Emisar D18

This is the Emisar D18, produced by Hank Wang of Intl-Outdoor. Also available from Mtn Electronics.

Disclaimer: I’m not an impartial reviewer. I helped make this, so I’m biased.

Official Specs

From Emisar’s site:

• CNC machined from aircraft-grade aluminium
• High Efficiency optics
• Driven direct by FET, from 5A and below linear constant regulation
• Anti-reflective coated glass lens
• Unihead construction
• High lumen output and low moon output
• Ramping interface
• Electronic soft touch switch (ALPS)
• Beryllium-Copper springs with 45% IACS superior to stainless steel alloys used for springs with only 2% IACS.
• Temperature step-down
• Waterproof and dustproof to IP67 standard (up to 1 meter)
• Powered by a high drain 3×18650 cell (button top only)
• Dimensions: 102mm(length) x 58mm(head) x 46mm(body)

Output:

• SST-20 6500K: 14000lm / 65,000cd
• SST-20 5000K: 14000lm / 65,000cd
• SST-20 4000K/3000K 95CRI: 10000lm / 45,000cd

Price: $99

Measurements

Manufacturer specs are nice, but more direct measurements are nicer.

Disclaimer: These measurements won’t quite match any of the regular versions of the D18. Hank made a really special version for me which has four different tints mixed together, so the performance is halfway between the warm white and cool white versions.

However, the values should still be in roughly the right ballpark. Just make the numbers bigger for the 5000K and 6500K versions, or make the numbers a bit smaller for 3000K and 4000K versions.

Also note: My lux meter seems to measure throw a bit lower than the numbers other people get.

Output:

• Beam type: Spot, corona, and spill
• Power: Oh geez… a lot. Like 50 A or potentially up to 200W. I can’t measure that much.
• LEDs: SST-20 3×3000K + 6×4000K + 6×5000K + 3×6500K
  • Lux: TBD kcd / TBD m
  • Min: 0.1 lm
  • Max: TBD (~12000?) lm (measured at start, not at 30s) (used 3×35E cells @ 4.18V)

Turbo will be brighter with higher-amp cells. I used Samsung INR18650-35E cells, which are good for mAh capacity but not very good for maximum turbo brightness.

Modes in the default stepped ramp:

• 1 (25/150): 13 lm
• 2 (37/150): 44 lm
• 3 (50/150): 125 lm (1×7135)
• 4 (62/150): 234 lm
• 5 (75/150): 495 lm
• 6 (87/150): 924 lm
• 7 (100/150): 1734 lm (14×7135)
• Turbo (150/150): TBD lm

The 5000K and 6500K models should have higher output, while the 3000K and 4000K models should have lower output. This particular light is halfway between the two.

Dimensions:

• Length: 102.5 mm
• Bezel diameter: 58 mm
• Body diameter: 46 mm
• Clip groove diameter: 45 mm
• Clip groove width: 3.0 mm
• Tail diameter: 48.5 mm
• Driver: 44.2 mm diameter, 2.2 mm thick PCB, ~33 mm x ~39 mm space for components.
• Maximum cell length: 67 mm
  (protected cells probably won’t fit)

Host / Build:

• Weight: ~342g without the batteries, according to toobadorz.
• Switch type: Momentary / electronic side switch
• Body material: Hard anodized aluminum: black or sand
• Lens: Glass with AR coating (pic below)
• Anti-roll shaping
• Tail stands very well
• Premium BeCu (Beryllium Copper) springs capable of high current without sagging
• Optic: custom, 18 pieces
• Driver held in place by screws
• No glue

Electrical:

• Operating voltage: 2.8V to 4.35V
• Battery: One to three 18650 cells, button-top, 65 to 67 mm
• Parasitic drain: ~0.04 mA (25 years to drain 3×3000mAh cells)
• Low-voltage protection: Yes
• Thermal protection: Yes
• Power can be locked out by loosening the body tube, or by entering a lock-out code on the button
• Reverse polarity protection (my DMM measured 0.00A with the leads connected backward, so RPP seems to work)

Interface

The UI is Anduril. In addition to that, it has a shortcut to thermal config on ten-clicks-from-off. The default brightness after a battery change is the highest regulated mode, or about 4.9A / ~2000 lm.

The full manual is here, though that link is a generic version and not specific to the D18. It’s pretty long, so I’m not pasting the entire thing into this review.

The short version is:

• Click to turn the light on or off.
• Hold the button to change brightness.
• Hold from off for moon.
• Double click for high or turbo.
• 3 clicks while on: Toggle between smooth and stepped ramp.
• 3 clicks from off: Battery check and some other modes (sunset, beacon, temperature check)
• 3 clicks from off, but hold 3rd click: Blinky modes (candle, biking, party strobe, tactical strobe, lightning storm)
• 4 clicks from off: Lockout (and momentary moon/low mode)
• 5 clicks from off: Momentary / signalling mode (uses memorized level)
• 6 clicks from off: Muggle mode (safer for kids and careless people)
• 10 clicks from off: Thermal configuration

This is summarized in the UI diagram:

Default config values:

• Smooth ramp: 1/150 to 100/150
• Stepped ramp: 25/150 to 100/150 in 7 steps (levels 25, 37, 50, 62, 75, 87, 100)
• Ramp shape:
  • 1 to 50: 1×7135 modes (0.1 to ~130 lm, regulated)
  • 51 to 100: 14×7135 modes (~130 to ~2000 lm, regulated)
  • 101 to 149: FET + 14×7135 modes (~2000 to ~13500 lm, hybrid)
  • 150: FET only (~14000 lm turbo, direct drive)
• Thermal ceiling: 45 C (but the sensor needs calibration before this will work correctly)

Packaging

The packaging is Emisar’s usual box. It’s a cardboard box with generous amounts of foam inside to keep the light safe. The box won’t be very difficult to damage, but you would have to throw it around pretty hard to cause any damage to the light inside. Plus, the light ships inside of its holster pouch, which gives it additional protection.

Accessories include a lanyard and a couple spare O rings.

The included pouch is nice:

Comparison to other lights

The Emisar family so far: (left to right: D4, D4S, D18, D1S, D1)

Other 3×18650 lights: (left to right: FF ROT66, Emisar D18, Sofirn SP36)

BLF Q8, Emisar D18, Emisar D4S:

Noctigon Meteor M43 vs Emisar D18

Host, Components, and Build Quality

Overall build quality is very high. Mine seems flawless aside from a bit of dust and lint which got stuck to it during use.

The heat fins at the front are rounded just enough to not be sharp.

The knurling is very, very grippy. One might even call it “gnarly knurling”…

Let’s open it up a bit to see the parts better.

Inside the body tube, where the batteries go:

The threads are square cut at both head and tail.

The bezel is a nice silvery steel, to protect it from damage and look good at the same time. The optics are 18 individual cups with a PCB fitted over them to keep the optics aligned and protected.

The button is recessed to avoid accidental activation.

The lens has AR coating. I’m not sure what each type of coating is, exactly, but the D18 has the reddish coating on the left:

The tailcap is not required to operate the light. It’s mostly cosmetic, and provides waterproofing.

Detail inside the tailcap:

The tail pcb is solid copper, to minimize resistance.

The tail PCB itself is just a slab of copper with some springs on it, and screws to hold it in place. Just as it should be.

The lanyard hole has rounded edges to avoid cutting the lanyard itself, though the area cut out around it is still a bit sharp.

Modding / Upgrade Options

Most mods are going to involve taking the light apart. Here’s a bit of detail about how to do that.

The driver is held in by screws, but the screws are covered by a reverse polarity protection ring. We need to remove that ring before accessing the driver.

I found the butt of a nail was the easiest tool for popping the RPP ring out.

Without the ring, the screws are accessible.

Taking apart the front end of the light… The optics are kept in alignment by a thin PCB:

The optics themselves are individual pieces, a clear optic with a black shroud holding the optic in position:

You’ll probably want to remove the inner optics before trying to disconnect the LED wires.

When accessing the driver, be careful not to damage the switch wires. Flip the driver over in place, without pulling on the thin wires connecting it to the switch.

Emisar always has really clean driver circuits. I don’t know how they keep the board so clean.

The MCPCB shelf is quite thick. This makes for good thermal performance.

As for modifications people might want to do, here are some ideas:

• Add aux LEDs or button LEDs. The driver already has pads to connect aux LEDs to, and the firmware has support for it. It just needs the aux LEDs themselves wired up.

• Add a minus green filter to the lens, to make the tint more rosy.

• Add some diffusion material to the lens, to make the beam more floody.

• Add a diffuser cap to make the light work as a lantern.

• Flash different firmware. Other UIs are available in the repository, and are relatively easy to make. Or perhaps you’d like to simply customize something about the stock firmware.

Beamshots

I’m no good at beam shots, but here’s the best I was able to get without whiting out the center:

I also made an attempt at doing a HDR beamshot from several exposures, but it’s not really much different:

Output / Runtime

Output varies from about 0.1 lm to about 14000 lm. There are 150 total steps in a perceptually linear ramp, so virtually any brightness level is available. The ramp takes about 2.5s to go from one end to the other, so adjustments are quick but not very precise. For precise levels, there is also a stepped ramp mode.

The D18 uses PWM to control brightness. This runs at 15.6 kHz though, so it’s not visible to a human eye. It also does not turn off between pulses in most modes, so it doesn’t look at all like it has PWM. The behavior is:

• From moon to the highest 1×7135 level (ramp 50/150): pulses quickly between 0 and ~130 lm
• From ramp level 51 to 100: oscillates between ~130 and ~2000 lm
• From ramp level 101 to 150: oscillates between ~2000 and ~14000 lm

Each level up to 100 should maintain steady brightness over time until the battery is low. Levels above 100 will sag as the battery voltage drops, and the shape of that curve depends on the battery type.

Since people always want to know the lumen output, power use, and runtimes for every mode in the ramp… here is a very rough estimation of what to expect. These are calculated by a script, not actually measured from a real light. Hopefully it at least gives people some idea what to expect, but if you really care about runtimes and lumen levels, you should measure a light directly.

1: 0.10 lm: 1.70 mA, 8.46 months
2: 0.14 lm: 1.81 mA, 7.96 months
3: 0.18 lm: 1.94 mA, 7.40 months
4: 0.24 lm: 2.11 mA, 6.81 months
5: 0.31 lm: 2.32 mA, 6.19 months
6: 0.40 lm: 2.58 mA, 5.57 months
7: 0.50 lm: 2.89 mA, 4.97 months
8: 0.63 lm: 3.27 mA, 4.40 months
9: 0.78 lm: 3.71 mA, 3.87 months
10: 0.96 lm: 4.24 mA, 3.39 months
11: 1.17 lm: 4.86 mA, 2.96 months
12: 1.42 lm: 5.59 mA, 2.57 months
13: 1.70 lm: 6.43 mA, 2.23 months
14: 2.03 lm: 7.41 mA, 59.05 days
15: 2.41 lm: 8.53 mA, 51.29 days
16: 2.85 lm: 9.81 mA, 44.59 days
17: 3.34 lm: 11.3 mA, 38.82 days
18: 3.90 lm: 12.9 mA, 33.84 days
19: 4.53 lm: 14.8 mA, 29.57 days
20: 5.25 lm: 16.9 mA, 25.88 days
21: 6.05 lm: 19.3 mA, 22.71 days
22: 6.94 lm: 21.9 mA, 19.97 days
23: 7.94 lm: 24.9 mA, 17.60 days
24: 9.05 lm: 28.1 mA, 15.55 days
25: 10.2 lm: 31.8 mA, 13.77 days
26: 11.6 lm: 35.8 mA, 12.22 days
27: 13.1 lm: 40.2 mA, 10.88 days
28: 14.7 lm: 45.1 mA, 9.70 days
29: 16.6 lm: 50.5 mA, 8.67 days
30: 18.5 lm: 56.3 mA, 7.77 days
31: 20.7 lm: 62.8 mA, 6.97 days
32: 23.1 lm: 69.8 mA, 6.27 days
33: 25.7 lm: 77.4 mA, 5.65 days
34: 28.5 lm: 85.7 mA, 5.11 days
35: 31.5 lm: 94.7 mA, 4.62 days
36: 34.8 lm: 104 mA, 4.19 days
37: 38.4 lm: 115 mA, 3.80 days
38: 42.3 lm: 126 mA, 3.46 days
39: 46.4 lm: 139 mA, 3.15 days
40: 50.9 lm: 152 mA, 2.88 days
41: 55.8 lm: 166 mA, 2.63 days
42: 61.0 lm: 182 mA, 57.80 hours
43: 66.5 lm: 198 mA, 53.00 hours
44: 72.5 lm: 216 mA, 48.67 hours
45: 78.9 lm: 235 mA, 44.75 hours
46: 85.7 lm: 255 mA, 41.21 hours
47: 93.0 lm: 276 mA, 37.99 hours
48: 100 lm: 299 mA, 35.07 hours
49: 109 lm: 324 mA, 32.42 hours
50: 117 lm: 350 mA, 30.00 hours
51: 126 lm: 372 mA, 28.22 hours
52: 134 lm: 395 mA, 26.58 hours
53: 143 lm: 419 mA, 25.04 hours
54: 153 lm: 445 mA, 23.60 hours
55: 163 lm: 472 mA, 22.24 hours
56: 174 lm: 501 mA, 20.97 hours
57: 185 lm: 531 mA, 19.77 hours
58: 197 lm: 563 mA, 18.65 hours
59: 210 lm: 597 mA, 17.59 hours
60: 223 lm: 632 mA, 16.60 hours
61: 237 lm: 670 mA, 15.68 hours
62: 252 lm: 709 mA, 14.80 hours
63: 268 lm: 751 mA, 13.98 hours
64: 284 lm: 795 mA, 13.21 hours
65: 301 lm: 841 mA, 12.49 hours
66: 320 lm: 889 mA, 11.81 hours
67: 339 lm: 940 mA, 11.17 hours
68: 359 lm: 993 mA, 10.57 hours
69: 380 lm: 1.05 A, 10.01 hours
70: 402 lm: 1.11 A, 9.47 hours
71: 425 lm: 1.17 A, 8.97 hours
72: 449 lm: 1.23 A, 8.50 hours
73: 475 lm: 1.30 A, 8.06 hours
74: 502 lm: 1.37 A, 7.64 hours
75: 530 lm: 1.45 A, 7.25 hours
76: 559 lm: 1.53 A, 6.88 hours
77: 590 lm: 1.61 A, 6.53 hours
78: 622 lm: 1.69 A, 6.20 hours
79: 655 lm: 1.78 A, 5.88 hours
80: 691 lm: 1.88 A, 5.59 hours
81: 727 lm: 1.98 A, 5.31 hours
82: 766 lm: 2.08 A, 5.05 hours
83: 806 lm: 2.19 A, 4.80 hours
84: 848 lm: 2.30 A, 4.57 hours
85: 892 lm: 2.41 A, 4.35 hours
86: 938 lm: 2.54 A, 4.14 hours
87: 986 lm: 2.66 A, 3.94 hours
88: 1035 lm: 2.80 A, 3.75 hours
89: 1087 lm: 2.94 A, 3.58 hours
90: 1142 lm: 3.08 A, 3.41 hours
91: 1198 lm: 3.23 A, 3.25 hours
92: 1257 lm: 3.39 A, 3.10 hours
93: 1318 lm: 3.55 A, 2.96 hours
94: 1382 lm: 3.72 A, 2.82 hours
95: 1449 lm: 3.90 A, 2.69 hours
96: 1518 lm: 4.08 A, 2.57 hours
97: 1590 lm: 4.28 A, 2.46 hours
98: 1665 lm: 4.48 A, 2.35 hours
99: 1743 lm: 4.68 A, 2.24 hours
100: 1824 lm: 4.90 A, 2.14 hours
101: 1909 lm: 5.17 A, 2.03 hours
102: 1996 lm: 5.49 A, 1.91 hours
103: 2087 lm: 5.81 A, 1.81 hours
104: 2182 lm: 6.15 A, 1.71 hours
105: 2280 lm: 6.51 A, 1.61 hours
106: 2382 lm: 6.87 A, 1.53 hours
107: 2488 lm: 7.25 A, 1.45 hours
108: 2598 lm: 7.65 A, 1.37 hours
109: 2712 lm: 8.06 A, 1.30 hours
110: 2831 lm: 8.48 A, 1.24 hours
111: 2954 lm: 8.92 A, 1.18 hours
112: 3081 lm: 9.38 A, 1.12 hours
113: 3213 lm: 9.85 A, 1.07 hours
114: 3350 lm: 10.3 A, 1.01 hours
115: 3491 lm: 10.9 A, 58.04 minutes
116: 3638 lm: 11.4 A, 55.35 minutes
117: 3790 lm: 11.9 A, 52.81 minutes
118: 3948 lm: 12.5 A, 50.42 minutes
119: 4111 lm: 13.1 A, 48.16 minutes
120: 4280 lm: 13.7 A, 46.03 minutes
121: 4454 lm: 14.3 A, 44.01 minutes
122: 4635 lm: 15.0 A, 42.10 minutes
123: 4822 lm: 15.6 A, 40.29 minutes
124: 5016 lm: 16.3 A, 38.58 minutes
125: 5216 lm: 17.0 A, 36.95 minutes
126: 5423 lm: 17.8 A, 35.41 minutes
127: 5637 lm: 18.6 A, 33.94 minutes
128: 5858 lm: 19.4 A, 32.55 minutes
129: 6087 lm: 20.2 A, 31.23 minutes
130: 6323 lm: 21.0 A, 29.97 minutes
131: 6567 lm: 21.9 A, 28.77 minutes
132: 6819 lm: 22.8 A, 27.63 minutes
133: 7079 lm: 23.7 A, 26.54 minutes
134: 7348 lm: 24.7 A, 25.50 minutes
135: 7625 lm: 25.7 A, 24.51 minutes
136: 7911 lm: 26.7 A, 23.57 minutes
137: 8207 lm: 27.8 A, 22.67 minutes
138: 8512 lm: 28.9 A, 21.81 minutes
139: 8826 lm: 30.0 A, 20.99 minutes
140: 9151 lm: 31.2 A, 20.21 minutes
141: 9485 lm: 32.4 A, 19.46 minutes
142: 9830 lm: 33.6 A, 18.74 minutes
143: 10186 lm: 34.9 A, 18.05 minutes
144: 10553 lm: 36.2 A, 17.40 minutes
145: 10931 lm: 37.6 A, 16.77 minutes
146: 11320 lm: 39.0 A, 16.17 minutes
147: 11721 lm: 40.4 A, 15.59 minutes
148: 12135 lm: 41.9 A, 15.04 minutes
149: 12561 lm: 43.4 A, 14.51 minutes
150: 13000 lm: 45.0 A, 14.00 minutes

Thermal Regulation

The Emisar D18 has full thermal regulation, so when it’s running at high power it should regulate itself to whatever temperature the user configured. In my testing, it was able to sustain about 2000 lumens without getting uncomfortably hot.

The thermal regulation has three zones:

• 1 to 50: no regulation
• 51 to 125: PID regulation
• 126 to 150: paranoid zone, will ramp down to level 125 as soon as it thinks it might be getting close to hot

In practice, this is how the thermal regulation looks. This is the same test, shown at three different time scales:

This test was gently fan-cooled, and the light stayed at about 45 C the whole time.

Firmware / Source Code

The D18 uses open-source code distributed under the terms of the GPLv3 license. Full source code is available at https://tiny.cc/TKAnduril .

A few other UIs in my repository should also work on this light if desired, but they are not tested and may require modifications in order to work correctly.

Potential Issues

Since this light uses SST-20 emitters, the tint varies from green-ish to pink-ish as the power level changes. Lower levels are more green, higher levels are more pink.

This is a very solid light, dense, full of copper, and it’s fairly heavy. The mass gives it good thermal performance for its size, but it’s a bit inconvenient to carry.

Turbo is so bright that, even with the large thermal mass, it is still basically a burst mode. Don’t expect the light to stay at turbo for very long unless it’s cooled by water.

There are no aux LEDs or button LEDs. However, the driver has pads to hook up one channel worth of aux LEDs, and the firmware has support built in… so this should make those things easier to add later. It still won’t be easy though, since it’s really not designed for that.

The lanyard connection part of the body tube is a little bit sharp. If you use a lanyard, use a thick one to make sure it’s secure.

Quick comparison to other Emisar lights

In a nutshell, this is the current Emisar lineup in chronological order:

• D4: smallest, most EDC-friendly light… floody and powerful
• D1: medium-range compact thrower, still small enough to EDC, in a class by itself
• D1S: longer-range thrower, unusually small and lightweight for its throw class
• D4S: bigger D4 with more throw, more runtime, a nicer body, nicer UI, and aux LEDs
• D18: compact photon grenade / soda can light, 3×18650, well-balanced beam for use both indoors and outdoors

Repeating the pic from earlier, this is the Emisar family so far: (left to right: D4, D4S, D18, D1S, D1)

Summary

What I liked:

• It’s basically an upgraded Meteor, and it fixes everything I disliked about the M43:
  • It’s quite a bit brighter.
  • Has a nicer-looking throwier beam.
  • Has my favorite UI.
  • The recessed button is a nice upgrade too.
  • Slightly decreased width and slightly increased length makes it more comfortable to hold.
• Good build quality. Emisar consistently makes good quality, and this is no different.
• Low lows: goes down to ~0.1 lm
• Super high turbo: goes up to ~14000 lm
• Good regulated modes in the middle, up to ~2000 lm
• Beam is throwy enough to be useful outdoors, yet still floody enough to be useful inside
• Hank did an awesome custom modification on mine to make it use emitters of four different color temperatures. The end result looks like it’s about 4200K or 4300K and has lower tint duv values (less green, more pink) than a stock light would have.
• Available in a wide range of different color temperatures, to suit all tastes
• The UI is my favorite one… but I’m pretty biased.
• Lightning storm mode is particularly nice with a light this bright.
• Open-source firmware means it can be customized easily and extensively. Use whatever UI you want.

Things I feel neutral about:

• The individual optics are a little inconvenient for modding
• The spill area has a mild flower petal effect in an irregular pattern, but it’s not very noticeable and it might just be because I took the optics off when I took the light apart.
• As a stubby / soda can / soup can light, it’s a bit large and heavy for daily carry… but that’s not really what it’s for, so it’s okay.
• No aux LEDs… but those are kind of superfluous anyway.
• Protected cells don’t fit… but they wouldn’t be able to withstand turbo mode anyway.

What I didn’t like:

• Tint is a bit green (+duv) on low modes (or cyan or yellow, depending on CCT)
• Lanyard attachment area is a bit sharp

_The_ End

Most jokes have a butt. This one certainly does: