Review: Noctigon KR4 (18650, 4xE21A 5000K, CRI96)

Disclaimer: I received the Noctigon KR4 free of charge from the manufacturer

The Noctigon KR4 could be considered as a tail switch variant of the popular D4 but it comes with some significant design changes. The version tested here utilizes small domeless Nichia E21A emitters with a constant current driver up to 5 amps. Unlike the other KR4 variants (SST20, XP-L HI, 219C) the E21A version is based on a noFET version of the Anduril firmware, so the output is lower, but the tint and color rendering should be better.

If you’re here only for the numbers, here they are:

-Didn’t have enough free hands to test the current draw on turbo. Any 10A 18650 should do fine.
-Default ramp floor is 3/150 and lower modes are not guaranteed to work. At 1/150 the minimum output is about 0.2 lumens but may exhibit some visible flicker.

Manufacturer’s specifications
Battery: 18650, not included (accepts unprotected button and flat tops), 18350 tube optional
LED: 4x Nichia E21A 5000K (several LED options available, also as mules)
Water and dustproof: IP67 up to 1 meter
Impact resistance: N/A
Mode memory: configurable
Low voltage protection: yes
Thermal regulation: yes, adjustable
Lockout: electronic and physical
Tripod socket: no
Tailcap magnet: no
USB charging: no

Manufacturer’s output specs
No specifications available at the time of writing

Measured dimensions and weight
Length: 97.8 mm
Head width: 29.0 mm
Weight: 89 g with lanyard ring, 95 g with pocket clip, plus 47 g for an 18650 battery

Box and contents

Bundled in the box with the light:
Pocket clip + optional lanyard ring
2+1 extra o-rings
Stainless steel bezel

There was no manual included. For that you need to browse to the Anduril firmware repository

Physical appearance

Four E21A leds behind a glass lens and a clear Carclo 10622 optic.

RGB Aux led colors are configurable in several different ways.

Light is operated by an electronic switch at the tail of the light. The switch has some extra travel before activation so it’s not as precise as Noctigon’s praised side switches. Clicking the switch from its edge feels better than at the center. The switch is very reliable and I didn’t have any activation failures during my testing. Albeit slightly too easy to activate in my opinion, I didn’t have any accidental activations happen.

Both ends have a spring contact, but only unprotected batteries fit. Flat and buttons tops both work.

The firmware flashing pads are compatible with the optional reflashing kit

Tail switch signal is transferred via an inner tube to the driver. The edge of the tube makes contact with a large spring at the driver. The setup seems more robust than on the FW3A, which has had issues with tight tolerances and contact issues.

My sample was equipped with 4x5000K E21A LEDs. The tightly populated MCPCB consists of two separate boards and is securely attached to the flashlight body with two screws.

RGB aux leds on their dedicated board are configurable in brightness and color. From the factory they change color according to battery voltage. Voltage indicator with aux leds isn’t always stable and switches between two colors. For example between blue and magenta on an almost full battery.

Aux leds in action when configured to rainbow mode:

Size comparison

Zebralight SC64c, BLF/Lumintop FW3C, Emisar D4, Noctigon KR4

(forgot to screw the FW3C head fully in, sorry about that)

User interface

The Noctigon KR4 is operated via an electronic tail switch. It uses ToyKeeper’s Andúril software.

The Andúril UI is very intuitive in basic use. Give the light to someone and they’ll figure it out right away. The light is ready to go with the stock settings, but it offers lots of things to configure to you liking. Such as setting the minimum and maximum points at the end of the ramp and configuring the temperature limit. In addition to the smooth ramp there’s a stepped ramp with however many steps you prefer. Default is 7 + turbo.

You can always get the latest operating instructions from the Anduril firmware repository.

Beam and tint

The clear Carclo 10622 optic combined with the small domeless E21A emitters produce quite a throwy beam for a quad. The hotspot is only 19 degrees (full width half maximum) with the spill covering up to 58 degrees (1% lux output). At 5.9 candela/lumen ratio it is about twice as throwy as most triples or quads such as the FW3A using XP-L HI emitters and a clear optic.

The absolute throw is limited since the maximum output only reaches 1200 lumens. At about 7000 candelas at turn on it’s about on par with similar multiemitters hotrods.

The clear 10622 optic isn’t necessarily the best option for the light as it has visible tint shift within the beam. Beam tint is fairly stable at the hotpot (50% output) but turns warmer and greener towards the spill. It can be remediated with a floodier Carclo 10623 optic which available as an option from the manufacturer. This will decrease throw significantly.

As is typical, the tint gets cooler and more magenta on higher modes.

I didn’t have a floody 10623 optic available so I sanded down an extra 10622. The result was a much more consistent tint throughout the beam as expected. Throw was reduced almost 60% from 665 cd to 273 cd on mode 3. The resulting candela to lumen ratio of 2.4 is similar to many other triples/quads and still throwier than the Skilhunt H04 RC for example.

The floody beam with a positive 0.0020 duv tint is visually nicer than the SST20 at similar duv. All the measurements were done after 30 seconds, but on brighter modes the tint will change while warming up. After several minutes on mode 4 the tint lands on the black body line.

Spectral data and color rendering

For spectral information and CRI calculations I use an X-rite i1Pro spectrophotometer with HCFR, Babelcolor CT&A and ArgyllCMS spotread for the graphs and data. For runtime tests I use spotread with a custom script and an i1Display Pro because it doesn’t require calibration every 30 minutes like the i1Pro.

Explanation of abbreviations

If you have an hour to spare, I recommend watching this presentation on IES TM-30-15 which also shines light into color rendering in general.

Color rendering overview on different output modes measured from the hotspot.

CRI data on turbo measured from the hotspot at 30 seconds:

On default mode 3 (113 lm):

When a floody optic is used, the tint averages out so the center is slightly warmer and greener:

CRI data on other modes
Ramp floor
Mode 1
Mode 2
Mode 4
Mode 5
Mode 6
Mode 7

Output and runtimes

As can be seen from the increasing output at the end of the cooled turbo runtime test, the decreasing voltage difference between the battery and the LEDs improves efficiency as the driver isn’t wasting the difference as heat anymore.

Driver + optics efficiency is mediocre. At 400 lumens (100 lumens per emitter) a Nichia E21A should have an efficacy of about 133 lumens per watt, but the total Noctigon KR4 system averages 83 lumens per watt over runtime test on T2. 38% losses are significant. The result is close to the Skilhunt H04 RC E21A, which did 71 lumens per watt with the less efficient 3500K emitters.

However, for a super high CRI light the total system efficacy is not bad at all. The KR4 will sustain about 400 lumens indefinitely in room temperature without external cooling when the temperature limit is set to 55°C. This is higher than other small 18650 triples or quads using the popular 7135+FET configuration, which can reach about 250 lumens even when using low CRI emitters. Of course the KR4 also has a bit more mass and surface area to help with shedding the heat.

The lowest output depends on user configuration. The ramp default minimum is 3/150, and the lower modes aren’t guaranteed to work, but on mine the two lowest setting exhibited a very slight flicker. This should bring the minimum level down to 0.2 lumens.

The temperature regulation works well and there’s no sudden overshoots to either direction with this LED configuration. Use the light in a cooler environment or with a minimal amount of wind blowing on it and the sustained lumen output will be higher.

The driver uses a noFET version of the Anduril build with a maximum constant current of 5 amps. This also means that unlike on the direct driven other LED versions whose output tracks battery voltage, the maximum output of 1200 lumens on the E21A model is available even with a mostly discharged battery.

Standby drain

There’s a negligible parasitic drain on the battery when the light is switched off and the aux leds are disabled. This is understandable due to the electronic soft switch. It would take several years to drain the battery inside the light. However, if you intend to store the light for prolonged periods, you should remove the battery or open the tailcap just a bit to break contact.

With the aux leds enabled, the drain understandably increases. On low brightness the jump is minuscule from 40µA to 50µA, but on high it increases to 390 µA. This would drain an 18650 in about a year.

The drain is not constant. Rather is jumps every couple of seconds to measure the battery voltage. This is to turn the aux leds off when the battery is getting low. Following values are average current measurements over one minute, but since the power spike is very short, it may not register on my meter very precisely.

Standby current with aux leds off: 40µA
With aux leds on low: 50µA
With aux leds on high: 390µA


No visible flicker.

Being a constant current driver there’s no PWM. Only the slightest amount of low amplitude 3.9kHz ripple is present on modes 3 to 7, but it’s nowhere near visible. The flicker snob index of 0% guarantees that. The light works well for all photo and video applications.

Worst case flicker on mode 6:


The thermal calibration was bad out of the box and read 0°C in room temp. If calibrating at the factory is not viable there should be a little note in the package urging you to do the calibration first thing.

The light was tested with factory default settings after calibrating the temperature sensor and setting the temperature limit to 55°C (30°C+25°C) in the thermal config menu. Despite this, the outer surface of the light reached 60°C on modes 7 and turbo.

As evidenced by the lack of stepdown on modes 5 and lower, the temperature stayed below 55°C on those. When using your hand as an extra heatsink, the temperature will be even lower.


With the Nichia E21A option the Noctigon KR4 is the highest CRI flashlight from Noctigon to date. Combined with a new type of constant current FET driver similar to led4power’s offerings, the KR4 aims to improve on the efficiency of the traditional 7135+FET driver especially on higher outputs. On the E21A version the driver offers regulation up to 5 amps with only the temperature being the limiting factor. This results in a maximum output of 1200 lumens, which is plenty for EDC.

The KR4 feels quite a bit bigger than the FW3A. I’d like it to be a bit smaller, but at less than 10cm long, I can’t complain as it fits the coin pocket of my jeans. Thanks to the driver and the higher mass of the body, the KR4 manages to maintain 400 lumens at 60°C surface temp without additional cooling, which is higher than most small triple/quad flashglights in this price range. It doesn’t quite reach the high end boost driver efficiency from the more premium manufacturers such as Acebem, Olight or Wuben, which can maintain higher output in similar conditions.

The small E21A leds combined with the clear optics produce a quite throwy beam in its category. It’s nearly two times as throwy (candela per lumen) than other small multiemitter lights. However, with the stock clear Carclo 10622 optic the beam exhibits some visible tint shift being warmer and greener at the edges. I would recommend getting the floody 10623 optic as an option, which will improve the tint consistency.

The only thing I’d complain is the slight play on the switch. It also takes less force to to activate than I’d like. Electronic or physical lockout makes the light safe to carry though. Overall I’m very happy with the build quality. The construction feels more solid and less prone to contact issues than the similar FW3A. Operation during one week of testing was very reliable. Combined with the excellent color rendering and tint with my favorite flashlight UI, the KR4 is easily worth the 55 bucks.

+ Very good color rendering and tint

  • Flat regulated output
  • Thermal regulation works well
  • Full output available with low battery state of charge
  • Anduril firmware is highly customizable with regard to output and temperature settings
  • Fancy RGB aux leds
  • No PWM or flicker
  • MCPCB securely mounted with screws
  • Easily swappable Carclo optics
  • Low parasitic drain even with aux leds on
  • Seemingly reliable operation

- Default clear optics show tint shift within the beam, floody Carclo 10623 or some sandpaper/d-c-fix takes care of this

- No temperature calibration done in the factory

- Lowest stable output is pretty high at 1.5 lumens (may vary between samples)

- Mediocre driver+optics efficiency

- Some extra travel on the tail switch

- Actuation force of the switch could be higher

- Slightly bigger than competing multi-emitter 18650 EDC lights

  • Pocket clip is big, no deep carry version available
1 Thank

Thank you for this excellent review. :+1: :smiley:

Absolutely, otherwise it is a waste of these LED’s advantages, based on your custom sanded optic looks excellent.

Very nice review. I was waiting for the release of the E21A KR4 and your thorough review answers a lot of questions I had.

Fantastic useful and detailed review!

Awesome review as always. I finally got round to watching the TM-30 video as well so can understand the charts a bit better too.

I’m surprised there’s tint shift in the E21A with the default optic, I hadn’t heard of them having any tint shift before. I wonder if E17A with frosted optic might be a better choice overall?

Thank you for the excellent review. I was originally planning to wait to order the KR4 w/E21A LEDs, but I’m glad I caved and got one with SST-20 4000k FA3. I love the KR4, but I doubt I’ll buy the E21A version. The FA3 bin SST-20s have good enough tint to my eyes. If I’m going to downgrade output by over 50% I want a beautiful tint and beam pattern with decent efficiency. This one unfortunately has none of the above.

Hank did say he would offer multiple CCTs, wonder if 4000k or 3500k would look nicer with the stock optic? Efficiency unfortunately will be worse I’m assuming.

Your question was already answered in the review:

“I would recommend getting the floody 10623 optic as an option, which will improve the tint consistency.”

I assume you do not know that Carclo 10623 is the frosted optic.

I am familiar with the optics thanks, I was talking about E17A emitters. E17A is the smaller brother to E21A so is throwier. Under a frosted optic I was proposing that they could have similar throw to the E21A with a clear optic but still have the same benefits of the blended tint a frosted optic would give.

Are there actually ultra high CRI E17As? I know Clemence offered them for the Jet-mini but they were 80CRI max IIRC.

Yeah there are. They were available in the H04 group buy and are available on his store.

Absolutely amazing review and lovely images. Thanks for the write up and time taken to do this review.

Hank is shipping KR4s with FA3 SST-20 4000K? If so, I may have to get one.

Interesting. Did you have the aux LEDs on high mode?

It seems like it could flip between two colors if certain conditions… like if the voltage was close to a threshold, the aux LEDs were on high, and the higher of the two levels had two colors turned on instead of one. The way that could happen is:

  1. Battery measures 4.11V, so aux LEDs turn purple (R+B).
  2. Red+blue aux LEDs use more power than just blue, so voltage drops.
  3. Battery measures 4.09V, so aux LEDs change to just blue.
  4. Blue uses less power than red+blue, so voltage increases a little.
  5. Repeat from step 1.

IIRC, it checks every 8 seconds, so it could change colors about 8 times per minute. In sleep mode, it wakes up very briefly about 8 times per second to handle the blinking aux LED mode… so the animation can happen at 8 fps. Every 64 frames though (every 8 seconds), it turns on the voltage measurement circuit just long enough to get 2 samples (the first sample is junk, so it needs to read twice), and then it checks the result and decides whether to change the aux LED color. If it’s below 2.9V, it turns off the aux LEDs no matter what mode they’re in.

I don’t have a way to measure it directly, but if my math is right, the entire process should take about 1 millisecond every 8 seconds.

Awesome review Maukka!

Any idea when the E21 emitter version will be available for sale? If it’s going to be soonish, will wait and get both the K1 SBT90.2 and KR4 E21 at the same time.

Also, on the intl-outdoor site’s search bar if you type in D1S, there’s an interesting suggested autocomplete for “37D1S v2”. Sure would like to see that come out soon.

Our E21A LED package is stuck on the way for long time due to the virus issue, not sure when it will move, so, we are not sure when it can be offered. The 10623 optic will be default optic for the KR4 E21A version.

IMO pretty mediocre tint but that was to be expected for a D240 binned E21A. The D220 4500K should be much rosier.

Thanks for the review. Very useful graphs as usual! The driver looks really good.

Mine will hopefully be here tomorrow. I need to get some 18350 cells, has anyone verified the max length yet? I would like to get a protected cell to avoid over-discharging if I forget to not use turbo. Assuming turbo for the xpl version draws more than an 18350 can handle.