Tag Archives: 28600

Lumintop D3

The D3 is a compact thrower-style flashlight, running on an included single 21700 battery or optional 26800 battery. Features the sophisticated Anduril user interface.

  1. Introduction
  2. Manufacturer Specifications
  3. Package Details
  4. Build
  5. User Interface
  6. Circuit Measures
  7. Emitter Measures
  8. Beamshots
  9. Testing Results
  10. Runtimes
  11. Pros and Cons
  12. Overall Rating
  13. Preliminary Conclusions
  14. Acknowledgement


Upon my recent return to reviewing, I was glad to see that Lumintop is still in operation. I’ve had a number of mid-range and very good quality lights from them over the years – typically at budget prices (or close to).

Despite the low prices, their lights often have some distinctive features, showing they have an innovative side. Enter the D3, which has a number of interesting characteristics for this class. It also features the Anduril 2 user interface, which I’ve heard good things about. So I thought it would be a good light to review and compare – let’s see how it does in my testing.

Manufacturer Specifications

Note: as always, these are simply what the manufacturer provides – scroll down to see my actual testing results.

Max Output (Lumens)6,000
Min Output (Lumens)-
Max Runtime-
Max Beam Intensity (cd)91,600 cd
Max Beam Distance (m)605 m
Mode Levels-
Weight (w/o battery)138 g
Weight (with battery)-
Length135.5 mm
Head Diameter45.5 mm
Body Diameter30.5 mm

Package Details


The D3 comes in a hard cardboard box with cut-out foam to hold everything in place securely. Inside you will find the following:

  • Lumintop D3 flashlight
  • 5000mAh 21700 battery
  • 21700 Battery holder (not needed for 26800)
  • Wrist lanyard
  • USB-C Charging cable
  • 2 Spare O-rings
  • Manual

It’s a decent package, and I like seeing the charging cable (don’t need another one, but it is good to see everything you need to charge the light included). That said, I would have liked to have seen a holster or pocket clip.


From left to right: LiitoKala 21700 (5000mAh), Fenix ARB-L21-5000U 21700 (5000mAh), Sofirm IF25A, Fenix E35 v3, Convoy S21E, Imalent MS03, Armytek Wizard C2 Pro Max, Acebeam E70, Nitecore P20iX, Nitecore MH12SE, Lumintop D3, Convoy M21F.


The D3 is a bit longer and wider than most lights in this class – consistent with its larger than typical reflector, and its ability to take 26800 cells. That said, it fits very comfortably in the hand – I don’t find it too large (although I do have large hands).

There is a side-mounted electronic switch on the side of the light near the head, with a stylish electric blue switch ring. The switch features the Lumintop rabbit logo, which glows faintly green when a battery is installed and the light is ready to use (making it easy to find the switch in the dark). Feel and traverse of the switch is good, and easy to activate. Note that it wouldn’t be too hard to accidentally activate the light, so I strongly recommend you keep the light stored locked out at the tailcap when not in use. A simple twist of the tailcap will do the job, thanks to the anodized screw threads. The light can tailstand stably, even with the cut-outs for the wrist lanyard.

The knurling is an attractive square block design. While not very aggressive per se, it actually serves to enhance grip pretty well due to the space between the blocks. With the additional head fins, I’d say overall grip is decent.

Anodizing looks to be very good quality, relatively matte in finish. I didn’t notice any flaws on my sample.

One thing I really like about the light is the square-cut screw threads in the tail – always a nice touch, and I found thread action to be very good.

Inside, the light comes with a gray un-branded standard-sized 21700 flat-top battery (rated at 5000mAh according to the wrapper), and a battery holder to help make contact. Note the build is designed to accommodate 26800 cells, so a 21700 cell could easily rattle around or be displaced without the extra holder. I like seeing the standard-sized 21700 here – on many lights that use proprietary cells with integrated chargers, it can be hard to find another battery that fits and works in the light. And in a pinch, you might be able to fit in one of those 21700s with an integrated charger. So, no worries here – battery flexibility is a real strong point of this build.

Update September, 2023: the no-name brand 21700 battery included with my sample had begun to leak corrosive acid while in storage (inside the light, but locked out). If I hadn’t noticed it in time, it would likely have destroyed the flashlight circuit. This is first Li-ion battery I have seen leak like this, and is very disappointing. STRONGLY RECOMMEND Lumintop replaces these with a better quality cell.

The light includes a USB-C charging port on the body, on the opposite side from the switch (with a well-fitting rubber cover). The switch bunny flashes blue when charging, and goes solid blue once the cell is fully charged. As previously mentioned, the switch bunny glows green when ready to use – and stay green while the light is in use.


The light uses the SFN55.2 emitter, which is a relatively large 7x7mm LED made by San’an Optoelectronics in China (I don’t know much about the company – someone please leave a comment if they know more about these emitters). It is coupled with a reasonably large heavily-textured reflector. You should expect a good amount of output with a relatively throwy beam. I didn’t notice any significant tint artifacts in the beam, it seems very clean.

The bezel is flat black aluminum, with a standard level of crenelations (i.e., enough to headstand stably, and see when the light is on). The light comes with an AR-coated  mineral glass lens.

Overall, I find this to be a nice looking light with decent ergonomics and a good beam pattern. It fits well in the hand, despite its larger than typical size.

User Interface

The D3 uses the Anduril 2 user interface (UI), which is a popular open source UI with many advanced features. It has two distinct UIs mode sets: Simple and Advanced. The labels are a bit misleading, as both are fairly sophisticated – it is just that the Advanced UI has a lot of extra options not available on the scaled-down Simple UI. Advanced UI also has an option for a discrete Stepped level mode, in addition to the continuously-variable smooth Ramping mode (which is the only mode present on the Simple UI).

To switch from the default Simple UI to Advanced UI, you need to do 10 clicks from Off, with a hold on the 10th click.

It’s easier to show it the UIs than explain them in words, so here is a pic I scrapped off the internet:


It’s not as complicated as it looks (and Simple UI is quite straight-forward). But let me try to break it down for you.

From OFF:

  • Press and hold: Turns On in lowest output, in either Ramping mode or Stepped mode depending on which mode is enabled (and which UI you are in)
  • Single click: Turns on in last memorized mode used (Ramping or Stepped again)
  • Double click: Turns on to Turbo (aka the Ramping max output)
  • Triple click: Battery check (voltage read out a single time)
  • Triple click and hold: Special strobe modes, but only when in Advanced UI (remembers last strobe mode used)
  • 4 clicks: Lockout mode. In lockout mode you have different options available:
    • Press and hold: Momentary Moonlight
    • Double click and hold: Momentary Low
    • 4 clicks: Turns On in memorized output level
    • 4 clicks and hold: Turns On in the lowest level
    • 5 clicks: Turns On in Turbo
    • 10 clicks and hold: Configure the lock timeout threshold (in Advanced UI only), allowing you to pre-set the timeout time of the lock.

From ON:

  • Press and hold: Ramps up (or Steps up, depending on the mode). Ramps/steps down if you do it again.
  • Single click: Turns Off
  • Double click: Jumps to Turbo
  • Double click and hold: Ramps down (or Steps down)
  • Triple click: Switch between Ramping and Stepped modes (in Advanced UI only)
  • 4 clicks: Lockout mode (see above for options)

Mode memory:

Yes, the circuit memorizes the last constant On output level in either Ramping or Stepped modes.

Strobe/Blinking modes:

Yes, quite a few actually. The strobe/blinking modes are accessible from off with a triple click and hold in Advanced UI only. You can switch between strobe/blinking modes with 2 clicks, in the following sequence (see testing results below to see what these look like):

  • Candle mode
  • Bike flasher
  • Party strobe
  • Tactical strobe
  • Lightning mode

Low voltage warning:

Yes. In operation, the light drops in brightness gradually until shutting off when the cell is ~2.9V.

Lock-out mode:

Yes. In either Simple UI or Advanced UI, lockout is accessed by 4 clicks from on or off (repeat to unlock). The lockout mode is unusual here, as it actually enables momentary operation in the Moonlight/Lo modes. There are other lockout modes available, as explained above. As always though, I recommend you physically lock out at the light at the tailcap, if you want to guarantee no accidental activation.

Temperature check and thermal calibration mode:

This is a little complicated (and beyond the needs of most users), so I will just refer you to the diagram from the manual above.

Reviewer Comments:

This is fairly sophisticated setup – a choice of simple or advanced UI, well thought out (although of course you will never please everyone). While I can see flashaholics enjoying some of the extra items, some of them are really novelties (e.g. candle mode and lightning storm are particularly well done, but when would you practically ever use them other than as a party trick?). At the end of the day, I expect Simple UI is fine for most users. That said, the Stepped mode and bike strobe are something that many could use, so it’s nice to have them as options in the Advanced UI.

Circuit Measures

Pulse-Width Modulation (PWM):

There is no sign of PWM on any level, the circuit appears to be fully current-controlled.

That being said, my oscilloscope was able to detect high frequency noise at most levels, as depicted below. Note that these are not a concern, as they are not visible to naked the eye – the light remains flicker free in actual use.


Stepped L1:

Stepped L2:

Stepped L3:

Stepped L4:

Stepped L5:

Stepped L6:

Stepped L7:

Stepped L8:

As you can see, noise begins with L2, and the frequency remains constant at ~16 kHz, but increases in amplitude as output levels rise (which is fairly common, as more light is being emitted) – except for L4, which is noise free. Noise at these super high frequencies is impossible to detect visually, and so not a concern. I am just including for completeness, as I like to present all my findings.


Tactical Strobe:

Tactical strobe is a fairly typical 10.4 Hz, although the light spends more time in the On phase than the Off phase on each cycle.

Party Strobe:

Party strobe is a super-fast (and annoying) 25 Hz. I don’t know what kind of parties Anduril fans like to hold, but I won’t be attending any. 😉

Lightning Strobe:

I’ve shown three consecutive 10 sec cycles above, so you can a feel for the frequency and intensity of light flashes. Lightning strobe is a fairly realistic lightning simulation, with variable intensity and time between flashes.

Bike Strobe:

Bike strobe is a nice, slow 1 Hz signalling strobe – but with with 4 rapid flashes on every signal (shown in more detail below).



Beacon strobe is a slow 0.5 Hz (again with multiple flashes detected on every signal).


Candle strobe is a continuous flicker, of varying intensity (again, accurately simulating a candle).


Resting voltage <3.0V

Resting voltage >3.0V

I measured two distinct charging currents for the 21700 using the D3’s USB-C charging port; 0.14A when the cell is heavily depleted (<3.0V resting), at a jump to 1.5A once the cell is above >3.0V resting. Two-stage current charging is a common feature for integrated chargers on 21700 lights, and these are good levels. You can expect the cell to be charged rapidly.

Standby / Parasitic Drain:

I measured the standby current as 110 uA, which is negligible and not a concern (i.e., it would take many years to fully drain the cell). Regardless, I always recommend you lock the light out when not in use to prevent accidental activation (and cut any standby drain). A single twist of the tailcap will lock out this light, thanks to the anodized screw threads.

Emitter Measures

This section is a new feature of my reviews, where I directly measure key emitter characteristics in terms of colour temperature, tint, and colour rendition. Please see my Emitter Measures page to learn more about what these terms mean, and how I am measuring them.

As explained on that page, since I am using an inexpensive uncalibrated device, you can only make relative comparisons across my reviews (i.e., don’t take these numbers as absolutely accurate values, but as relatively consistent across lights in my testing).

The key measures above are the colour temperature of ~6190K, and the slight positive tint shift (+0.0080 Duv) to yellow-green at this temperature.

For CRI (Ra), I measured a combined score of 68.

I don’t have previous experience of SFN emitters, but these values seem reasonable for a typical high-output cool white emitter, and match my visual experience of this light.


All outdoor beamshots are taken on my Canon PowerShot S5 IS at f/2.7, 0.5 secs exposure, ISO 400, daylight white balance. The bend in the road is approximately 40 meters (~45 yards) from the camera. Learn more about my outdoor beamshots here (scroll down for the floody light position used in this review).

Click on any thumbnail image below to open a full size image in a new window. You can then easily compare beams by switching between tabs.

Sorry, the label for D3 is mislabelled – my sample has a SFN55.2 emitter.

As you can see above, the beam pattern is more on the throwy side, but with a good amount of spill.

Testing Results

My summary tables are generally reported in a manner consistent with the ANSI FL-1 standard for flashlight testing. In addition to the links above, please see my output measures page for more background.

All my output numbers are based on my home-made lightbox setup. As explained on that methodology page, I have devised a method for converting my lightbox relative output values to estimated lumens. My Peak Intensity/Beam Distance are directly measured with a NIST-certified Extech EA31 lightmeter.

D3 Testing Results

ModeSpec LumensEstimated Lumens @0secEstimated Lumens @30 secsBeam Intensity @0secBeam Intensity @30secsBeam Distance @30secs
L1-2.72.7---NoNo0.14 A1.5 A110 uA170 g265 g
L2-2424---No16 kHz0.14 A1.5 A110 uA170 g265 g
L3-7575---No16 kHz0.14 A1.5 A110 uA170 g265 g
L4-180180---NoNo0.14 A1.5 A110 uA170 g265 g
L5-660660---No16 kHz0.14 A1.5 A110 uA170 g265 g
L6-1,6001,600---No16 kHz0.14 A1.5 A110 uA170 g265 g
L7-2,8002,450---No16 kHz0.14 A1.5 A110 uA170 g265 g
L8 Turbo6,0004,4003,90039,000 cd37,400 cd387 mNo16 kHz0.14 A1.5 A110 uA170 g265 g
Candle-------No0.14 A1.5 A110 uA170 g265 g
Bike Strobe------1.1 HzNo0.14 A1.5 A110 uA170 g265 g
Party Strobe------25 HzNo0.14 A1.5 A110 uA170 g265 g
Tactical Strobe------10.4 HzNo0.14 A1.5 A110 uA170 g265 g
Lightning-------No0.14 A1.5 A110 uA170 g265 g
Beacon------0.53 HzNo0.14 A1.5 A110 uA170 g265 g

You can tell from above that my D3 sample comes in well below the rated ANSI FL-1 max lumen output spec – and even further below the rated beam distance.

To see full testing results for all modern lights in my testing, check out my Database page.


As always, my runtimes are done under a small cooling fan, for safety and consistency. To learn more about how to interpret runtime graphs, see my runtimes methodology page.



All runtimes above L5 show a “noisy” pattern, with fairly steep step-downs early in the runs, and then significant fluctuations in output over time at their “regulated” level. In practice, the fluctuations shown above are not noticeable (i.e., they are actually fairly slow over perceptual timescales, and unnoticeable). Below L5, the light shows a semi-regulated pattern at best, that looks a lot like direct drive.

Overall efficiency on the sub-Turbo modes is lower than the competition, with runtime duration on L7 and lower somewhat on par with the lights containing lower capacity 4000mAh cells. Despite the flashy Anduril UI, it seems the actual D3 circuit drivers are fairly basic. I would recommend Lumintop look into the improving both the regulation and the efficiency.

Note that some reviews of the early release version of this light showed differing output levels on Turbo/ramping max depending on the mode and UI used. I didn’t observe anything consistent, as shown in two different max runtimes (next to a L7 runtime) below:


That said, I did find a slightly higher output (and greater initial step-down) on my initial runtime tests when the battery was new. On successive runs, it seems to have settled into a standard pattern, as shown below (i.e., the red chart is the very first run I did, the others all came later after more testing – and more charge cycles).


I’m not quite sure what to make of the variation in initial output. But the circuit does seem to produce mildly different step-down patterns on different runs. Could be thermally-mediated, but all my runtimes are done under pretty consistent conditions (i.e., under a fan, in a climate-controlled room). Again though, you will not notice the differences by eye, given the relatively gradual shifts over the timescales involved. I’m just including the graph for completeness sake.

Finally, as mentioned above, most runtimes are very “noisy” looking – though this is not actually visible in practice, due to the time frames involved and the limits of human perception in terms output changes. To demonstrate, here is a blow-up of a 2-min portion of the run, when the changes in output were the most extreme (on the max output):


Believe it or not, you will not be able to notice those kind of outputs changes in real life. That said, this is probably the most extreme example of rapid output shifts I’ve seen to date. For example, note how this compares to the Fenix E35 v3 here.

Pros and Cons

Comes with the advanced Anduril 2, with has both simple and advanced UI options (with lots of features).Max output and throw are not as high as rated.
Very good throw for the class (but not as high as I was expecting).Output runtime efficiency is lower than other lights I've tested with this rated capacity battery.
Can support a range of 21700 cells and 26800 cells, thanks to the bundled adapter.Driver regulation is poor, with inconsistent and "noisy" shifts in output.
Good build quality and hand feel.Lacks a true Moonlight mode, but has a good lowest mode level
Good charging implementation.

Overall Rating

Preliminary Conclusions

I was more favourably inclined toward this light initially, until the runtime results came back.

Don’t get me wrong, the D3 has a lot of nice features, generally well implemented. The Anduril 2 UI is sophisticated, and includes a lot of extras that I can see flashlight enthusiasts having fun with. And the larger thermal mass means it doesn’t step-down from max output as quickly as many competing smaller lights. But the performance of the actual circuit – in terms of overall efficiency, output stability, and regulation patterns – are all disappointing in comparison to other lights in my testing.

Moreover, I find the max output rating to be a little too far off of spec – and the published throw rating completely out of sync with my actual measures. Since many consumers base their purchase decisions on the rated specs, I really can’t justify anything higher than a 3 star rating.

These performance issues are a shame, as I find the physical build to be otherwise excellent, with a lot of nice touches. The light is a little large, but that’s in keeping with its dual 28600/21700 battery support (and again, brings a thermal mass benefit).

This is my first time testing a SFN55.2 emitter, and I find it does well for the large die-size class (especially coupled with the heavily textured reflector here). I would encourage Lumintop to consider sourcing an alternate driver, as I’m afraid it is really holding back an otherwise very nice build and advanced UI. Considering the low cost of this light, it could be a real winner if coupled with a better, more efficient, current-controlled driver.

Update June, 2023: the Lumintop website currently lists a SFN60 emitter available in this model, so output may be slightly higher than the SFN55.2 sample I reviewed.

Update September, 2023: the no-name brand 21700 battery included with my sample had begun to leak corrosive acid while in storage (inside the light, but locked out). If I hadn’t noticed it in time, it would likely have destroyed the flashlight circuit. I’m lowering my rating by another half-star, due to the bundling of such a low-quality battery.


The D3 was provided for review by Lumintop. All opinions are my own however, and the light received the same rigourous and objective testing as all other lights that I have reviewed. At the time of review, this light retails for ~$65 USD (~$75 CDN).