[ 40,000 Lumen Manker MK38? - A Review and Teardown ]

Hello BLF,

Wishing everyone a happy 2022 filled with fulfilling days, health, and joy.

To begin the new year, I purchased the Manker MK38 flashlight and would like to share some of my initial thoughts with the community. I'm expecting many people to create their own standard reviews soon with runtimes, brightness measurements, UX overviews and so on.. that's not what I will be doing; instead, I'm hoping to share a slightly different perspective from a more engineering point of view.

The Manker MK38 is one of the new 'soda-can' flashlights of an emerging class, powered by three 21700 cells (either integrated or separate), with a claimed output luminosity of around 40,000 lumens. The MK38 was released towards the end of 2021, and appears to be a direct competitor to the Acebeam X50 (which was released in mid 2021).

These two flashlights have a very similar form factor and are both powered by essentially a 3S 21700 battery pack. They both claim about the same output, and their brightest configuration uses eight XHP70.2 LEDs. They are both offered in 6500K and 5000K variants, as well as other LEDs (such as the SFT40 for the MK38 with more throw, and the GeTain FC-40 for the X50 for higher CRI).

There are lots of people out there who conduct flashlight reviews from a general point of view, talking about the general construction, UX, runtimes, and so on. I'd like to do things a little differently so I'll be focusing a bit less of the general aspects of the flashlight such as runtime, brightness measurements, charging rate etc, and instead look at it from a more engineering and EE point of view. This will take some time since it will require disassembly of the flashlight - I will update this page as it progresses.

The MK38 I purchased is the one in a very nice off-white / light grey micro-arc oxidation finish. I also chose the more expensive option with a XHP70.2 5000K, and the built-in battery pack with USB-C charging. I will be taking the flashlight apart including the charging and battery module.

The flashlight can be purchased here: http://www.mankerlight.com/manker-mk38-satellite-multi-purpose-handheld-searchlight-power-by-3x-21700-batteries/ .

I purchased this flashlight with my own money so I do not have any affiliates or disclosures to make. This was a very expensive flashlight, so I hope this write-up is useful to you all!

First Impressions

The first impressions of the flashlight are very good. It comes in a nice box with good foam padding, as well as an instruction manual, a 60W USB C charger and cable, and a wrist-lanyard (which seems.. much too insufficient for a flashlight of this size.. a shoulder strap would have been more appropriate). It also comes with a carrying handle which has two e-switch buttons and two small fans.

I will be comparing this flashlight with a Zebralight SC700d as a reference, since it's a bright flashlight which many people here are familiar with. I do not own any other 'soda-can' flashlights so I cannot compare with any of them. It would be nice to compare this with the original Emisar Meteor M43 just to see how things have evolved over the years.

The SC700d weighs 158g with a 21700 battery, and the Manker MK38 weighs 890g with the included battery back, charging base, and carrying handle. While physically compact, this makes the flashlight fairly heavy, though it feels balanced with the carrying handle. There is lots of space for the fingers even when wearing bulky gloves, and the ribs on the handle definitely make it much more secure to hold (esp. with gloves). I couldn't help but think that this flashlight would do well with a shoulder strap, and a DIY version seems possible with the inclusion of the 1/4-20 threaded tripod mount. That said, the tripod mount hole is a little shallow, and may require a small washer/spacer - I tested this with one of my tripods and the hole is just not quite deep enough.

The white finish is bound to get dirty when in use, but I think that's ok. It's a tool after-all and is meant to be used (despite the packaging saying proudly, 'More than just a tool'). The finishing seems excellent and the construction and machining feel solid. I would love to see other flashlights adopt this finish. I raised this up with Fireflies and they said they would consider it for future flashlights.

Subjectively, I am also a fan of the aesthetics, and I prefer its look compared to the Acebeam X50, even though the Acebeam seems to have much more effective cooling fins. I appreciate the industrial design of the MK38 and I'd love to see Manker build some other (smaller) flashlights in the future with a similar design language. It has a storm-trooper feel to it.

I was impressed with how the the e-switch buttons are integrated into the handle (which the X50 lacks), allowing easy control of the two small fans as well as the main switch. The effectiveness of the fans is questionable since they do not provide much airflow, and they are merely blowing air around the channels around the front of the flashlight (I struggle to call them fins). That said, I'm sure they will provide some sort of improvement since I can indeed feel the airflow, and there is likely some Coanda effect in play.

Regardless, the flashlight is large enough and LEDs efficient enough that the MK38 is likely be able to sustain several thousand lumens until the battery runs out. Manker claims 9000 sustained lumens with the fan on, and 5500 without. Given the similar form factor and specs to the X50, and that the X50 has been verified by other reviewers to sustain about 6000 lumens, the claims seem plausible.

One thing that struck me immediately was the glass lens covering the beautiful 8-up orange peel reflector. This lens has what appears to be a very effective anti-reflective coating, so much so that reflections are difficult to see, and it almost looks like there is no glass. Manker claims that the lens "Utilizes dual-sided AR coated, 99% transmittance toughened glass lens" - and I believe them, it's a nice detail.

Manker claims "Integrated precision digital optics technology provides extreme reflector performance", but I have no idea how this reflector is 'digital'! Maybe because it has 8 cavities and 8 is a digit! Jokes aside, the reflector appears to be very nice and I especially like its compact size (for 8 XHP70.2 LEDs).

The LEDs in my MK38 are XHP70.2 LEDs with a 5000K colour temperature. I contacted Manker directly to request which specific LED they used, but they seemed hesitant to provide the actual part number. Comparing with my Zebralight SC700d, I'm quite sure it's the 70CRI variant since it has a slightly poorer colour rendering of reds.

Interestingly, I also asked Acebeam about their X50, and they claimed that they were using 90CRI XHP70.2 LEDs in their flashlight. This seems a little suspect to me given that both the MK38 and X50 have about the same claimed brightness and are likely being driven at close to the same power level. The 90 CRI LED is much less efficient and I would expect something closer to high 20,000 lumens if it were the case. Note: Acebeam claims that their 6500K version is CRI 70 and the 5000K version is CRI 90, in a chat conversation I had with them.

While I do not have equipment to measure the brightness of the flashlight, I can say that it is immense, making the quoted 40,000 lumens believable.

In the above beamshot, I have both flashlights at the same distance illuminating the room, with the camera set on a fixed exposure. The middle photograph shows the SC700d at 3000 lumens, and the right most shows the MK38 on turbo, with the middle of the photo saturated. I'll definitely have more beamshots coming soon.

So whether the flashlight is 40,000 lumens, or somewhat less at 30,000, I can't say for sure at the moment. Regardless of the actual numbers, the MK38 is overwhelmingly bright indoors, and my initial estimates put the driving power around 350W, depending on the exact LED used, and maybe less if the top-bin LEDs are used and the output is less than advertised.

It was a foggy day, so it was perfect to light up the sky with the MK38. The beam is fairly floody with a bright central spot - the SFT-40 version should throw a lot better.

Next, let's move on to the insides. I'm expecting to find a really big inductor (or inductor array) inside, with discrete fets and a standard switch-mode controller IC with current sense feedback similar to my Lume drivers. For the charger, I'm expecting a simple USB type-C port controller with PD support, together with a bi-directional buck booster IC with integrated battery charging such as the SC8915. I'm hoping to see if it's done properly with cell balancing and such, but unfortunately I'm not expecting it to be.

Carrying 'Fandle'

The MK38 comes with a very nicely machined handle and comes in the same gorgeous finish as the rest of the flashlight.

The handle attaches via a large custom screw. Manker provides a simple tool to help screw this in. Other option they could have done is to have use a large Hex or Torx screw, and have a hole through the handle to allow the use of a screwdriver. However, the implementation is decent enough. A nice touch is that the screw acts as a plug for the hole should the user want to remove the handle. An O-ring around the screw provides water-proofing.

The handle interfaces with the flashlight via 4 gold popo-pins. This seems reliable enough, but the pogo-pins are exposed if the handle is removed from the flashlight and needs to be treated with care. Manker has tried hard to ensure water-proofing and added an O-ring around the port.

In future updates, I hope to look at the effectiveness of the two small centrifugal fans on the handle. These provide a small but noticeable airflow around the head of the flashlight and I suspect that they may be more effective than initially meets the eye, so don't dismiss it too quickly!... It will be interesting to compare this implementation vs. the Acebeam X50, which has much deeper fins, but no fan.

Battery Charger and Powerbank

Caveat - please do not disassemble your flashlight; it will likely void your warranty of a very expensive product. I am a electrical engineer and I do these kinds of things as a job so I (hopefully) know what I'm doing and I can fix something if I mess up. I'm doing this with the hope of sharing some knowledge with the community.

Manker offers the MK38 in two different models - one with an integrated battery pack and charger / powerbank, and another one which uses three individual 21700 cells. Due to the huge power that this flashlight draws (over 100W PER cell), personally I think that using raw batteries can be a dangerous choice and creates limitations with the reliability of the overall performance of the flashlight.

The one I have is this powerbank version, and it differs from the individual-battery version since the tail now houses the charging electronics. Let's unscrew this back module and see what we uncover..

The insides are revealed. I was fairly impressed with the overall quality of this charger / powerbank unit.

The system is housed on a single PCB with a very nice matte-black finish - I don't see many of these in mass-produced consumer electronics because it's expensive - some engineer is taking pride in their work!

The key components are the Injoinic IP2716 USB-C Power delivery controller with source-sink capability, hooked up to a Southchip SC8902 bi-directional buck-boost converter with integrated 1-4S battery charging. A 7150 SOT-89 LDO provides 5V power to the Injoinic VSYS. The batteries are wired in a 3S configuration. My initial measurements with a USB-C PD charger show a roughly 33 to 37W input charging power (when battery is about half full), which is in-line with the claims made by Manker (1.5hr charge time to 90%). IMO this quick charge and power-bank function is a significant benefit, and could be worth the price over the individual-battery variant. Manker also provides a no-brand '60W' USB C charger but I haven't used it yet (because it does look a little cheap, and improperly engineered power supplies always worry me!); charging has been done instead with an Anker 60W USB PD charger.

Overall, I'm glad to see that this seems like fairly standard and decent implementation with no major corners cut*. The above photo shoes the charging PCB with all the components on the top layer; there are no components on the bottom layer, other than two pads which the battery wires are soldered to.

* - note this is just based on an initial visual inspection; I haven't made any effort to analyze the actual EE design.

PCB quality looks unexpectedly good and assembly looks decent. The USB C connector is of the waterproof variety, and also features an o-ring to seal it from the outside, in addition to the rubber flap.

However, there are some areas of concern at first inspection. First, I would have preferred to see would be some silastic (or other dampening method) around the larger components like the large inductor and tantalum caps, which could potentially shear off on a hard impact or vibration. I hope that Manker addresses this since it could be a point of failure for a product that presumably is meant to take some abuse in the field. I'm going to call it out now since I'm expecting that there are going to be people who have some failures with their charging system since that big inductor is just asking to be sheared off its pads.

I would also have preferred to see two LEDs on the board for better charge indication (there is only one red LED, which indicates both charge and discharge modes; the LED turns off when fully charged and I would like to have seen a blue or green LED turn on instead).

I did not see any cell balancing circuitry on the board, nor any heat-sinking. The 30W charging rate is fairly high, and personally I would have gone the extra mile to add some thermal pads on some of the components to use the body of the flashlight as a heatsink. Based on a quick soldering test on the wire pads, the PCB looks like it has some internal copper layers (I'm guessing a 4 layer board), so thermal management should be OK in practical situations. I'm going to assume here (for now) that Manker have done their homework. The overall layout of the board seems decent as well.

Taking off the back module, we can see the battery pack. It's a custom-made 3S 21700 pack, with a claimed rating of 10.8V / 4200mAh. Given that it needs to be able to support >=35A continuous discharge, my guess is that it is made up of three Molicel INR-21700-P42A - one of the best performance high-drain 21700 cells on the market. It is shrink-wrapped and padded with foam, and looks reasonably well constructed - we will take it apart at a future time. The two flexible silicone wires connect to the charge/powerbank board and seem adequate for the purpose. The entire weight of the battery only held by the PCB though (and 3 screws), and that could be another point of failure during a hard axial impact.

Ideally I'd like to see some sort of active charge balancing inside, but I'm not holding my breath for it. They could also have been carefully matched in the factory, and the hope is for the electrodes to be properly tab welded together; personally if I were to design this flashlight, I'd make sure the cells have a balancing circuit, and do away with the individual cell variant from a safety point of view (though this is probably not a discussion I want to have since a 40,000 lumen flashlight is inherently dangerous :P.. )

One safety point though - this 3S battery pack is extremely powerful. If something were to happen on the charger (or driver) PCB and a short were to occur, there is no safety mechanism and you could have the entire battery shorted, causing a huge fire hazard. I'd like to have seen a fuse somewhere. Personally, I'll be adding a small fuse to this section, maybe on the PCB or in-line with the battery wire. In the event of some short, the fuse will blow.. Note that it is possible that there is one in the battery pack - I would be very happy if it is indeed there!

Despite some small nitpicks, overall I'm happy with what I see. For an extra $60, you get a battery pack (three P42A cells would have been about $18 by themselves), and a properly designed powerbank and charger implementation (effectively a 12600mAh battery pack). I would have preferred to see a smaller cost delta (say $40), but my guess is, unlike EDC flashlights which sell in much larger volumes, Manker are not going to be selling a huge number of MK38s.

Just the engineering required for the charger electronics alone, battery pack development, material cost, and a different SKU required for the flashlight body, is arguably more than the majority of super-simple AMC7135 flashlights, and makes the $60 cost easier to swallow. They do have to recoup back their engineering and tooling cost afterall.

[ .. continued on the next post .. ]

.. Part 2 of the review / teardown.

Today I opened up the front half of the flashlight. The version of my flashlight comes with 8 XHP70.2 5000K LEDs. Personally I would have preferred a 4000K option. However, as we will soon see, this may not be too difficult to pull do..

Before further disassembly, I did a quick ceiling bounce test.

Ceiling Bounce Test

I used a Extech LT40 lux meter and a ceiling bounce to estimate the brightness of the flashlight. Note that this is not a very accurate method to estimate the brightness of the flashlight since the bounce method doesn't take into account the beam profile and so on. However, it does give a general idea of the brightness when compared to other flashlights.

  • Zebralight SC700d - 136 Lux - 3000 lumens (claimed by Zebralight, let's use this as a baseline)
  • Fireflies E12R prototype - 245 Lux - ~5400 lumens (note that this is a prototype and has some damaged emitters.. I'm not sure which LEDs it uses sadly I don't own a production E12R)
  • Manker MK38 - 1595 Lux - ~35000 lumens (again this is just a number based on extrapolation, inaccurate as it may be)

The brightness measured was in the correct rough ballpark of the claimed 39425 lumens. It's more than 10x brighter than the already very bright SC700d. For actual runtimes and such, I'm sure someone else will conduct those kinds of review since I do not have the proper equipment to do luminosity measurements.

LEDs and Driver

What's inside the MK38? Will Manker go the extra mile to laser off the markings on the IC (don't worry, we'll reveal them if they do)? What is the the LED layout like, and how well does the buck topology work (claimed 98% efficiency)?

Before that, I opened the lens and reflector.

The front bezel is very nicely machined and has a very nice polished black finish in my model. Opening the front bezel reveals a beautiful 74.5mm x 2.0mm AR-coated glass lens, together with a clear o-ring for waterproofing.

Under the glass lens is definitely one of the stars of this flashlight - the custom full aluminium reflector. It's very well machined and is very heavy, definitely contributing a huge amount of the total mass of the flashlight. I was quite impressed by this part. The reflector lifts to reveal the MCPCB.

And the MCPCB is revealed. It's a thick, full copper DTP MCPCB and a quick measurement shows that all LEDs are arranged in parallel, in the 6V configuration. Overall quite simple, but effective. Heat sinking seems robust and the MCPCB is held down by three hex screws fairly securely, which bodes well for thermal management. The main cables connecting to the driver are 16 AWG flexible silicone wires. One nit-pick I had with the layout of the MCPCB is the fairly thin copper pours around the outside edge of the PCB. They could be increased in width for better current carrying capacity.

The good thing is that the MCPCB is very easy to remove. This means that it would be fairly easy to replace this MCPCB for one with different LEDs. I'd recommend against reflowing off the existing LEDs and reflowing new ones since the LEDs on this board cost at least $80 USD by themselves. I wonder if I can get a bare MCPCB from Manker - I'd reflow my own LEDs on it myself if I could.

With the MCPCB removed, I was able to remove the driver from the cavity. The driver cavity is well machined with a huge 8.7mm Z-height. This Z-height allows for the use of wire-to-board connectors, all nicely done with silicone wires. One cable assembly goes to the lighted side-e-switch, and the other to the fandle connector. I should also note that all ingress points and body interfaces had o-ring seals. Note that there are two posts that stick out from the driver head - these act as heat-sink posts for the driver.

And here is what I assume people are most interested in. The driver came out easily enough.

Note - I bought this flashlight also because I had some plans to use it in the field, so I won't be doing a full breakdown and analysis of the driver since I need to reassemble it for actual use. In the future I may do more comprehensive testing, or if I can find a way to get another driver from Manker.

The overall design is fairly straightforward and was as expected.

The main switching coverter is the LM5145 Synchronous Buck Controller set for a switching frequency of 213kHz. The high and low side FETs are comprised of a HYG013N03LS1C2 dual-array each (in parallel). These N-FETs are rated 30V with a R_ds_on of 1.3mR at 10V V_gs each, with a surprisingly reasonable gate charge. Input and output capacitance is at least 100uF from the large electrolytic + tantalum capacitor, plus a few more large MLCCs. The star of the buck converter is a huge 1710 3.3uH inductor. I do not know which exact part it is, but the saturation current is between 35 and 40A, based on similar parts I can find on the market.

Power control is via constant-current feedback using two pairs of sense resistors, 200mR for low-modes, and a 5mR for the high modes (using two paralleled 10mR current sense resistors). Another of the same G013N03 FETs is used to switch the 2nd current sense resistor array in and out of the feedback circuit, just like my Lume X1 driver. This sense voltage is fed into a generic feedback amplifier.

Controlling the whole system is a ATTINY816 MCU, powered by a HT7150 5V LDO. Another small Micro One DC/DC controller is on the left side of the board (as pictured), which I presume is for powering the two smaller fans.

Power input from the battery pack (this will be different for the individual-battery version) is via two fat cables, which are soldered directly to the PCB.

I soldered on a slightly longer cable to conduct some peak-power measurements. Using a Uni-T UT210E clamp meter (not very accurate), I measured a turn-on current of 43.0A at 6.30V to the LEDs, corresponding to about 270W output power (note this may not be an accurate measurement since I had to use a longer wire loop to make space for the current probe I also did not measure the current from the battery pack). This PCB is definitely not going to be able to dissipate that much heat (5.5W even at 98% efficiency, plus the massive heat from the LEDs), so the system will throttle after about 30s or so depending on the starting temperature of the flashlight. Based on some measurements by Texas_Ace on the XHP70.2 LED, I'm expecting about 4500 lumens per LED, or a total output of 36,000 lumens.

Overall, here are some of my initial thoughts:

  • PCB quality looks good. I like how they used a matte-black finish. Overall layout looks OK to the eye.
  • Inductor looks a bit under-specced for the power level, if the inductor is what I think it is. 43A_avg output means a higher peak current on the inductor and I suspect that the inductor is likely going to be operating close to saturation.
  • Heat-sinking is not ideal, but the entire situation is not ideal as is; the switching transistors have some sort of heatsinking to the posts (with a silicone thermal interface pad), but the inductor is not - however the body of the flashlight gets ridiculously hot so quickly at full power that I'm not sure if it acts as a heat sink or source at that point. The flashlight is expected to operate around 6000 lumens continuous, likely around 30-35W? In that case, the driver electronics should be running well within spec, with the LEDs contributing most of the heat.
  • Just to be clear, the driver you are seeing above is NOT a 300W driver - it can handle 300W peak, but only for a short while. A 50W continuous rating would be a reasonable estimate (with proper cooling). Many parts of it are not built for continuous 300W operation, and definitely not the max output. For example, consider the losses just in the sense resistor array. At 43A, that's a total of 9.25W dissipated in just 5mR. It appears that metal element R_sns were used, but they are fairly small - I would have used a physically larger sense resistor (like some of the wide-varieties).
  • Main interconnects are soldered onto small pads - can this handle 40A continuously? Probably not safely, but should be OK for 1 minute or less. Personally I would have use larger pads on the PCB, design the PCB with thick copper (2oz or more), and use maybe a cable assembly (such as two 18 AWGs instead of one 16 AWG) or other method (solid copper post, or copper strips) to connect the MCPCB to the driver board.
  • Battery pack is directly soldered to driver PCB - this is a good thing, since it eliminates the high resistance metal-to-metal contact which will be present on other flashlights (or the individual-cell variant).
  • 3S is a good idea. I don't understand individual-cell lights that use 3P since it can be quite dangerous. Likewise, I'm so glad this driver is a proper current-regulated design, not a driver with a direct-drive FET. While I have designed drivers with FETs before (client requested), it's still something I advocate against.

It was fun to take apart the flashlight to see what makes it go. A close to 300W buck driver on a compact PCB is a fun engineering challenge. Overall the flashlight has been fun to disassemble (and reassemble), not too difficult, and even more fun to operate.

What's Next?

At the moment, I will be using this flashlight in the field since I didn't just buy this to take apart! Perhaps I'll write an update in the future w.r.t real-life use. My main use case will not be running this on turbo, but likely using the flashlight at moderate 2000-4000 lumen modes (Mid 2 and Mid 3) for long durations. As a result, I decided not to take apart the battery pack for now since it doesn't look like I can open it non-destructively (since it has a nice shrink-wrap over it and I don't have any wraps that diameter to repack it for now!).

One idea I had for this flashlight was to also develop an even more ridiculous driver for it (500W multiphase GaN?), with Anduril, and with a much better low mode (the lowest mode for the MK38 is a claimed 40 lumens... quite high, but usable and much better than Acebeam's 200 lumens), but that's probably a project for another time...

I'd also love to compare this with a Acebeam X50, but I don't have one and it's too expensive for me to buy two $300++ flashlights.

[Update - I improved some solder connections and replaced wires with nicer, slightly longer ones, re-cleaned all threads and lubricated o-rings during re-assembly.I reassembled the flashlight and everything is working great; glad I didn't break anything! ]

This post is mostly complete for now; I'll do a few more measurements and edits over the coming days and weeks, so do check this thread from time to time. As always, I'll be curious to hear what questions you have. Let me know below and I'll update this post as I get time to continue my thoughts on this flashlight.

Thanks for reading!

3 Thanks

Thanks for doing this, it’s it your first review on BLF?

My first question would be: what’s the light intensity it can sustain for a longer time? Alternatively a graph of Lumen vs time.
Having that heavy brick in hand I hope to be rewarded for my effort with high constant brightness.

Just waiting for your deep dive. Lets see what those 2 fans do. Cooling a handlebar or a electronics? :slight_smile: Its 3x 21700 in series for 12V ( from manual). Driver DD (easiest way)?

With 3S input to x*6V output it’s very likely a buck driver, probably arround 350W output for 40000 OTF (FL1) lumens, which is already very hard on 3x21700s.

Rather than more power, what would greatly benefits these multi XHP70 soda can lights would be the new XHP70.3 HI, they would produce a much more intense and usable beam.

Edit : the product page actualy mentions that it is a step down (buck) driver.
Edit : Ah I see Loneoceans also guesstimated 350W :smiley:

I’m eagerly waiting for your teardown, especially the USB charging section is of particular interest to me since mine failed yesterday.

thanks for doing this

I see. And it can sustain 9K lumen with fan on. So what power off buck driver :slight_smile: ? I’ll say about 100W at max. And another DD channel for turbo? Someone mentioned wants to build 500W driver! I want to see fans for 500W system :smiley:


Thanks for the post. It got me really interested.

As I searched everywhere, I didn’t see either the official store or any dealer carrying the GT-FC40 option, nor is it listed in any official spec sheet. May I ask where you see that piece of info?

I love High-CRI lights and whether its available in FC40 decides whether I’m gonna buy it or not.

Thank you!

I think loneoceans talk about Acebeam X50 which have Hi Cri version with GT-FC40 LEDs.

Yes, it is particuarly tough on the cells, and most likely any higher power output will yield diminishing gains. But it could be a fun project to try to improve the driver even if it is an impractical endeavour. For example, a multiphase GaN converter would be fun. As for the LEDs, Manker calls it a 'searchlight', though I would definitely not consider the MK38 as a thrower; it's more similar to say the general-purpose SC700d, just a lot brighter.

I'm interested in the USB Charging section as well, and how the 'battery pack' is assembled. Having removable batteries in the 21700-only version seems like a tricky option since the flashlight will be drawing >100W per cell at peak draw, making interconnects and such potentially difficult to implement reliabily.

icpart is correct. At the moment, the Acebeam X50 comes in a high CRI variant. I asked Acebeam directly and they told me that it will be using GT-FC40 LEDs. They're not very efficient, but they do produce a really nice tint as I found in my Lume-X1 KR1 build. On the other hand, the Manker MK38 comes in a throwier variant using Luminous SFT40 LEDs, which supposedly halves the luminous output but doubles the intensity.

Photography Fill Light|AceBeam® Official Store | Flashlights, Tactical Lights


I have no interest in buying the light itself, but great interest in the engineering and assembly/design aspects of all lights, especially ones like this that aren’t the ordinary basics. Looking forward to following this thread! Thanks for putting in all the time to do this.

I would consider myself lucky, if I was ever able to see or handle one of these lights in person.

Thanks for the thread and future updates.

I look forward to the review! Your photography is beautiful as well!

Can you perhaps give more details on the “Micro-Arc” finish? How is it applied, and what makes it special? According to the site, the black version of the light is HAIII, it would be nice to know what (if any) difference that makes besides the color.

I have the SFT40/3*21700 version of the light coming soon for review, I assume it uses a buck driver as well, I’m very interested to see how the two versions actually compare in terms of specs/runtimes.

oooh I want one of these…
I am excited for updates!

Thanks for your comments. I'm not a metallurgical engineer so I can't comment definitively on the subject, but my general understanding is as follows (w.r.t. the treatment of aluminium / aluminum):

  • Untreated - raw aluminium is a reactive element, and quickly forms a natural, hard, and inert oxide layer on the surface in a process known as passivation. This layer is hard but quite thin (several nm thick only) and easily scratched off - in those cases, the oxide layer quickly reforms.

  • Anodized - Similar to the natural aluminium oxide layer which forms when AL is exposed to air (containing oxygen), but the layer of oxide is grown using an electric current while immersed in some acidic electrolyte such as sulphuric acid. For Type III anodizations, the layer needs to be thicker than 1mil (25um) (+ a few other specs I won't get into). This thicker oxide layer is harder than the raw aluminium and provides slightly improved wear resistance and desirable aesthetic qualities.

  • Micro-Arc Oxidation - takes anodization one step further; the process is similar, but a higher voltage is used compared to anodization. This causes an electrical breakdown through the oxide layer, and causing micro discharges / arcs to form. These arcs (which are plasma) increase the temperature locally, causing a change in the way the oxide grows. If done correctly, the oxide is converted morphologically from regular aluminium oxide to corundum (same compound but crystalline), and therefore has a significant increase in hardness, resistance and durability.

Anyway that's my general understanding of the process and I am definitely not the subject matter expert and I can't comment on how well the MAO process is done.

As to how it results in the actual product, I'm not exactly sure just yet (in terms of durability or toughness); all I can say is that it looks really nice in person and has a very pleasant, smooth but matte finish. I also like the colour (a very light grey) very much.

it also feels a bit different than anodized aluminium. Maybe it is just my mind playing tricks on me, but I had the impression it feels slighter thicker, making sharp edges less sharp. Sure, this could also be done mechanically before anodizing, but I attribute this to the treatment for now.

do you have a rough ETA for the next step in your teardown?

I’m really interested in the charging circuit.

Definitely one of the coolest and most inspiring designs in recent years. Low CRI 70.2s might be the worst part about it. It begs for a good tint + high cri combo. B35AM anybody?

Still super cool :sunglasses:

I’m an absolute tint-snob for lights which are used close up. So EDC and Headlamp must have a nice tint and high CRI.

For very high output or throw I can live with low CRI, because the main task of these lights is to either reach very far or just be ridiculously bright, which requires the most potent emitters.

Well said and totally agree :+1:

Personally, I would have to agree with Pobel and Glenn7. It's true that the specs definitely sells the flashlight so I don't fault Manker for going with the XHP70.2 LED here. That said, I think it would be nice to see this flashlight with a XHP70.2 4000K CRI90 variant.

I updated the post with more details about the charger and powerbank circuitry; enjoy!