Pretty simple DIY XR-E bike light

[daq]

Inspired by KonradC and especially nightrider, educated by newbie and SilverFox, I decided to build my own homemade bike light.
(Trying to make this post readable and informative only gives me greater appreciation for their efforts)

Summary:

• 3x Cree XR-E, Ledil optics, 8x AA NiMH direct drive.
• Pretty bright, pretty simple, pretty cheap.
• Each emitter is angled to lay down an even swath of light.

From Newbie's and SilverFox's NiMH posts, I figured the I could direct drive 3x XR-E with 8x AA NiMH. I already had AA NiMH batteries and a charger, so this seemed like a good project to "get my feet wet". 1idjack confirmed my thinking that this should work. My first test showed the current at 810 mA (I had first written 710mA. Fresh, fresh recharged batteries gives 810mA) with freshly charged batteries.

I'm not done yet, but here is an "exploded" view of the pieces ready for final assembly.


Here is roughly how it will look like when assembled.



Perhaps this light's one claim to fame is that the emitters are mounted on a copper flat bar that serves several purposes;

• be an angled "riser" for the emitters to result in a better distribution of light in front of the bike
• allow space for wiring underneath the optic
• act as heatsink

I used a hacksaw to cut about half way through the bar and then bent to form a 7 degree angle (1/8" rise is about 7 degrees). This angle was to match the +/- 4 degree angle of the Ledil optics. Here is a closeup (turn your head sideways ).



The pieces are;
3x P3 Cree XR-E
2x +/-4 degree and 1x +/-9 degree Ledil optics
1" x 2" rectangle aluminum tube in cut in half
1/8" x 1/2" rectangle copper bar
24 AWG stranded wire
3/4" conduit hanger
1/8" x 1" aluminum bar
8x AA battery holder with "9V" connectors
"9v" wiring harness
Arctic Alumina epoxy
heat shrink

I plan to use the Arctic Alumina to attach the copper riser/heatsink, top and bottom aluminum pieces and the conduit hanger for mounting to my handlebars. I think I'll use silicon caulk to try to weatherproof the LEDs and the optics. I don't plan to put a cover over the optics, I think they are pretty well protected now.

In subsequent posts below, I'll go into;

• Design considerations
• Material/supplies selection
• Tool requirements
• Build process
• Things I would have done differently

I hope this inspires someone to either do their first homemade light or to further improve the design.

Dale

 

[daq]

* Design considerations

The optics are basically 7/8" square. This helps determine the sizes you need for the housing.

For the copper heatsink/riser, I spec'd 2-3/4" for the length. This assumes about 1/16" for each cut with the hacksaw.

 

[daq]

* Material/supplies selection

To minimize costs, try to minimize shipments (since UPS is about an $8 minimum).

For metal parts, I found a "candy store" at www.onlinemetals.com. I bought rectangle tube, flat bar aluminum, flat bar copper for what I considered reasonable prices. If you live in Seattle, you can even get "will call" to pick up yourself. One funny thing is that if you order more than two pieces, they start to add a "cut charge" to each piece. You can work-around by ordering something Qty 2 of 2-1/2" and Qty 2 of 2-5/8" (if you're not too picky about exactly what the size is).

For LEDs and optics and Arctic Alumina, I bought from www.cutter.com.au. Remember to use the "CPF1" discount
It took only a week to get my package from Australia (during Christmas rush too)

I purchased the 3/4" conduit hanger from a local electric supply store for $0.24. Silicon caulk came from my local hardware store.

I bought the 8x AA battery carrier and the "9v" leads from my local Radio Shack.

Wire, solder, etc is listed in the tools section

 

[daq]

* Tool requirements

- Hacksaw with metal cutting blade
- Mitre box is helpful
- bench mounted vise makes bending copper heatsink/riser easier

A decent soldering iron goes a long ways and doesn't have to be expensive.
After buying TWO crummy $13 irons that burned up in no time, I found this thread which ultimately led me here

At this time, if you spend $50 (plus ~$8 shipping) at www.web-tronics.com you can get a pretty decent gift (I chose the pliers set)

Suggestions;
soldering iron
(You may also consider this iron. It looks like it has a nicer/smaller iron and comes with a replacement ceramic heater )
solder, stranded wire, desoldering braid, third hand, heat shrink (check other sizes too),

Can you tell I spent a lot of frustrating hours trying to do this "cheap"

 

[daq]

* Build process

 

[daq]

* Things I would have done differently

I clipped the corners of the XR-E's to isolate the bottom pads from the electrical paths. Next time, I'll just drill a hole a little bit into the ceramic with a bit that is just larger than the via. This will break the connection with the bottom pad but leave more room on top for soldering the wires (and there is not much room)

After trying 24 AWG solid wire, then 22 AWG stranded and finally settling with 24 AWG stranded, next time I'll just start with 24 AWG stranded wire.

I wired all three emitters together first and then epoxied to the copper bar. This was hard and resulted in less than ideal results. Next time, I'll solder the wires on the "top" and "bottom" emitters and then epoxy all three while they are loose. Then, I'll solder the wires to the middle emitter.

 

[daq]

* Alternative ideas

If there is a next time, I think I'll build a 4x Cree XR-E using the same Ledil optics and the copper heatsink/riser (in a square 2 x 2 array). I'll cut my 1" x 2" rectangle aluminum to to expose a 2" "face". (This will need a 1/8" spacer, or you can get the 1/8" thick tube instead of 1/16" thick)

From my reading of newbie's charts, the batterspace.com regulated Li-Ion pack should run the LEDs at ~200 mA (~50 lumens each) on 12V and ~800 mA (145 lumens each) on 13.2V. It would last ~25 hours on "low" (~200 lumens) and ~6 hours on high (~550 lumens). WOW!

 

[AndyTiedye]

Adding to bike light index

 

[luminari]

Cool, Daq, I'm about to embark on a similar journey with 2 Seoul SS P4's and a bFlex controller driven at 1 amp, and it was very interesting to read your direct drive observations. I have a friend who is trying to do it for less (the bFlex rocks but is $28) and is considering direct drive.

 

[daq]

luminari,

I really like the bFlex and the new UI specific to bikes.

To be honest, I just couldn't figure out how to mount the bFlex easily/robustly. This turned my thoughts to direct drive and I was lucky enough that a pretty reasonable combination of Crees and NiMH AAs came out "even".

Thanks for the comment. I was kind of wondering if I offended everbody.

 

[nightrider]

Hey daq. Nice going.

Have you got a chance to take her for a test ride yet, or at least get some beam shots. I was just curious about your lens choices and the way you angled them in. What kind of beam do you get?

What kind of usable runtime are you getting out of 8 AAs unregulated?

 

[p97z]

Nice...

I'm interested in seeing some beam shots. I'm in the process of making a triple but haven't decided on optics. It looks like nightrider inspired another!

 

[daq]

Nightrider,

I'm glad you saw my post. Obviously, I owe you a great deal of thanks. Thank you!

When I was bending the angles, I went a little "light" so the angles are probably about 6 degrees. I thought I could bend more later, but that really wasn't do-able.

Here's a white-wall beamshot (sorry, I don't really know how to take this photo properly). Remember to tip your head to the left.

If you look closely, you'll see two hotspots with just a small space between them. The Ledil +/-4 degree optics seem to have a small square hotspot, a larger square corona (I think that's the right term) and not too much spill. You can't really even see the +/-9 degree beam at the bottom (right), but I think it's doing a reasonable job of lighting up the area right in front of the bike.

So, I'm pretty happy with how the beams are separated.

Below is a picture from my backyard. The distance to the playhouse is 50-55 feet. The playhouse doors are almost 6 feet at the top. In this photo, the top of the highest hotspot was on the ground just in front of the playhouse. Remember to tip your head to the left.

I tried to adjust exposure to match how I perceived it, but it's not exactly how I saw it. To me, the brightest portion of the photo wasn't as bright as it shows in the picture and the dimmer portions (especially the playhouse) seemed brighter to me. Perhaps this is difference in how cameras and people perceive light. :shrug:

I'm doing some run-time testing now. I was _almost_ patient enough to get a full battery charge (probably at least 90-95% full). At the beginning of the testing, the aluminum angle was getting quite warm (have to take fingers off after pressing firmly for ~5 seconds) so I figure the heat sinking is working well.

p97z,
In general, I'm not wild about optics. However, it makes things so much easier when you can work with an optic/reflector that is square and fits into a cheap, available material. Yep, definitely inspired by nightrider.

Hope the beamshots help. I'm happy with the beam. When it's out on the road/trail, I don't think the optic "funnies" will be perceptible.

Dale

 

[daq]

Here are the results of my run-time testing.

Batteries: 8x Titanium 2000 mAH
Charger: Titanium V4000

4 batteries were completely charged, and the other 4 were almost completely charged.

Time      mA       Approximate Lumens*
  0      810       394
 10      690       356
 20      610       326
 30      540       305
 45      470       270
 60      410       245
 75      380       229
 90      340       213
105      320       205
120      300       191
130      300       191
140      290       183
150      280       180
160      270       175
170      260       172
180      250       167

* Approximate lumens estimated from newbie's chart . Includes 10% deduction for losses due to optics.

I'm getting too tired to finish. Let's do some estimatation (and be a little conservative). SilverFox's graphs for this battery show that 1.75 AmpHours is the max before you're probably abusing the battery. For this setup, we have the following averages;

Minutes    Avg mA      Cumulative AmpHours
   0-30     675         .34
  30-60     475         .58
 60-120     355         .93
120-180     275        1.21
-------     ---        ----
(subsequent numbers are guestimates - I don't know if they are likely)
180-240     230        1.44
240-300     210        1.65

We see the light runs for at least 3 hours. I'd be willing to guess that it will run at least one more hour (maybe even two if my batteries match SilverFox's tests).

Sure, it's not as bright now as it was at the start. However, you're not exactly "limping along" with 160+ lumens at the 3 hour mark since that amount of light is about what you'd get from three Luxeon III's being run pretty hard.

Dale

 

[chris_m]

Not a big fan of driving high powered LEDs direct. After all when you've gone to all that effort and expense of building something as nice as that with the latest LEDs and optics, which aren't that cheap (looks very similar to what I'm thinking for my next build, but I'm thinking of the 2x2 quad you suggest later), why not spend just a little more and have a light which is just as bright after 2 hours as it is at the start?

 

[nightrider]

daq. Beamshots look good. And not bad runtimes from unregulated 8 AAs. As Chris said, a constant current regulator will give you a more steady output, but for a shorter total runtime, and it will go quicker at the end... a flatter runtime curve, if you will.

The light will stay much cooler when you are riding. Have you got a chance to take a test ride with it yet? I might have missed it, will your light be for trail or road?

I'm still waiting for some Cree CR-E Q2-bins, and Seoul P4 U-bins.

Thanks for keeping us updated daq.

 

[daq]

Here are some finished photos.


You may notice that I didn't use the 3/4" conduit hanger to mount to the handle bars. I had a mount from a little 2.4W 4AA light that was easier to install and adjust so I used it.

Not a big fan of driving high powered LEDs direct.

Yes, direct drive is not an optimal solution. This is my first homemade light, so I deliberately made it as simple as possible to help ensure it would actually get finished. Even with as simple as it is, I'm kind of embarassed to think how much time it took me.

After all when you've gone to all that effort and expense of building something as nice as that with the latest LEDs and optics, which aren't that cheap, why not spend just a little more and have a light which is just as bright after 2 hours as it is at the start?

If I would have known the time and expense at the start, I might not have started.

The secondary reason I didn't add some sort of driver was because I couldn't figure out how to mount it easily. Feel free to offer some suggestions to how this could be done. Maybe some aluminum tube with some sort of spacers and caps?

(looks very similar to what I'm thinking for my next build, but I'm thinking of the 2x2 quad you suggest later)

I think a 2x2 would be a nice setup. After riding home tonight, I think I'd have 3 LEDs at the same angle (2x 4degree Ledil and 1x 9degree Ledil) and one at about an 8-9degree downward angle (with a 9 or 15 degree Ledil) for the area close to the bike. I think my setup gives a bit too much light up close. (However, I've not ridden it the rain yet and I think wet asphalt is one of the hardest things to light up well)

I'm "cooling" on the idea of the batteryspace 12V/13.2V regulated Li-Ion since it uses voltage regulation rather than current regulation. One thing to consider is I measured a voltage of 9.53V (while the light was running) on the battery connector at the 90 minute mark on my runtime tests (340 mA). This equals about 3.18 Vf which compares with newbie's 3.1 Vf for this same current. Or, on newbie's chart, 3.18Vf corresponds to 480mA. So, either I have some resistance with my stranded 24 AWG wire (possible, but I only have about 3 feet total wire) and/or the Vf on my LEDs differs some from newbie's. This would make me a bit hesitant to use a battery with a voltage regulator (like what I assume is in the Li-Ion pack) instead of a current regulator. Also, since this battery only does 13.2V, you can't run 4x XR-E with a buck regulator like the bFlex.

daq. The light will stay much cooler when you are riding. Have you got a chance to take a test ride with it yet? I might have missed it, will your light be for trail or road?

Yes, I was surprised what a difference riding made on the cooling. You can hardly feel that it's warm (of course was around 48F when I rode home tonight).

Almost all of my riding is for commuting. (Sadly, my full suspension is a shed queen). Since I'm just commuting, it works pretty well to put the batteries in a zip lock and in a Bento Box. For mountain biking, you'd want a sturdier system for holding things in place.

I'm still waiting for some Cree CR-E Q2-bins, and Seoul P4 U-bins.

At first I was drooling over the Q2, but then I realized that the performace increase between bins at 750-1000mA isn't quite as impressive as at 350mA. I keep having to remind myself that 100% brighter equals 50% brighter according to our eyes.


This is not intended to be a "downer", but for full disclosure, I compared my light to a co-worker's NiteRider Classic Plus. The NiteRider has a 12W spot and 20W flood. I was comparing during daytime which is not optimal, but it appeared that my light (after about 20 minutes usage) may have put out more light that the 12W but the 12W had a brigher hotspot. I was hoping to get the equivalent of the 20W flood, but I was no where close to that from what I could see today. I'll try to do the comparison outside tomorrow evening to to see if looks different then. Perhaps the Ledil optics are holding things back since the datasheet says "up to 90% efficiency".

 

[chris_m]

This is not intended to be a "downer", but for full disclosure, I compared my light to a co-worker's NiteRider Classic Plus. The NiteRider has a 12W spot and 20W flood. I was comparing during daytime which is not optimal, but it appeared that my light (after about 20 minutes usage) may have put out more light that the 12W but the 12W had a brigher hotspot. I was hoping to get the equivalent of the 20W flood, but I was no where close to that from what I could see today. I'll try to do the comparison outside tomorrow evening to to see if looks different then. Perhaps the Ledil optics are holding things back since the datasheet says "up to 90% efficiency".

You have a problem with the performance of your light I reckon (either that or your comparison isn't very good). Finally found the 20W bulb for my old Lumicycle system, and took that out on a ride to compare tonight (Lumicycle on the bars, Cree triple on my helmet). The Cree triple with the Cree 8 degree optics definitely has better throw than the 20W halogen, which is a 10 degree spot. Both systems are regulated - the Cree at 700mA, the Lumicycle with a custom PWM regulator at 13.2V equivalent, or 10% overvolt. Not much noticeable difference in spill between the two. So the conclusion has to be, along with my comparison with my 20W halogen headtorch which gave similar results, that a triple Cree is more powerful than a 20W halogen.

 

[daq]

Thanks for your observation Chris,

Your experience matches what I was hoping/expecting to get out of the light (about the equivalent of 20W halogen). In some ways, I'm surprised my light works at all because I'm certain that I abused the LEDs with my lack of soldering skills. Several times I know I heated them up enough to cloud the lense (but it seemed to "get better" after cooling).

Are you using Ledil optics in your setup too?

 

[chris_m]

No, I'm using the Cree optics (ie made by Cree, or at least marketed by them) http://www.cutter.com.au/proddetail.php?prod=cut680. However I don't believe there is anything wrong with the Ledil optics in performance terms, and was planning on using them for my next build (it's just my current light is a triple in a 50mm diameter round can and I picked the optics which fitted).

If you managed to cloud the lens, you may have damaged them and spoilt the performance - degradation of LEDs is certainly heat related - I'm not an expert on this though!