How about SinkPAD? I believe Meodex has some triple, but they are aluminum. I am interested in a light with the triple in series, too. Those MCPCB’s are star, not round. The size should be close enough, perhaps some sanding. Where to source them could be an issue.
Annyli fhashlight, did you try to send him a PM using the forum? You can try this: https://budgetlightforum.com/messages/new
This is how I manage to message to forum members.
So BLF, I finally had an opportunity to purchase a Zebralight SC700d.
Expensive, even with a good deal. But since my goal is to develop as good a flashlight (driver) possible, I figured I needed to have at least a good benchmark to challenge myself against.
The Zebralight SC700d is often considered one of the best flashlights on the market not only in terms of its physical construction and machining, but also its driver efficiency and performance, as well as its thermal design and durability. I have to say that my first impressions are very good. The flashlight, despite being a 21700 light, is extremely compact, even smaller than the 18650 KR1, and noticeably lighter as well (even with battery) - almost makes the KR1 look clunky! The machining, anodization and finish, are one step above the competition, even when compared to Noctigon / Emisar flashlights. The flashlight is unibody, with the only screw thread at the base - threads are fine-cut but anodized, precise, well lubricated. Switch design is well thought-out - recessed but tactile. I could go on, but that's for another post.
The bottom line - just using it for a short while really validates all the work I've been putting it for high efficiency switch-mode drivers. The performance of the ZL is really noticeably better than cheaper flashlights using linear drivers. I understand the definite need for lower cost flashlights (e.g. $30 or less) - for those flashlights, a linear driver is the sensible choice and is here to stay.
However, having used these switching drivers, I personally find it difficult to justify paying for a flashlight over ~$50 which still uses a linear driver; it's like purchasing a nice sports car with amazing bodywork, interiors, sound systems and tricked out suspension, but with an engine from a compact sedan.
One of the reasons why I purchased this was because the SC700d is also considered to be one of the contenders for 'King of the Moon Mode', having exceptionally low moon mode.
Here you can see the ZL SC700d on the left at its lowest mode (which is, in fact, very low, especially vs. almost all AMC7135 drivers). However, it's very bright when compared with the Lume X1, even when the Lume X1 is running at around ~10x its minimum brightness level (potential to go even lower)! This image was taken at ISO1600 f/4 1/10s and illuminated by early dawn light. I was glad that this test also validated the UDR performance of the Lume X1 driver.
For top-end performance, the SC700d with its more efficient but lower CRI LED (still high at 90 CRI though) XHP70.2 performs very well, producing 3000 lumens (from ZL's spec sheet), which should correspond to something like ~22W output. The KR1 with the Lume X1 and GT-FC40 reaches a peak brightness (from my eye) a little less than the SC700d, but this tells me that the Lume X1 performance is higher than the SC700d driver since the GT-FC40 is extremely inefficient (but CRI 95), and I was expecting ~2500 lumens after the optic, with about 33W power drive. Running the XHP70.2 with the Lume X1 should produce north of 4000 lumens without issues.
At this point I understand better why the XHP70.2 is a controversial LED - while its output efficiency is undoubtedly high, the tint of this 5000K 90 CRI emitter in the SC700d exhibits a greenish tint (compared with the peachy white of the FC40), and worse, exhibits moderately strong greenish yellow tint-shifting from the OP reflector. The center of the beam is pleasant but it falls off quickly into a frankly fairly ugly corona. Outdoors, this is not a problem. But indoors it's very noticeable and is the definite minus of the SC700d. I can't help but think that something like a Ledil Olga in the SC700d would improve things significantly (but too bad the Olga optic doesn't fit as it's a fair bit smaller). Alternatively, dedoming the LED could produce a better beam profile, likewise with the addition of a simple green-reduction filter / film.
Side note - I tested a few different ramping tables and I found that I did like x^4 curves a little better than x^3 or some others I tested, so I'll be testing that out for now.
annyli fhashlight, yes please feel free to drop me a PM and I'll be happy to see what your request is.
I’ve wished for years that some Chinese manufacturer would attempt to compete with Zebra quality, maybe even beat it while offering good LEDs, BLF UIs, higher outputs and no glue…
You drivers bridge a part of the gap. It is still wide in other areas though.
I’ve been wondering how a casting process would pay off for unibody or larger lights. Nitecore does it. Really opens up possibilities in thermal design while managing machining overhead. Requires volume to pay off I’m sure.
I spoke to some manufacturers and they seem open to the idea of unibody designs. The only concern is that it is a little more expensive to machine since regular flashlights can be machined from short stock for the head and pipes for the bodies, as opposed to a single long stock with a lot of material cut away.
Today I managed to open the ZL SC700d, and reassembled it back after (it's challenging for those without the right tools so I don't recommend anyone doing so without a risk of damaging it) and it's nicely designed; they used a PIC microcontroller, a TPS61088 regulator and PCB-side-mounted e-switch, all on a single but fairly large PCB (basically one inch square).The main inductor is a Coilcraft 7030 1uH, a perfect choice for the 61088. The design is very similar to most of the switching driver circuits including the Lume drivers, so I guess I'm glad that this shows the Lume is on the right path. I can see ZL swapping out the 61088 for the 61288 for future flashlights for even more power, but I think the Lume X1 is still the first driver to use the 61288 !
The LED is mounted in a channel cut in the PCB where a copper block provides thermal contact and heatsinking to the body. The regulator and inductor are potted, but in a soft silicone. It's compact, but perhaps too compact - and here comes my main criticism. The design does come at the cost of difficulty of reparability and warranty for the product, though it's robust enough that I think the failure rate is going to be quite low.
This also makes LED swaps difficult since the design is intrinsically package and footprint specific. It doesn't really save much room either since it uses a separate PCB as the positive-battery-spring contact (using pogopins), so it still uses 3 boards in total, just like regular flashlights if you count the MCPCB (driver, positive spring, tail spring), but it does improve thermal performance.
Regardless, I think there is still some value in a 3-piece design (head, body, cap) - it's cheaper to machine in general, and I guess it's required for easy re-flashing without using a ridiculously long pogopin programmer (going with the AVR-1 and 3-pin UPDI makes this almost possible, though). I'm not sure how many manufacturers will want to go with the casting method - perhaps there are other possible options but I'm not familiar with them. The unibody does improve on thermal sinking, though.
Unibody is one thing, the quality of the machining and anodization is another. I think so far Emisar/Noctigon comes the closest. I think the main thing that strikes me about the ZL is the economy of design - it's so compact and tidily built and I see no reason why other flashlights can't be designed similarly even if the machining isn't as good. I'd like to keep the LED and MCPCB swapping though, since that's really part of the ethos of the BLF community, and I think all these are not difficult to achieve at the same cost of machining.
The front bezel is a typical interference fit - very compact and well machined. I tried one quick attempt to pry it with a small pry tool, but the fit was too tight - didn't want to damage the ring or the flashlight... it is a $120 flashlight after all!
So, I heated up the outer shell with hot air and cooled the ring with an aerosol freeze spray. Then immediately, one sharp blow to the ring with a small hammer and I got the ring out. The glass didn't break but perhaps I was lucky. I have a fair bit of experience taking apart consumer electronics so this is probably the most tricky to do, but not too difficult with some experience I think.
The reflector looks custom and is very well machined - fits perfectly in the body, and the hole at the bottom is surprisingly big. There is one big O ring around the lip that creates the waterproofing seal and it's compressed in very well. However, you can't just press the glass down to move the O ring because the O ring looks like it's being compressed radially instead of axially, so there is basically no play in the bezel ring fit in the z-direction. The tight tolerances of the ZL machining help make this work. I was hoping to replace the reflector with a TIR optic to fix the tint shifting, but none I could find fit without modification and maintaining a good deal, so I decided to just leave it be just to keep this one as original as possible. I figured that the ZL SC700d makes for a great 'benchmark' light.
Removal of the board is more tricky because it's potted in, and one of the main connections is soldered down to the spring PCB with a solid metal post instead of a wire. The spring PCB is glued down and impossible to reach with a soldering iron, but I managed to desolder the joint. I used a very high power Metcal iron and I'm not sure if a typical soldering iron would do the job without risk of damaging the joint on the other PCB or damaging the plated TH pads.
Potting was the easiest to remove. I flooded the bay with isopropyl alcohol and used needle tweezers, and it came out safely.
The only thing I think can be done to mod this flashlight is to either replace the LED with another XHP70 (it's in 6V configuration), but I suspect most people will have trouble replacing it because it's bonded in two parts (half to the PCB and half to the gold-plated copper block, so reflow is going to be tricky if you don't have the right tools), or an easier mod would be to de-dome the LED to reduce the tint-shifting (likely you get 1 good attempt at it!), or to add some sort of filter film to the glass.
Fitting it back requires care as well. I cleaned up the internals, re-tinned and refluxed, and resoldered on the main connection to the spring PCB. I also cleaned up the silicone potting to make sure there wasn't any contaminants or loose pieces of potting, and reassembled it back, taking care to reprep the surfaces with thermal paste (I used arctic silver). The two screws serve both to clamp the LED onto the body, as well as a high current path for the negative terminal, so the screw are in fact soldered on. Those were easy to desolder, though. I resoldered them back during reassembly.
Finally, I relubricated the seals and pressed the ring back in with an arbor press, but I did file a very tiny notch in the bottom of the bezel ring with a micro file just enough for a pry tool, in case I wanted to reopen it again more easily.
I've been using the ZL for a short while and I'm starting to like the UI more. It's not as fun as Anduril for sure, but it definitely is very practical and works really well 'in the field' as a utility tool.
For a moment I thought about building a driver to replace the ZL driver with the Lume X1 design with Anduril, but it's challenging enough for most folks without the right tools to mod that I'm not sure if it would be practical. A ZL with Anduril though, that would be fun!..
How about a kapton PCB overlay to translate the format to 5050 or 3535?
PCB would move the anode and cathode closer together and leave the center open, so the thermal pad is soldered directly to copper? Bond line would be thicker but it should still work OK.
Thanks for sharing, it’s the first time I see this trick. Previously I’ve seen people opening Zebras by cracking the glass and prying on the ring from below.
You don’t need very long pogo pins to flash unibody light. The programming pads could be placed at the front of the head instead. Or one could use USB for that.
As to swaps…years ago, it would be easy to swap pills or drop-ins. Inside a pill or a drop it, it would be quite easy to swap the driver or the LED. With a simple spacer, one could also turn that to a triple or quad; spacers were available for the popular formats.
Nowadays, most lights come with solid shelves, so no pill swapping.
Drivers of e-switch lights are often of non-standard size which makes swapping harder. I’m not sure if I’ve seen an Emisar with a swapped driver before Lume X1 (though I’m not sure whether the RGB Emisar D4 had a heavily modified or replaced driver).
Keychain lights often choose LiPo cells and these always have custom drivers as well as custom LED PCBs (metal or not).
We’ve got over that, I think.
Should we give up LED swaps? IMHO - not easily. But the question is whether moving the driver to the front of the head precludes the ability to swap LEDs - and I think the answer might be no. Could driver be placed on a flex PCB wrapped around the optics, freeing lots of space for a regular MCPCB? Or driver overlaid on top of a regular MCPCB? Leave 10 mm in the middle of the light for a regular PCB and put the driver as a ring around it? I don’t know.
At the same time - it the driver-in-the-front strategy worth pursuing? If Zebra couldn’t figure out a good way to avoid a PCB at the battery positive contact, maybe there is no easy to manufacture replacement that would be more compact and as performant?
BTW, from my limited experience Lumintop anodization is more durable than that from Emisar. But I have only 1 Lumintop to compare (GT Mini).
BTW, Zebra’s approach to weight is the opposite from what most budget brands do. Zebra cares a lot about making lightweight lights. Budget manufacturers tend to prefer hefty.
I understand how huge shelves help Emisar or Fireflies increase Turbo duration. I recognise that they still get criticised for Turbo being too short. It was their choice, performance-to-size at the expense of weight.
The only budget maker that chose to make his lights small and light is DQG. But DQG is at the opposite extreme with his lights being poor performers.
I so wish there was a brand that would try to make their lights as small and lightweight as possible while maintaining proper thermal design…I’m a huge fan of DQG but I sometimes miss higher output in his lights.
A couple of years ago I made the GXF22 which was a 10A linear + FET driver for my Emisar D4 - https://budgetlightforum.com/t/-/57383, and fitted it with four Luxeon Vs. It looks like the newer Emisar / Noctigons are now using the same linear + FET configuration, but I didn't continue any development on that project because it was just... inefficient and I didn't see significant added value over cheap AMC7135 drivers other than the fact that it is PWM-free.
It's possible to make E-switch drivers swappable, but I imagine it would only apply to side-switch flashlights with a separate e-switch board. (I have to say I much prefer side-switches though, much more ergonomic IMO). There just needs to be a few standards which I suppose is challenging...
Personally I prefer a slightly lighter flashlight. I bought a copper FW3C for my Lume1 project and while it does feel nice in the hand, I've wound up not carrying it around much because it's just too heavy in any sort of pocket. Including the battery, the FW3C is 16g heavier than the much larger ZL SC700d with 21700 battery.
Regarding swapping the LED, keep in mind that the driver is configured for 6V. Perhaps you could swap it out for a XHP50.2 but I don't think you're gaining much there for the amount of work. Likewise the 144AM could possibly fit but at the expense of significant loss of thermal performance; best to just design a driver-led-sink assembly specifically for the flashlight. If I manage to get another SC700d for some reason it could be a fun project.
Ha! So my idea does work… Awesome work getting this open AND reassembled cleanly!
Do you really find the SC700 an EDC-able size? I personally set a soft cutoff at 26mm dia and 105mm length and a max of 28x120. Ideally its 23x95, personally.