. I tested three specimens of the Convoy S2+, purchased from two different sellers. The Series 3 and Series 5 data overlap completely and Series 3 is therefore not visible.

Torch peak wavelength (nm)
MTE 301 without window 367
MTE 301 with U-340 window 366
Convoy S2+ 365 nm 368
Convoy S2+ 365 nm 369
Convoy S2+ 365 nm 368

The difference in peak wavelengths between MTE 301 and Convoy S2+ is only 1-2 nm, which is not significant in practical use. The U-340 window shifts the MTE 301 emission peak by 1 nm toward shorter wavelengths, which is also not practically important. A more important effect of the U-340 is that it significantly cuts the "tail" of the MTE 301 emission at longer wavelengths. No doubt it would have the same effect on the Convoy S2+. As shown above, the reduction in emitted energy introduced by the U-340 is minimal, and likely caused mostly by the uncoated surfaces of this filter.

Summary of results

The Convoy S2+ 365 nm emits significantly more UV than the MTE 301 (a difference of about one stop). The MTE 301 was replaced by the MTE 303 model, which is not part of the present test and may or may not perform better than the original MTE 301. Both the MTE 301 and MTE 303 sell/sold for a much higher price than the Convoy S2+ 365 nm. Wavelength peaks of the tested Convoy S2+ 365 nm torches are 1-2 nm longer than the tested MTE 301.

The Convoy S2+ 365 nm does not appear to overload its LED in terms of current, but the possibility remains that it might exceed the thermal dissipation of the LED mount and torch barrel, which in turn might impact its durability. The Convoy S2+ barrel gets warmer than the MTE 301 after 5 minutes of operation, which could result in overheating if the torch is left on for a long time, especially in a hot environment.

web counter