ArmyTek Predator (XP-G R5, 1x18650 2xR/CR123A) Programmable: RUNTIMES, BEAMSHOTS +

Originally posted: January, 2011
Last Revised: June 12, 2011
Warning: even more pic heavy thas usual.

Specifications (abbreviated):The Predator is a new light from a new manufacturer, ArmyTek. And as you can tell from the stats above, it’s a very versatile one (this is the first time I’ve seen multiple user-configurable regulation patterns in a light ). The multi-modes are also fully user-configurable.

Packaging is a stiff carboard box, with ArmyTek logos and graphic design all over it. Inside, the light comes packaged between two pieces of styrofoam, with a good quality holster with closing flap, wrist lanyard, spare o-rings, boot cap replacement cover, and a 5mL tube of Nyogel 760G (again with ArmyTek logo). The manual is available for download on their website.

From left to right: AW Protected 18650, ArmyTek Predator, 4Sevens Maelstrom G5, Lumintop TD-15, Ray Tactical X60.

Predator: Weight : 134.6g (no batteries), Length 153.9mm x Width 36.7mm (bezel)
TD-15: Weight: 139.2g (no batteries), Length 148.7mm x Width 37.8mm (bezel)
4Sevens G5 Weight: 145.5g (no batteries), Length 156mm x Width 38.9mm (bezel)

Certainly well in keeping with others of the “thrower” XP-G class.

Nice touch the inclusion of Nyogel 760G – it’s one of my preferred flashlight lubes.

On a side note, the threads of my sample were swimming in Nyogel – I had to wipe most of it off. You don’t need to use much – a little goes a long way.

Build quality is very high - the light has a substantial feel.

In fact, the actual surface feel is fairly unique - the body has a very matte finish anodizing that feels almost moulded. According to ArmyTek, the Predator has a much thicker and harder level of anodizing than most lights (i.e. 35-50 microns, compared to typical 15-25 microns). Although the light lacks knurling as such, grip is better than you would expect, thanks to the unique finish. Still, I found it somewhat slippery at times, especially when programming.

The bezel rings and clip are supposedly stainless steel with a titanium nitride coating for additional hardness (ArmyTek claims it is 10x harder than Type III anodizing). Apparently a more sandblasted titanium look is also available.

Lettering is sharp and clear, in bright white against the dark grey matte finish.

Threads are anodized at the tailcap for lock-out.

The light can’t tailstand, despite the scalloped tailcap ring.

There is a raised contact disc in the head, so flat-top 18650 batteries work fine on my sample.

I like the flat disc at the end of the tailcap spring - no risk of scratching your batteries.

Clip is removable (have to remove the o-rings first to get it off/on). Note the clip tends to swivel around the light unless the tailcap is screwed tightly on. There is a small notch in one area of the body which the ring slides into, preventing it from moving once the tailcap is screwed down all the way.

The Predator features the latest emitter (XP-G R5), in a deep smooth reflector. Although the reflector's max diameter a little narrower than the Maelstrom G5 or Lumintop TD-15, it is definitely deeper. I would expect excellent throw.

Lumintop TD15 on the left, Predator on the right.

BTW, the red reflections you are seeing above are due to the red centering ring visible around the emitter (at the reflector opening). It also somewhat visible around the periphery of the beam, up close by the bezel. Below is a shot from my famed “integrating carpet”, to show you what I mean. It is not otherwise noticeable in real use.

Which brings us to the requisite white wall hunting . All lights are on Hi on AW protected 18650, about ~0.75 meter from a white wall (with the camera ~1.25 meters back from the wall). Automatic white balance on the camera, to minimize tint differences.

As you can see, the Predator throws very well. The main difference in the reflector design is that the Predator has a tighter and more defined hospot than other lights of the XP-G “thrower-class”. However, it also has a greater corona immediately around it. My sample has only the mildest evidence of the center-void darkspot common to XP-G lights with smooth reflectors (your experience may vary, however).

I've recently updated my 100-Yard Outdoor Beamshot Round-up, including the latest XP-G lights. Check out that round-up thread for more details on the testing method, plus higher quality JPEG images of all lights. For now, here is an animated GIF of relevant XP-G comparisons:

Since I was little off in my aiming of the lights, I decided to retest this recently on my replacement Predator:

Note however that the full snow cover causes a lot reflection and diffusion (both ambient light and the flashlight beam, leading to a less clearly defined hotspot).

User Interface (UI)

The manual for this light is 8 full-letter pages long, so I won’t be able to go through it all here. However, the manual is a little unclear in places, so I thought I would summarize the main points. Note that the light is reasonably straight-forward to use once programmed, and has a complex and sophisticated programming interface (somewhat similar to the Novatac/HDS/Ra and LiteFlux lights).

UPDATE JUNE 12, 2011: There have apparently been some tweaks and updates to the programming interface on the Predator. See this post and discussion thread from CPFer LightForce for a discussion.

In simple terms, the light is turned on/off by the forward tailcap clicky switch (press for momentary on, click for lock on).

Output modes are arranged in what ArmyTek calls two “lines” (i.e. head tight for “First line”, and head loosened for “Second line”). Within each of these head states/lines, there are multiple modes that you can switch between. Switching is controlled by twist cycles of the head relative to the body (i.e. loosen/tighten or tighten/loosen to advance modes within each “line”).

How many modes per line, and what they are, are fully user-programmable. What is rather unique about the Predator is that the precise regulation pattern (“stabilization”) is also programmable (more on that in moment …)

To explain what you can do with the light, let me start with the default shipping settings. The light comes with two factory pre-set states, called “Military” and “Outdoor”. Military comes set by default, but you can switch between them in the programming menus.

Military is set with Full regulation, with the First line (head tightened) having three modes: Max (“500 lumens”/1.5A), “100 lumens”/250mA, and “7 lumens”/15mA. Second line (head loosened) has two modes: Strobe 15 Hz, and Firefly/“0.1 lumens”. Again, you switch between modes by tighten/loosen (second line) or loosen/tighten (first line) head twists in under 1 sec.

Outdoor mode uses Step regulation, with First line having 2 modes: “350 lumens”/1A, “65 lumens”/150mA and Second line having 3 modes: Strobe 15 Hz, “7 lumens”/15mA, and Firefly/”0.1 lumens”.

Note that I put the lumen values in quotes above, as they refer to emitter lumens, not actual OTF lumens (see my Summary Tables later on for more on that).

UI Features

So, what else can this light do? Well, to start, you can change any of those output modes to any level from Firefly to Max in the programming state (using a continuously-variable interface, with flashes at 25%, 50%, 75% and 100% output over a 40sec timeframe).

In addition to constant output modes, you can select SOS, Beacon or Strobe mode. Within Strobe, you can set the frequency from 1 Hz to 50 Hz (claimed).

You can change the number of pre-set modes in First line from anywhere from 1 to 10 modes, and Second line from 1 to 5 modes (default is 2-3, depending on whether Outdoor or Military defaults are chosen).

You can set the auto mode-memorization feature for each line independently (i.e. remember last mode for both lines, neither, or only one or the other).

You can specify the battery voltage being used (2xCR123A 3V batteries, 1x18650 3.7V Li-ion, 2xRCR 3.7V Li-ion, 1x18650 LifeP04 3.2V). Unprotected batteries can also be used.

What is truly innovative is the ability to set three different regulation patterns for the light. FULL regulation is as you’d expect – perfectly flat regulation for as long as the battery can handle it, then an immediate drop to off or a low output state. SEMI regulation is what you typically find on a number of multi-power lights – the light maintains flat regulation for awhile, then drops into direct-drive as the battery power dwindles. STEP regulation is a similar to SEMI, but shows a step-down pattern of lowering outputs rather than a smooth drop-off. All of these patterns are shown in my runtimes graphs below.

How to Set Output Levels

To select a variable output level in the first line, start from head tight and loosen the head for at least 1 sec (i.e. wait for the light to switch lines), then immediate tighten, wait again for 1 sec (i.e. for the line to switch back), and then immediately turn off-on at the clicky. All the above has to be done in under 3 secs total, so timing is very tight (i.e. don’t delay on the head twist any longer than necessary to switch lines). It is highly unlikely that you would ever enter the programming state by accident!

The light will then begin ramping from Firefly to Max, which takes about 40 secs in total (with flashes at 25%, 50%, 75% and 100% output). Once Max output is reached, the light restarts at Firefly and cycles through again. Make your selection by a loosen-tighten head switch.

As you can see the output is not visually-linear across the ramp. The quarter percentage flashes thus seem to reflects actual drive currents. As such, you will have to be quick to select any of the truly low outputs.

At the end of the ramp, the cycle repeats from low to hi again.

Note that my testing has revealed a bug in the software in the initial batch of lights (subsequently confirmed by ArmyTek). If you leave the light in the continuously-variable output selection (or any other programming mode) for more than 2 minutes with no activity on your part, the light exits and locks you out of these features permanently (i.e. can no longer enter programming or output selection). The light still functions normally for how ever it was previously set, but you can no longer program any changes. ArmyTek has fixed this bug on currently shipping lights, and is working to see if they can find a solution for existing owners. For current owners in the meantime, I recommend you set your output levels as quickly as you can.

How to Program the Light Settings

To enter the general programming menus, you start in the second line modes. Do the same as the first line output selection above, only starting with the head loosened (i.e. tighten, wait 1 sec, loosen, wait 1 sec, click off-on – all under 3 secs in total). Once inside the programming state, you double-click the tailcap to advance through menu levels, and loosen/tighten or tighten/loosen the head to select the menu entry you want. Menu levels and sub-levels are presented as a series of bright flashes. Please refer to the manual for a description of the menus.

If the above sounds complicated, that’s because it is. You can’t really re-program the light without the manual in front of you (unless you’ve memorized all the menu and sub-menu levels!). Basically, this will seem familiar to LiteFlux users (especially those that used the earlier twisty version of their programmable lights). But once programmed the way you want, the light is straight-forward to use (as described at the top of this section).


Light has no evidence of PWM on any mode, leading me to believe it is current-controlled.


Beacon mode was 1 flash every 7 secs in my testing.

Strobe speed in user-controllable, from 1-50 Hz (according to the manual). However, that’s not entirely true. First off, here is the default “Military” Strobe (which is reported to be 15 Hz, but was 12.5 Hz in my testing)

According to the manual, the strobe freq will change from 1Hz to 50Hz over 30 secs in the programming mode. In my testing, I found it took about 35 secs to run through the range. However, the ramping speed is far from linear – although it starts at 1 Hz and slowly increases, all the frequencies between 15-50 Hz are presented in the last 1-2 secs of the cycle (and the cycle restarts at 1 immediately after 50Hz is reached).

Despite repeated attempts, the maximum strobe speed I could detect was ~30 Hz, as shown below:

As you can see, the strobe went from 19Hz to 32Hz in about 400msecs (i.e. those are the same traces above, I am just highlight different frequencies over a 500msec period). I wasn’t able to capture any speeds faster than that. Of course, I don’t really see why you would want anything this high – by 50 Hz, you are getting into nearly PWM constant light frequencies.

Testing Method:

All my output numbers are relative for my home-made light box setup, a la Quickbeam's method. You can directly compare all my relative output values from different reviews - i.e. an output value of "10" in one graph is the same as "10" in another. All runtimes are done under a cooling fan, except for any extended run Lo/Min modes (i.e. >12 hours) which are done without cooling.

I have recently devised a method for converting my lightbox relative output values (ROV) to estimated Lumens. See my How to convert Selfbuilt's Lighbox values to Lumens thread for more info.

Throw/Output Summary Chart:

Effective November 2010, I have revised my summary tables to match with the current ANSI FL-1 standard for flashlight testing. Please see for a description of the terms used in these tables.

the table above is based on my original Predator sample. The replacement Predator ArmyTek has sent me (due to the circuit glitch described above) has lower max output, very similar to the TD-15. Throw is just slightly reduced from the first sample. [/I]

To get a better idea of how some of the pre-set modes compare to actual lumens, here is a comparison table:

Output/Runtime Comparison:

One of the most unique features of this light is the option to choose between 3 different regulation (“stabilization”) patterns. Here are some comparative runtimes on different battery sources, starting with 1x18650:

As you can see, FULL regulation gives you exactly that – perfectly flat regulation, followed by a sharp drop. Note that in some cases the light came back on in a low output level after the protection feature had been tripped. This effect was variable, and I suspect has to do with the cell recovering sufficiently to activate the light at a low level.

SEMI and STEP regulation seem to give very similar overall runtime results – the difference is just in the pattern of drop off once the light falls out of regulation. Both typically last longer than FULL regulation.

Note that the differences between regulation patterns at lower output levels are not as noticeable. It is really only on Max or near-maximal output levels where you can really see the difference.

Also, at least on 1x18650, there is no difference in output levels between the regulation methods.

In any case, I recommend you run the light in SEMI or STEP regulation, as they are the most efficient and won’t potentially leave you in the dark with little warning.

Here’s how 2xRCR compare:

As before, runtime on FULL was less than SEMI or STEP. Although there was a brief step-down near the ends of the STEP run, this is only on the order of a few seconds, not many minutes. The key difference being STEP and SEMI drop down to a low level rather than abruptly shutting off as FULL does on 2xRCR. I again recommend SEMI or STEP.

On my original sample, there was a slight increase in output at "350 lumens/1A" on FULL regulation on 2xRCR. I have re-tested the replacement sample, and it shows the expected consistent output level across all three stabilization settings. I have updated the graph above to show this revised testing result. As you can see, there is not much difference in the regulation pattern on 2xRCR at this output level.

And now 2xCR123A:

Note that there is an interesting step-down pattern on FULL on 2xCR123A. But once again, I recommend you run the light on SEMI or STEP for maximum runtime without abrupt shut-down (i.e. especially as the light drops to lower outputs).

To compare to other lights, I’ve used SEMI regulation pattern for the comparison graphs below (except on Max, where I show both FULL and SEMI).

Note that all the runtimes were performed on my first Predator sample. A second sample was received (due to the software glitch lock-out of my first sample - see User Interface discussion above), which showed slightly lower Max output, similar to the TD-15.

Potential Issues

As with all lights with complex programming modes, you will need to refer to the manual if you want to re-program any of the features or output levels of the light. Once programmed (or with the default set), the light is straightforward to use.

The programming modes can be a bit tricky to enter (i.e. timings are very tight, requiring multiple twists and clicks in a narrow time window). The switch is bit stiffer than average.

The body texturing is interesting, and helps with grip, but not as much as actual knurling would. The light can be a bit slippery, especially when doing repeated head twists quickly during programming. I found myself occasionally unscrewing the tailcap during programming (due to gripping too firmly during twisting), thus cutting power and accidentally exiting the menus.

UPDATE: There was a bug in the circuit software on the initial batch of lights, causing you to be permanently locked out of the programming/output selection modes if you delay more than 2 minutes without making a selection (i.e. remain inactive for 2 mins). ArmyTek has fixed this software issue, and all currently shipping lights should be working fine.

Preliminary Observations

The Predator is certainly an extremely sophisticated light for a first-outing. Overall throw and output is right up there with other recent XP-G R5 “thrower” lights in this 1x18650/2xCR123A/RCR class. But what really distinguishes this light is its distinctive build and incredibly detailed programming interface.

Starting with the build, I quite like this “matte” finish heavy anodizing. It gives light something of a molded look and feel. I would like a bit of actual knurling though, as the light can still be slippery (which is a challenge when trying to program). The titanium nitride coating of the stainless clip and bezel rings is also very distinctive! It’s a small touch, but I am also impressed with the inclusion of Nyogel 760G with the light.

The sophistication of the programming interface is really the key aspect of the light. You can set as many pre-set levels as you like (from 2-15), with an extremely wide range of constant outputs and special modes available (e.g. strobe from 1 to 30+Hz, etc.). This light has at least as many options as other sophisticated programmable lights I’ve tested (e.g. Liteflux, Novatac, etc.)

A truly unique feature of this light is the choice of three stabilization (regulation) patterns, over a wide range of battery types. This is not something I’ve seen before. The main difference is really that FULL typically leaves you abruptly dropping to zero output with no warning. SEMI and STEP are still regulated in a very flat-output manner - they just drop off slowly as the battery nears depletion. And as a general rule, SEMI and STEP are slightly more efficient at all levels on all batteries - I recommend you stick with one of these two modes.

I am also glad to see that throw and output/runtime performance is right on par with other lights of this class. Over the range of outputs I’ve tested, performance seems comparable to good current-controlled lights. This is rather surprising, since every other light I’ve tested with a continuously-variable output-selection mechanism uses PWM (which is less efficient than current-control). I can detect no signs of PWM with my oscilloscope-sound card setup, and the output/runtime performance tracks with a good current-controlled light (like Fenix, Klarus, etc.). That is very impressive, given the incredibly wide range of outputs on the Predator compared to other limited multi-level lights.

There is really little to criticize here, except perhaps the complexity of the programming menus (which require the manual to be on hand). But that is to be expected on any light with this many features and controls. That being said, the manual is a little confusing in places, and could benefit from a professional re-write. Also, having done a lot of re-programming over my testing, I will say I find it a challenge to consistently enter and move through the menus given the tight timings and the need for both head twists and double-clicks. The relatively stiff switch and lack of body knurling doesn’t help.

But in actual use (once programmed), I find the light fairly easy to use. It has great throw and output/runtime performance. Coupled with an incredibly sophisticated programming interface and loads of features, you get a lot for your investment here.


Predator provided by ArmyTek for review.

To follow the online discussions for this review, please see the full review thread at CPF.

Return to the master review list (at

For a list of all my CPF flashlight reviews in chronological order by battery type (direct link to CPF), please see here:
Candlepowerforums Threads by Selfbuilt

If you are interested in sending a flashlight to me for review, please see my Terms and Conditions here:
Terms and Conditions

For more information on my testing methodology, please see my Methodology section here:
Testing Methodology

Unfortunately, my flashlights are expensive to feed with all the runtime tests I perform. I don't accept any payment for any of my flashlight reviews, but I will gratefully accept donations to my Paypal battery fund. Your contributions will go toward helping defray the costs of creating all my detailed reviews.


For cash donations, please use my personal Paypal account (note that ONLY cash transactions are possible on this account).


For all credit/debit card donations, please use my regular Paypal account.

Site last updated on June 12, 2011 - selfbuilt (at) sliderule (dot) ca (replace the "at" and "dot" labels with the appropriate symbol for e-mail)
All material © 1999, 2011 by Eric Marcotte.