Triple MC-E 0D Maglite based bike light, MADE!

I wanted to build a really bright bike light with lots of runtime, and I wanted to do it with all the parts and batteries I had already purchased for, or salvaged from other projects.

The end result is a Mag 4D host mounted to the frame and used as the battery carrier, and a Mag 0D host housing basically everything else and mounted to the handlebars. Both hosts still have their switches.

For the LEDs, I had three cool white MC-E that I had pulled from three MC-E modded lights to upgrade them to XM-L. I mounted all three onto a triple XRE star, wired as 4p3s.

Now how to drive 'em like that? I decided to see if the DX 3-mode 3A XML driver, sku 57779, which is capable of up to 18v input could drive three in series. Answer = yes.

Now how to keep 3 MCE at 3A cool? I combined those components with a huge Aluminum heatsink that snugly fits the Mag D neck. I then drilled thru-holes for screws and wires to pass and tapped three holes into the bottom of a DX MOP tri-flector, which I'd already modified the threads on to fit a Mag head. The result is the tri-star sandwiched between the triflector and the heatsink. Thermal grease was applied between the heatsink and star and heatsink and interior of the neck.

Now, how to drive 3 MC-Es in series at 3A and still have decent run-time? Enter 4 UltraFire 32600 LiFePO4 4AH batteries and a Mag 4D host. The big giant host might not appeal as a hand-held, but here's the cool part...

This will be a convertible bike light! My plan is to take a Mag 2D of the same color and cut the top section, including the switch section and install the completed head assembly to that. The 4D body will remain unmodified and will mount to the bicycle's down tube with Maglite's D-cell wall mounting brackets, secured to the frame by the bottle carrier bolts. 16 awg speaker wire with plugs will connect top of the switch of the 4D body to the bottom of the switch in the cut 2D body. The cut 2D body will mount to the handlebars. Both switches will remain operational (both must be on and either can be used to select modes). The point is that the head can be moved from the part of the 2D on the handlebars back to the 4D body at any time with relative ease.

This way, I've got mega runtime on low and car high-beam brightness on high. Not sure I'll ever need strobe unless I feel I'm in danger of being hit and need a driver to take notice of me.

Photo journal of the build...

Assembled light engine with giant heatsink


Thru-holes viewed from bottom of heatsink to screw tri-star to triflector thru the heatsink so all three are sandwiched together:


The face after assembly:


Fully assembled light, for now:


LiFe batteries provide ample power and runtime


Tailcap draw on low


...on strobe


...draw on high



Beamshot comparison with other triple floody buddies:

Triple XML with triple DX optic:


Triple 4500K SST-50 with SMO triflector:


This light - Triple cool white MC-E with MOP tri-flector:



*Edit*

Cut down a 2D to a 0D and did what I planned...






Pics of it installed on the bike are coming up...

Re: Triple MC-E bike light MADE!

Installed!

I installed the 4D body with the Maglite D-Cell wall mounts where the 2nd bottle carrier would go. For the 0D body, I used hose clamps. It's a crude solution, I know, but they might help spread heat to the handlebars and I hated the way the rubber & velcro light mount I had would allow the light to bounce and jiggle over every bump.

Re: Triple MC-E bike light MADE!

And some more pics...







Here's a view inside the first head of the hi-amp quick connectors that allow the 4D body to be removed and replaced easily for carrying the cells inside to recharge:



Full blast during sunny daylight:

Just took it for a spin now that it's dark out... WOW!

I'm VERY pleased with the result.

Here's a YouTube video taken from one of my surveillance cams of my old bike light in action on 7/25, and then this new one on 7/29 (video is timestamped in top left). This new light appears ~38 seconds in.

Techjunkie said:

Here's a YouTube video taken from one of my surveillance cams of my old bike light in action on 7/25, and then this new one on 7/29 (video is timestamped in top left).

Click to expand...

Insanely bright man. Awesome build!

Any further info. on how the LED's are mounted to your awesome heatsink and what you made that heatsink out of? Did the mag need any boring for those cells? In your very first pic, in the upper left, is that something used to bore the inside of mags with a drill? Are the LEDs all on one star of three separate stars?

simplec6 said:

Insanely bright man. Awesome build!

Any further info. on how the LED's are mounted to your awesome heatsink and what you made that heatsink out of? Did the mag need any boring for those cells? In your very first pic, in the upper left, is that something used to bore the inside of mags with a drill? Are the LEDs all on one star of three separate stars?

Click to expand...

The heatsink is 1 3/8" T6061 aluminum bar that I had a machinist turn down to the same diameter as the internal diameter of a bare Mag D neck. The thickness of the anno on the inside of the Mag necks varies from color to color, so I remove the anno with a brake hone and grease the neck with Arctic silver 5. before pressing in in place. No honing is necessary for the batteries to fit. They're not even snug.

The LEDs are surface mount soldered to a triple LED star like this:



That's actually the same tri-MCE on the right in the pic above.

I don't rely on the small trace routes to carry the 3A of current though. I reinforce them with 22awg copper wire, like this:


To mount the star to the heatsink, instead of screwing the star down from the top, I use the triflector to push the star down onto the heatsink. That's done by drilling the holes smaller on the top of the heatsink than the bottom. The threaded portion of the screws enter the bottom of the heatsink, pass through the star and catch the threads of holes I tapped on the underside of the tri-flector. When the heads of the screws reach as high as they can go on the inside of the heatsink, the screws pull the triflector down onto it, squeezing the star in between them. Small washers shim the space between the star and the triflector to adjust the focus and elminate some but not all of the donut effect of the MC-E. There's Arctic Silver 5 between the star bottom and the heatsink top.

The following pics from a similar triple SST-50 build that I did a while back explain the mounting a little better than I did with words above:

3 holes tapped in triflector


Srews seen from beneath the heatsink:


Everything screwed together as seen from the side:


Unlike that 2D triple SST-50 build, the larger heatsink and the lower power emitters and larger power supply allow this one to run on full blast for 1.5 - 2 hours without getting crazy hot.

On low, this light could go for 8 hours, and low is still pretty impressive at that drive current (~600mA per 4p, or 150mA per die * 12 dies.)

Thanks for that, i've been digging around trying to work out what the draw was on low for this driver, looks like very roughly 20%.

nice bike
was it made by Raleigh?

Here I was out for a ride tonight wondering if my frame mounted XM-L Solarforce was too bright?:thinking:

nanotech17 said:

nice bike
was it made by Raleigh?

Click to expand...

Thanks. It's made by Kona. The model is "Dr Good". I think the closest Raleigh model to it is the Cadent i11, which is admittedly a big step up from my Dr Good and even a step up from the higher Kona hybrid-commuter model, Dr Fine. I did quite a bit of comparison shopping and the Dr Good checked every box I had and came in at the lowest price. My requirements were:

Commuter-style bike i.e. hybrid (MTB-style) frame, but with no suspension, 700c wheels, flat bar (in this case, the handplant V shaped bar)
Lightweight (bike without torch is 25-27 lbs.)
Road tires (700x23c)
Disc brakes, 'cause my rims never stay true (technically, my rear brake is a cable-pull hub brake)
Internal gear hub (IGH)

I found about 7 or 8 bikes that fit the bill. Some had fatter tires that I'd have had to change, like the Cadent, some had nicer hydraulic brakes instead of cable-pull, all the others had the newer Shimano Alfine 8-speed IGH (the Cadent i11 has the new 11-speed hub), whereas mine has the older Shimano Nexus 7 speed IGH, but all the others ran about double the cost my Dr Good, so overall, I think I chose very well.

Lightdadark said:

Here I was out for a ride tonight wondering if my frame mounted XM-L Solarforce was too bright?:thinking:

Click to expand...

I've never met a flashlight that was "too bright" :devil:

Oh, the light survived multiple 1+ hour long runs on high and some bumps on the trail. I was worried the Maglite wall-mount wouldn't hold up, but I'm not exactly plowing down hillsides and jumping stumps. Now, to clean, or not to clean?



...to clean...

(really, I just wanted to show off the new seat)

thanks for the bike info.
i like IGH - simple looking.

hmm this looks pretty interesting to me... i ride/race dirtbikes and have been looking at getting a helmet light, how much would you say the head weighs? i could keep the rest of it in my camelbak.

Why is there a school bus in your yard? Looks like a real chick magnet.

redswr said:

hmm this looks pretty interesting to me... i ride/race dirtbikes and have been looking at getting a helmet light, how much would you say the head weighs? i could keep the rest of it in my camelbak.

Click to expand...

It's probably ~1/2 lb. (Imagine about 1/2 the weight of a 2D Mag loaded with Alkaline D cells.) I don't think you'd be too comfortable with it on your head, even if the cord running from the back pack up to your helmet didn't bother you. I tried a helmet light first, but got too many bugs in my face. The handlebar mount suits me much better.

VegasF6 said:

Why is there a school bus in your yard? Looks like a real chick magnet.

Click to expand...

Yeah, sexy, right? My neighbors love seeing it there too. My wife drives the camp bus for the day camp she works at / my kids go to.

This old thread needed updating for accuracy... I was tempted by the newer KD S020121 driver, which promised buck AND boost, so the LEDs could be driven at a full 3A for the duration of the discharge cycle. With the DX buck-only driver, once Vbatt drops below Vf of the emitters @ 2.7A (+ whatever the driver consumes), then the current draw and brightness would slowly wane. This setup had decent battery overhead, and after a 1.5-2 hour ride in the dark, the light would still come back with no detectable reduction in brightness. (In truth, it was dimmer, but over the course of the ride, I noticed no difference. The batteries would have to be really abused for me to see that the light was dimmer.)

Really, I was motivated to change the driver because I found the memory on the DX driver annoying (always goes to the next mode at next power-on), and knowing the driver regulated at 2.7A, I was hoping to see 3.0A for the entire discharge (wishful thinking, I know). Sacrificing the strobe would not be a big deal - I rarely had occasion to use it.

Long story short - the KD driver was a waste of time and money. A clamp meter test between the driver and the LEDs in this exact light revealed only 2.5A is pushed to the LEDs after initial settling in. Worse is that after exactly one minute of runtime using the thermal pads provided by KD for heatsinking in the Mag neck, the driver would throttle back to a visibly lower brightness level. Current draw at the tail could be measured rapidly dropping from 3.1A down to ~ 1A.

On advice from others, I potted the driver in a homemade batch of AAA, using premium Alumina powder from eBay and Devcon 2-ton epoxy. Here's the result:

That pic was taken after removing potted driver from light and repairing the positive lead that was torn in the removal process. I'll explain why I took it out later.

Here's what it looks like from the side after potting. If I ever need to separate it from the switch, I'll have to use a dremel to cut it off:


Sadly, all the materials, time, and money trying to improve this $15 driver to make it as good as the $5 driver it replaced was a waste. At room temp, it lasted a little longer before throttling back... a whole 3 minutes this time. It didn't dip as low at the end of the 3 minute period as it did at the end of the 2 minute test period using the thermal pads, but it still dropped down to between 1- 2 Amps. The neck of the light wasn't even hot at this point. Slightly warm at best.

I had to prove to myself that I could get it out if/when I wanted to put the DX 57779 driver back in. I also wanted to try wrapping the aluminum sheet collar that the driver was potted to in aluminum foil, just in case the thermal grease wasn't enough to fill any gaps between the sheet and the mag neck. In the process, I tore the positive lead, but luckily was able to drill into the AAA and patch it.

With potted driver wrapped in foil and then crammed back into the light, I tried some super cooling tactics to see how it fared. To give you an idea of how bright it is by comparison, that thing circled in red is the reflection of the bathroom incan lights in the brass faucet):


That bath has got to be as good, if not better than riding along at a steady clip with the light mounted to my handle bars by steel hose clamps. After 10 minutes on high in the bath, a tailcap measurement showed 2.5A draw. That's reasonable, considering that by the time the emitters are heated up, less draw from the battery should be required to maintain their current. However, after a minute of holding the leads at the tail, draw dropped to 2.1A and stayed there. Out of the bath, after a 45 minute run on my desk, the light is scorching hot in the hand, and current draw at the tail is 1.25A. A visual comparison of the light in that state to my Lil Giant (also 3x MCE in DX MOP triflector) shows it to be about 90% as bright.

Long, long story short - the potting worked well enough to convince me to leave this driver installed, but I had I known this would have been the result, I would have stuck with the DX driver. On the bright side, runtime on med probably significantly increased and this driver has better mode memory. The low battery warning kicked in on me after ~2.5 hrs of runtime on high (the first 30min in short runs, then the last 2 hrs a single run.) That, I suppose is nice, but open voltage of the four LiFePO4 cells that I'm using showed them at 2.6V each when that happened. Now they might sag worse than any other LiFePO4 cells under load, and 2.6V open is a good time to charge them up again, but all that aside, unless the driver is smart enough to calculate that I'm running four cells, I think someone running only two Li-Ion cells would be very dissapointed with how quickly low batt must kick in if it always does that at the same Vin.

A member on BLF reports that the driver doesn't trip the thermal protection nearly as quickly when it's boosting from a lower Vin, but even with that in mind, I still can't recommend it.

Greetings, I enjoy reading threads where not everything goes according to plan and I am grateful to you for sharing your difficulties with us. I learn more this way and it saves me repeating attempts already made by others. From the picture of the potted driver it's hard to see how close the driver is to the metal it's being potted to but it seems far. Would a better fitting sleeve or maybe a wide strip of copper wound and potted from in close out to the case improve thermal conductivity? In the mods that I have done and mostly due to the recommendations of those who came before me I have tried to get a solid piece of copper, even a small one, as close as possible to the heat source and "pipe" the heat to ambient. This has worked for me on the kd 5-15v buck driver that I have running 2 triple xre q5's in mr11 lamp heads and a handheld in a plastic 8AA case. All three are ~ 2 years old and no thermal issues. In all three the sink for the driver is not directly connected to the led sink. I have done 2 Maxflex lights (4xpg 4s) and the one with the driver seperated from the leds(even by only 1/16") runs cooler than the one on a common sink.
Best of luck and I look forward to seeing more of your work. Mine tends to be less photogenic.
Scott

rufusbduck said:

Greetings, I enjoy reading threads where not everything goes according to plan and I am grateful to you for sharing your difficulties with us. I learn more this way and it saves me repeating attempts already made by others. From the picture of the potted driver it's hard to see how close the driver is to the metal it's being potted to but it seems far. Would a better fitting sleeve or maybe a wide strip of copper wound and potted from in close out to the case improve thermal conductivity? In the mods that I have done and mostly due to the recommendations of those who came before me I have tried to get a solid piece of copper, even a small one, as close as possible to the heat source and "pipe" the heat to ambient. This has worked for me on the kd 5-15v buck driver that I have running 2 triple xre q5's in mr11 lamp heads and a handheld in a plastic 8AA case. All three are ~ 2 years old and no thermal issues. In all three the sink for the driver is not directly connected to the led sink. I have done 2 Maxflex lights (4xpg 4s) and the one with the driver seperated from the leds(even by only 1/16") runs cooler than the one on a common sink.
Best of luck and I look forward to seeing more of your work. Mine tends to be less photogenic.
Scott

Click to expand...

Hey, Scott. That's a good point. Replacing as much alumina epoxy as possible with solid copper would certainly provide better thermal conductivity. You can only see the inductor coil in the potting picture. What you can't see is the stack of 26mm PCBs beneath it. As I recall, there was very little room between them and the wall of the Mag tube. A few weeks ago I was in Home Depot and almost bought a copper coupler to shim a solid Al heatsink with, but balked at the $4 price tag. Two or three of those cut open and overlapped would have probably been an improvement over the Al sheet, but at that point, I'd have spent more on heatsinking than on the driver. At this point, I'm convinced that while that would have been an improvement in heatsinking, I still think the driver would not perform as desired. Honestly, I'm just glad to be done tinkering with this light for as long as this driver survives.

On a side note, I received a "FandyFire 3900" triple XML light as a gift. I measured 2.5A draw from the tail on two fully charged AW IMR18650. Some crude math translates that to ~20W pulled from the batteries (8.0V under load x 2.5A). Factoring in a 90% driver efficiency (wishful thinking), that leaves 18W for the LEDs, or 6W each. An XML consumes 6W at a drive current of ~1900mA. I was, at first, convinced that the light should be modded by replacing the silicone glue with good thermal grease and the OEM driver with this KD driver. Now having had this experience with the KD driver, I'm happy to be leaving the FandyFire light, and its thermals, and its runtime alone. Besides, as it is, it's already 90% as bright as my multi LED Mag mods, and who'd want a cheap Chinese light to be brighter than their best Mag mod creations anyway?

I've seen what those drivers look like and the new high power ones from kd all look a nightmare to sink for the simple reason that you can't get a normal sink to a close enough proximity. George and Datzip both design their drivers with the need to heat sink in mind but the cheap price of these alternatives makes me forget what a royal pain they are to cool. My personal reason for favoring them has to do with how the modes operate. On Taskled drivers a separate switch and switch path are necessary for the ui. I use Vistalite lamp heads for my bike light mods and they are sooo small. I did manage a Maxflex in 2 of them, one with a digikey mom switch replacing the stock switch and the other with a small jack in and a remote mom by my thumb, but both are very tight. The ui on the sharkbuck/remora is such that to change levels you have to turn off the light and wait a few seconds. This was not acceptable to me in a bike light so I asked him to reprogram a couple of remoras for "on state" programming. This was expensive and I've only just assembled the second set and haven't tested it yet(the first stays in one mode, yes, I think I botched the soldering. Shark okay but not, I think, the remora. Anyway, I want to use my Makita 18v tool batteries for power(SS has a power block that the batt will lock onto) and this means a buck drive. Bflex wont fit and I really want the driver in the lamp head. I have read extensively about the multilux method as well as paralleling drivers and will use them for some lights but not all because of battery limitations. All of these options and nothing quite right. I guess that's what makes it all so much fun.
Happy modding!