Considering a custom build - seeking feedback

[idleprocess]

I recently obtained a bicycle and with the onset of fall, it's getting dark during typical riding times.

A friend uses the Fenix BT20. While it produces a bright, automobile-friendly beam that's great for road biking, its area of illumination and light to the sides leaves a bit to be desired when transitioning between lit/unlit areas and on trails.

I'm considering a custom light with 4x XP-G triples. Two will be collimated with 24° narrow optics and the other two will have 43.6° X 23.3° elliptical optics. The idea is to provide a wider spread of light nearby for riding on trails and unlit streets while also sending light forward to see beyond this immediate pool.

I'm expecting to wire up all the modules in parallel and use resistors to control current. Power will be in the form of a 9.6V NiMH pack. I will use a short length of 2"x1"x0.125" aluminum U-channel (at least 6", mounted horizontally to the bars) to house the entire affair with a polycarbonate window.

Using Excel to do some hasty extrapolations of the output and forward voltage curves based on current, I derived the following with likely drive levels:

If   | Sum Vf | Array P | Array Lm | V_Ohd | Run | Fudge-Lm | Fudge-Run
-----+--------+---------+----------+-------+-----+----------+----------
0.10 | 8.31   | 3.33    | 358.28   | 1.29  | 300 | 268.71   | 225      
0.35 | 8.95   | 12.53   | 1408.56  | 0.65  | 86  | 1056.42  | 65       
0.70 | 9.67   | 27.08   | 2666.21  | -0.07 | 43  | 1999.65  | 32       
1.50 | 10.52  | 63.12   | 4608.87  | -0.92 | 20  | 3456.65  | 15

If and Sum Vf are per-module current and forward voltages
Array P and Array Lm are power and spec-sheet lumens for the entire device
V_Ohd is simplistic anticipated voltage overhead for the array verses the battery pack
Run is anticipated runtime assuming flat pack voltage and uniform amp-hours, based on 2000mAH cells
Fudge-Lm Multiplies lumens by 75% to account for heat, optics, etc
Fudge-Run Multiplies runtime by 75% to account for pack voltage drop, capacity loss at higher discharge, etc

Looks like 700mA drive level is iffy and I'm simply not going to be able to drive the array at 1.5A with a 9.6V pack. Of course, I don't account for thermal effects on forward voltage and surely will not direct-drive anything. Otherwise, 150/350/700mA drive levels will provide some OK output distinctions.

My challenge is fine-tuning the optics. Angling the entire assembly down by about 16° at a height of 36" will make the eliptical beam illuminate from 5 to 35 feet. Angling the narrow optics up about 4° (a ~0.60" shim) will allow the narrow optic beam will intersect the ground slightly further out and proceed to infinity on the upper edge. I can sort of imagine what this will look like, but lack the geometry skill to project it in a drawing, can't really tell how the light will be distributed, and I'm not sure how effective it will be as a headlight. I expect that the foreground will be much brighter the distance in spite of my efforts to try to throw some more light forward. Intensity is clearly never going to equal the BT20's hotspot, which in my mind is acceptable. I was hoping to introduce some vertical cutoff as a courtesy to automobiles on the road, but since the beams aren't as neatly horizontal as 2D projections suggest, should I rethink the geometry and angle up? Perhaps any LED's dedicated to closer illumination should be driven at a lower level? Look at other optical solutions? Rethink the array components?

Appreciate any feedback.

 

[Steve K]

What were the design objectives? Was this supposed to be a light for off-road riding, or for both road and off-road?

If this is for road riding, I think even the narrow lens is far too wide. For comparison, I use a Ledil Rocket SS for road riding, and feel it is about the correct width. It is described as +/- 4 deg FWHM with XR-E LEDs. I had tried some slightly wider optics, and felt that it required me to angle it down too far, which put too much light up close to me.

As far as off-roading, I don't have any experience with it, and can't offer any suggestions.

 

[idleprocess]

Road, bike trails (which vary immensely in their illumination), offroad - would like this to do a bit of everything.

Friend's BT20 is great for riding on the road where things are generally predictable. I found it troubling for sudden changes such as alleys and trails. I felt that the lack of spread would make it a poor choice for offroad.

Simple 2D projections definitely make the narrow optics seem far too wide. I lack the ability to even begin to model relative brightness ... not sure if my old copy of AutoCAD's raytracing will simulate illuminance levels.

 

[Steve K]

For road illumination, I'd recommend moving away from the triples, and consider perhaps just two discrete LEDs each with narrow optics. I've got a dynamo setup with four XR-E's with the Ledil Rocket SS running at 0.5A each, and it seems quite sufficient. If anything, I would consider something a bit narrower in the vertical direction. There is a company, Gaggione, that makes some very interesting narrow elliptical optics, but they don't appear to actually sell them.
http://www.lednlight.com/

It's been a while since I've searched for other narrow optics for modern LEDs, but the Ledil Heidi RS looks okay. It is advertised as having a FWHM angle of 9 degrees for the XP-G. The diameter is 21.6mm, so it's not too huge.

 

[idleprocess]

Did some simple experiments and observation this evening.

I have some driveway lights with similar triple boards to what LED supply is selling running at 350mA specced at ~300lm each with no optics other than crude diffusion. At approx 9 feet off the ground they produce a nice pool of light on the driveway - about half of which is reflected off the garage door. I could fairly comfortably read under that light from about 5 yards away.

I did another experiment where I held my PD35 at about 36" off the ground and angled it down roughly 15 degrees. On the 170 lm level it seemed pretty adequate and very adequate at the 450lm level ... save for the lack of side spill.

So ... I think I'll need appreciably less lumens than I assumed unless I want to do something nutty like offroad downhill runs. A 3p2s (or even 2p2s) array of single LED's would likely do the job admirably and reduce the pack size. A pair of eliptical optics will generate the wider close-in/lateral illumination I want while some reflectors will generate a narrower throw beam.

This is assuming I don't just settle on the PD35 using the bar mount that's in the mail.



EDIT - The PD35 in a TwoFish block will do it quite adequately for now.