Maglite Mini 2AA Mod Questions

[yazkaz]

Thanks again, yazkaz.
I am definitely NOT planning on putting the XHP50.2 into there. I am aware it is a 6V LED and was planning on a 2S 14500 build but I'll have to find something suitable to put it in. I take your experience on this very seriously and of course will not attempt such a thing. Since you've discouraged it so strongly.

Looks like I'll need to be picking up an NG1000 at Sandwich Shoppe after all.

I hope this is the right one. It would be nice to use a single 14500 with a spacer if possible though. I was having a hard time figuring out the GD Buck/Boost at SS though. Can I just choose the blank and put in two 0.10 resistors for 1A output? I think that's all that's necessary. It looks like that would allow either 2AA or 1x14500 and the spacer which would be a very nice option to have.

Thanks again so much for the insight. I know I am being very ambitious for a first build here.

If you have a strong desire to use 1x14500 then definitely go for the GD1000. Currently the SS is out of the 1000 so you'll have to configure one from the GD blank board, which is more expensive than the standalone GD1000 when listed. Perhaps you can contact the SS to express interest in the GD1000, so that they can put up a few for you?

Of course you can use ordinary 2xAA with the GD, but output will be a bit less (even with 2xL91 Lithium AA). In other words, the system may not run as efficiently as with one single 3.7V.

If however you want to squeeze as much juice as possible from the AAs then the NG1000 is no escape. I have two MagAAs run on this driver and it simply feels great -- no worry about LiIons when out in the fields.

 

[Reziarfg]

Hmm,
So I've received all the parts and wired them up but got no output to the LED. I'm not sure if I've made a mistake or if there's an issue with the buck/boost converter from sandwich shoppe.

Soldering two battery holders to the LED directly with two NiMH AA in series (2.4V) lights up the LED as expected.

I wired up the converter (GD1000 buck/boost see pic below) J5 to LED+ pad and J6 to the LED- pad then VIN is a short soldered wire going to the grounding screw in the heat sink.
I get continuity between the heatsink and positive battery terminal on the obverse of the converter board. When i attempt to touch the positive battery end to the terminal on the other side of the board and the negative side (still 2.4V) to the heat sink, I get no output.
Have I missed something?

Thanks all.

 

[yazkaz]

Hmm,
So I've received all the parts and wired them up but got no output to the LED. I'm not sure if I've made a mistake or if there's an issue with the buck/boost converter from sandwich shoppe.

Soldering two battery holders to the LED directly with two NiMH AA in series (2.4V) lights up the LED as expected.

I wired up the converter (GD1000 buck/boost see pic below) J5 to LED+ pad and J6 to the LED- pad then VIN is a short soldered wire going to the grounding screw in the heat sink.
I get continuity between the heatsink and positive battery terminal on the obverse of the converter board. When i attempt to touch the positive battery end to the terminal on the other side of the board and the negative side (still 2.4V) to the heat sink, I get no output.
Have I missed something?

I think you've got one connection wrong: The VIn should be empty. Instead, the second J6 should be used for GND.
FYI the J5 and J6 terminals are in pairs (ie. two J5 holes, two J6 holes).
In other words,
J5 #1 -- LED+
J6 #1 -- LED-
J6 #2 -- GND (connect to heatsink)
VIn -- empty

BTW the SS boards require extra care, and that one simple mistake could kill the board. Also for testing, it's absolutely necessary to apply a load to it. Failure to do so and your board will be instantly wasted.

 

[Reziarfg]

Thank you again Yazkaz. You should charge an hourly rate. :laughing:

I was aware that there is no open load protection for this board so all testing was with the LED in circuit (soldered, to ensure constant connectivity).
It looks like my mistake was definitely the wiring then. The way you laid it out makes a lot more sense as it puts the GD1000 in a more obvious parallel circuit with the LED.

I accept that my mis-wiring could have toasted this board though I hope not. I will re-wire and test as indicated.

EDIT: Anyone know if it matters which J6 is used for the grounding screw? Generally buck/boost converters have "in" and "out" contacts that are specific.

 

[nzoomed]

Im wanting to mod my mini maglite too and use 2x 14500 cells instead.
I see some people are getting 1800 lumens with similar mods.
Does this mod allow you to focus the flashlight?
Ive been looking at the driver and think that a buck driver would be better with 2x 14500
Where abouts can I find one?

I dont think that the zener diode option is going to be as efficient?

 

[yazkaz]

Im wanting to mod my mini maglite too and use 2x 14500 cells instead.
I see some people are getting 1800 lumens with similar mods.
Does this mod allow you to focus the flashlight?
Ive been looking at the driver and think that a buck driver would be better with 2x 14500
Where abouts can I find one?

I dont think that the zener diode option is going to be as efficient?

Your problem with approaching the MagAA mod is that you're greedy for output while hoping to keep the light form small. That's NOT going to work.
Remember the MagAA is designed for convenient everyday use, so common battery type and accessibility is one factor. And 2x14500 does NOT fit into that.

Generally you want the MagAA to run with 2xAA or up to 1x3.7V (ie. 1x14500 or 1x14670 plus dedicated dummy spacers). Also maximum current is preferred to be limited to 1-1.2A, nothing more, or else the system will run very hot.

If you insist on running on 2x14500 then your only choice would be a buck driver (eg. SOB driver from the SS) which canoot handle voltages lower than 3.6V. In other words, in case of emergency you won't be able to use AAs to power up the light.

And NO the MagAA mod CANNOT be focused as the front twisty mechanism has been totally defected for custom pill installation. Power is controlled via the Kroll tailcap switch (or similar).

@Reziarfg
Either J6 is fine for GND, although personally I would prefer the right one -- the LED- takes preferential treatment to the first J6.

 

[nzoomed]

Your problem with approaching the MagAA mod is that you're greedy for output while hoping to keep the light form small. That's NOT going to work.
Remember the MagAA is designed for convenient everyday use, so common battery type and accessibility is one factor. And 2x14500 does NOT fit into that.

Generally you want the MagAA to run with 2xAA or up to 1x3.7V (ie. 1x14500 or 1x14670 plus dedicated dummy spacers). Also maximum current is preferred to be limited to 1-1.2A, nothing more, or else the system will run very hot.

If you insist on running on 2x14500 then your only choice would be a buck driver (eg. SOB driver from the SS) which canoot handle voltages lower than 3.6V. In other words, in case of emergency you won't be able to use AAs to power up the light.

And NO the MagAA mod CANNOT be focused as the front twisty mechanism has been totally defected for custom pill installation. Power is controlled via the Kroll tailcap switch (or similar).

@Reziarfg
Either J6 is fine for GND, although personally I would prefer the right one -- the LED- takes preferential treatment to the first J6.

Considering this is a modded flashlight, battery types are not an issue for me, besides most people are using torches that only take 18650 these days anyway. For sure, I would keep an emergency light that accepts common battery sizes.
I really couldnt care less if i didnt get that much light output, but 1000 lumens should be doable and give the same runtime as other 18650 torches out there on the market, possibly more if you are using 2x 14500.

As you say, it would be getting very hot and drawing alot of current at 1800 lumens, fun to impress others with on short bursts, might be nice to have in multiple modes perhaps, but not necessary.

Im currently looking at designing such a pill that would allow a focus and switch off with the head, but means I would have to use a shorter 14650 or 2x 14430 cells to give more space inside the barrel.
I think it can be done, but ideally you would want the pill making more surface contact with the barrel to sink heat away rather than just contacting the edge where it makes contact to switch the torch on.

Just been looking at the SOB 1000 driver, looks quite a good driver. Trying to find the datasheets on it now. Does it allow different brightness modes with memory if you turn on/off?

 

[yazkaz]

Considering this is a modded flashlight, battery types are not an issue for me, besides most people are using torches that only take 18650 these days anyway. For sure, I would keep an emergency light that accepts common battery sizes.
I really couldnt care less if i didnt get that much light output, but 1000 lumens should be doable and give the same runtime as other 18650 torches out there on the market, possibly more if you are using 2x 14500.

As you say, it would be getting very hot and drawing alot of current at 1800 lumens, fun to impress others with on short bursts, might be nice to have in multiple modes perhaps, but not necessary.

Im currently looking at designing such a pill that would allow a focus and switch off with the head, but means I would have to use a shorter 14650 or 2x 14430 cells to give more space inside the barrel.
I think it can be done, but ideally you would want the pill making more surface contact with the barrel to sink heat away rather than just contacting the edge where it makes contact to switch the torch on.

Just been looking at the SOB 1000 driver, looks quite a good driver. Trying to find the datasheets on it now. Does it allow different brightness modes with memory if you turn on/off?

Simply put, you should not compare apples with oranges. 18650 lights is one thing: AA lights is another.
Plus, how can 2x14500 give "the same runtime" as other 18650 lights? Besides, how much capacity does a single 14500 have??

"Fun to impress others with on short bursts" -- a flashlight is for usage, NOT play. What you said is merely useful for LE uses, that's pretty much it.
Do this in a civilian way and you could get yourself into trouble......

SOB1000: the specs are already on the SS web. NO multimode, NO latch memory. Further questions, please contact SS.

 

[nzoomed]

Simply put, you should not compare apples with oranges. 18650 lights is one thing: AA lights is another.
Plus, how can 2x14500 give "the same runtime" as other 18650 lights? Besides, how much capacity does a single 14500 have??

"Fun to impress others with on short bursts" -- a flashlight is for usage, NOT play. What you said is merely useful for LE uses, that's pretty much it.
Do this in a civilian way and you could get yourself into trouble......

SOB1000: the specs are already on the SS web. NO multimode, NO latch memory. Further questions, please contact SS.

Firstly, where did I say I wanted to have a "toy" to "impress" others? I was just stating a fact, based on someone elses mod. Im not interested in something that bright.


Anyway, Im obviously wanting something as bright as possible, but with the most practical battery life.

I know you are not recommending the XHP502, but Im only going to run it at 500mA instead of its maximum 3A rating, which according to the datasheet would give me round 1000 lumens and at least 2 hours or more of run time on a pair of 14500 batteries.

Been looking at the specs for the Nichia and it seems it takes 6000mA to give its max output of 1300 lumens, so does not look very efficient with those numbers.


Would be nice to find a driver with different brightness modes

Ive decided i will just do this mod and find a suitable tail switch for now.

And as far as the SOB goes, all the links on their page are dead
see here for yourself
http://www.anlighten.com/shop/test.php
https://my.xfinity.com/~theledguy/cpf_store/sandwiches/techical_data.htm

Anyway the upside is i can use the lower current driver and go with the SOB 500, not sure what is more efficient though? Use a driver that runs at its max output or a higher rated driver running under less of its maximum rated load?

 

[Reziarfg]

@Reziarfg
Either J6 is fine for GND, although personally I would prefer the right one -- the LED- takes preferential treatment to the first J6.

Hypothetically, would there be any issues if I had accidentally shorted the two J6 pads and left them that way? They look like they're electrically continuous. Other than maybe a small effect on resistance I don't think there should be any ill effects.

I should have some pics up as early as tomorrow. This has been a long on and off project for me but I'm finally getting some results!

 

[yazkaz]

Hypothetically, would there be any issues if I had accidentally shorted the two J6 pads and left them that way? They look like they're electrically continuous. Other than maybe a small effect on resistance I don't think there should be any ill effects.
I should have some pics up as early as tomorrow. This has been a long on and off project for me but I'm finally getting some results!

No comment on the J6 pair since I'm an electronics noob. But just make sure nothing is shorted and you should be fine.

I agree these builds could take a long time to complete. Which is why I'm thinking of offering some completed module pills for sale. Module specs will be restricted to limited driver board and LED choices. Also they won't come cheap as not only premium materials are used, but also the labor cost for professional module assembly.

 

[Reziarfg]

Alright! After much tinkering I have managed to put together something resembling a flashlight :twothumbs

It's a bit late so I will try not going into excruciating detail. This was my first build so it's definitely not perfect.

Of my original parts list, here's what changed:

1. Driver: 0.55 inch (13.97mm) diameter driver from SS. Custom blank with 1A constant current
2. Emitter: Nichia 219C 4000K reflowed onto Sinkpad XP 10mm MCPCB
3. Switch: Terralux TCS1 Tail Cap Switch I found online. Might go back to the Kroll though
4. Power: 1x 14500 Nitecore NL1485 OR 2x AA NiMH OR 2x Alkaline AA

So basically my original idea of stuffing a XHP50.2 with two 14500 was shear heat-generated madness.

So without further ado here is the build. I tried taking some pics but I'm a bit of a noobie on this so..follow any of this at your own risk.

Some Kapton tape after soldering the 26AWG silicone wire to the pads:

Seeing how it fits:

Closed circuit test (no open circuit protection so this is how I tested the electronics). Also I put some black RTV silicone gasket maker in there before final installation but don't have any pictures of that:

How it looks inside the tube. You can see my really terrible sanding job to get the copper heatsink flush for proper fit of the MCPCB:

Here it is soldered in place after snipping the silicone wires. Of note is that the silicone wires were a pretty tight fit when screwing in place. I applied a bit of SuperLube to the copper threads though I don't know if that was a great idea or not:

Another angle:

Here is the front. I probably need to trim the reflector height a bit. Yazkaz mentioned this but I haven't gotten around to it yet. There is a small gasket in there as well to provide insulation between the silicone wires and reflector:

Not too bad:

Another angle:

Nitecore 14500 and dummy AA:

Beam shots!
Both beam shots below were taken on my Samsung Galaxy S8 at 1/30s and ISO 200 from approximately 3 meters away against the top corner of my ceiling where it meets a brick wall. I also set the white balance manually but don't remember to what temp.
Also just realized I don't have a before MagLite shot. I think I might have a stock one in working condition lying around somewhere I might edit this later as a point of reference.

MagLite Mini Mod below. I still may be modifying the reflector to smooth this out a bit (the pattern there is likely the gasket sitting too far from the LED). I also have some d-c-fix I picked up in case I wanted to add that on so I might do that:

And a SST-20 4000K Emisar D4 at highest regulated for comparison:

I haven't done any multi-meter measurements yet so I don't know exactly what it's drawing. But if the Sandwich Shoppe driver delivers the 1A constant current as it claims we should be seeing ~353 lumens as Texas_Ace reports here
According to ToyKeeper here the D4 with Nichia 219C at highest regulated output should be at 150 lumens. Even though mine is the SST-20 4000K model they output similarly and I can't find a ramping lumens table for the SST-20 so that's the best I've got so far.
Also tested with Alkaline and NiMH AAs and there's no perceptible change in brightness. Which makes sense with this buck/boost driver.

Fun fact I learned along the way:
There is a reasonable variance in diameter in the AA spec. Wikipedia lists the diameter at "13.5–14.5 mm". Though all the batteries and tubes fit within this spec, there's just so much slop. Inner diameters on MagLite bodies from the 90s are narrower than modern ones and won't fit anything but Alkalines and unprotected Li-Ions. Since the SS driver doesn't support LVP, I really wanted to use a protected cell.

For anyone looking into a project like this, the Nitecore NL1485 has so far been the only protected cell I've had fit the tube without trouble. The KeepPower 14500 cell could be coaxed in but it was a tight fit and takes some shaking to get it out.

I measured the following with my calipers (narrowest to widest):

Device	Diameter
Dummy AA	14mm
Alkaline AA (Energizer)	14.01mm
Efest IMR 14500 650mAh	14.135mm
90s incandescent MagLite tube	14.19mm
Nitecore NL1485 14500 850mAh	14.198mm
IKEA Ladda NiMH	14.32mm
Modern Xenon MagLite tube	14.4mm
KeepPower (EVVA) 14500 800mAH	14.4mm

And that's it. It's late and I'm turning in. Looking forward to some questions, feedback, and advice. Cheers!

 

[yazkaz]

@Reziarfg
Excellent and detailed documentation! Feedback below (in point form and in no particular order):

a. Sinkpad: it's recommended to use the ~8mm type such that the module itself can be easily removed for another MagAA application. With the 10mm pad it means (as shown in your procedure) the module assembly had to be done within the flashlight host, which renders the module setup permanent. The beauty of a module built upon this copper sink, however, is easy adaptability onto just any incan MagAA host due to easy assembly and disassembly (ie. when assembled properly);

b. Wiring: typically AWG26 is recommended, although AWG24 is usable. PTFE wiring is even better, but some claim it's more challenging to handle (due to increased difficulty in bending). Which wiring spec did you use?

c. Optics: which reflector did you use? It's true that the optics height needs to be maintained within 9.9mm so as to give the head assy enough threads to screw onto the body. Too high the optic and the head assy won't screw in properly, resulting in severe wiggling. On the other hand, modding the stock plastic reflector may not be a good idea, as effective focusing is not guaranteed;

d. Electronics insulation: I can see the importance of Kapton tape usage inside the driver, but I've been told it's not always a good idea, as effective heatsinking may be affected. The best way is to completely pot the internals. Sure, you may not be able to reuse the driver once fully installed, but that's the way the original sandwich-type module is being made;

e. Tail switch: I too have used TCS-1 before but don't always like it. So I gradually revert back to the Kroll. There's also reliability problem with the TCS-1, as prolonged use will result in eventual switch failure. And there's no way you can fix it even if you can crack it open.

OTOH, the problem with the Kroll is that the side spring may not latch onto the tail threads properly, causing the switch to fail. In this case, you need unscrew and rescrew the Kroll to turn back the light on. This is particularly the case when the tail threads are being meddled with. Say, if you lap the threads (even with fine wet sanding), even with just one turn would take off material enough to increase the chance of Kroll switch failure. So, even if the MagAA's tail threads feel coarse to the touch, better leave that as is;

f. I've been told that, depending on the quality of the aluminum (or other metal material), a dummy spacer may increase internal resistance and thus decrease the overall light performance. However, should anyone end up with this route, there isn't much one could do;

g. Battery: I've used (and still using) AW's 14500 and 14670 (with corresponding spacers of course) on my MagAA setups and most of them fit into the MagAA body without problem. Not quite so with Keepower cells though. The other day I was testing Keepower's 14650 and find that it won't fit in unless with the label peeled out. Then I start to realize that most of Keepower's cylindrical LiIons are wrapped in such a way that the total diameter is always a tad thicker/wider, thus causing fitment problem in various flashlight setups;

h. GD1000 performance: per SS'es claims it's best to run this driver with one single 3.7V LiIon. Ordinary AAs and NiMHs are perfectly OK, but performance may suffer a bit due to these cells' internal resistance. On my GD1000 MagAA setup I can eyeball the performance difference between 1x14670 and 2xAAs (either L91 Lithium or NiMH). For optimal performance with 2xAA however, NG1000 is recommended (but that it probably cannot run on 1x3.7V due to the low LED Vf, and that the over-voltage threshold on the NG is just +0.7V).

Also too bad you had to order the GD1000 in custom form (and likewise, the added cost). SS usually has a few GD1000s online but these days that's not the case. So if you need a GD1000 again, always inquire the SS to see if they can put up a few online, in advance.

 

[Reziarfg]

Some of these points I addressed in my post and some come from my original post. See below for a few responses.

a. Sinkpad: it's recommended to use the ~8mm type such that the module itself can be easily removed for another MagAA application. With the 10mm pad it means (as shown in your procedure) the module assembly had to be done within the flashlight host, which renders the module setup permanent. The beauty of a module built upon this copper sink, however, is easy adaptability onto just any incan MagAA host due to easy assembly and disassembly (ie. when assembled properly);

I agree with you on this mostly. I wasn't able to find an 8mm sinkpad. The hole-to-hole distance on the 10mm one is <8mm though so it's not terrible but yes I would have liked not to file down the copper ring and for it to not be permanent as you say. Is there a place to source 8mm sinkpads? I might be doing this again in the future with another MagLite host I have.

b. Wiring: typically AWG26 is recommended, although AWG24 is usable. PTFE wiring is even better, but some claim it's more challenging to handle (due to increased difficulty in bending). Which wiring spec did you use?

I started with AWG24 which was a bit thick. The red wire is actually AWG24 silicone sleeved. The black ones are AWG26. It was much easier to work with the 26 gauge.

c. Optics: which reflector did you use? It's true that the optics height needs to be maintained within 9.9mm so as to give the head assy enough threads to screw onto the body. Too high the optic and the head assy won't screw in properly, resulting in severe wiggling. On the other hand, modding the stock plastic reflector may not be a good idea, as effective focusing is not guaranteed;

Reflector is the McR-18 Reflector which admittedly isn't perfect but it's not too bad either. It does screw in ok. I think once I tweak the gasket a bit more I can get the LED to fit better to reduce the pattern in the beam. I'd be open to better suggestions though.

d. Electronics insulation: I can see the importance of Kapton tape usage inside the driver, but I've been told it's not always a good idea, as effective heatsinking may be affected. The best way is to completely pot the internals. Sure, you may not be able to reuse the driver once fully installed, but that's the way the original sandwich-type module is being made;

Check out a note I left on one of the early images above! I did buy some black RTV silicone gasket maker and used that to pot the internals in addition to the Kapton tape. I suppose in retrospect it wasn't necessary to do both.

e. Tail switch: I too have used TCS-1 before but don't always like it. So I gradually revert back to the Kroll. There's also reliability problem with the TCS-1, as prolonged use will result in eventual switch failure. And there's no way you can fix it even if you can crack it open.

I have both TCS-1 and Kroll. I'll probably switch the TCS-1 out though since I don't like the button as much. Both seem to work fine at the moment.

Thanks for all the support through this! It was a blast to put together.

 

[yazkaz]

@Reziarfg
I'll address the optic and potting here...

Reflector is the McR-18 Reflector which admittedly isn't perfect but it's not too bad either. It does screw in ok. I think once I tweak the gasket a bit more I can get the LED to fit better to reduce the pattern in the beam. I'd be open to better suggestions though.

Having the McR-18/S to work is tricky. To me it's the readjusted height of the reflector. Sure you can machine it down from the frontal side (back side not recommended as you don't want to increase the smaller bore diameter), but the metallic dust during the process is likely to permanently mar the reflector's surface. The marring is not normally noticeable until after the light is turned on. If however such marring can be avoided then the reflector will do a good job to yield a very round beam profile with decent (if not stellar) hotspot.

The other thing is the machining quality of the reflector. Turns out not every one is equal to this respect. And you won't notice the flaw until after it has been applied to the light setup and powered on. One of my local machinists told me that if an ellipsoidal beam is noted, then the reflector's QC is likely at fault.

Potting: most flashaholics use Arctic Epoxy (A+B) but the pros at the manufacturing side do otherwise. Was told a special heat-dissipating 3M epoxy is being used, but I forgot the model code.

 

[Reziarfg]

I think when it comes to the reflector I'll just have to get it "good enough" for me. I'm not super picky and other than the hotspot pattern above it seems acceptable.
I do recommend if anyone else wants to build this, find a better reflector. This one is a bit narrow, a bit tall, and if I don't screw in the front just right it doesn't light up at all. But I'm not experiencing any wobble or looseness once it's screwed in and since the front doesn't generally need to be removed, it's acceptable until I find a better option. Thankfully the reflector is easy to swap out once I do.

@Reziarfg

Potting: most flashaholics use Arctic Epoxy (A+B) but the pros at the manufacturing side do otherwise. Was told a special heat-dissipating 3M epoxy is being used, but I forgot the model code.

I considered various thermal epoxies when deciding on a potting solution but I didn't want it to be permanent yet since it's my first build and I wasn't sure it would work properly. I know black RTV silicone isn't the ideal candidate for thermal conductivity, but I think it's sufficient for the heat generated by a Nichia 219C at 1A (should be ~3W or maybe 3.5W accounting for efficiency losses?).