Light & Motion Vega: Battery and LED Replacement

By and large, I like my L&M Vega.

Basic characteristics:

• LED headlight
• Fairly tight-focus round beam
• One-piece battery + light + case
• Handlebar quick-disconnect
• Plug-in charging without removing batteries
• Nominal ratings: 85 lm for 2 hours // 55 lumens for 4 hours // 35 lumens for 8 hours // 35 lumens flashing for 24 hours.
• 240g with handlebar clamp
• Retail price US$180 (2006, 2007, 2008).

Some good attributes:

• One-piece, so:
  • easy on/off/stow
  • only one thing to find and connect when re-mounting
  • no cords to come unplugged while riding
• Reasonable — for my needs — tradeoff of size, light output, and running time.
• Flash mode runs a long time and is highly visible.
• Simple charging: just plug in and forget, no disassembly required, no "unplug after N hours" needed.

Some bad attributes:

• Hard to attach and remove: it uses a ball detent that takes a lot of force to remove. It would benefit from a spring-loaded latch, like most other lights.
• The beam is round, which makes it too narrow and too tall. Oval would be better; better yet, a pattern that is bright at the top and dim at the bottom, like European lights. The current beam is less than ideal for several reasons.
  • The beam is too tall, and bounces light off trees and other things that are not useful to see. The reflected light is worse than wasted, as it makes it harder to see light reflected from things you do want to see.
  • Light in the trees and on the ground up close is also wasted and would better be pointed at something useful.
  • The beam is too narrow, limiting the ability to see around corners, also making it is harder for others to see me.

A good/bad attribue is the light can be steered but does not have a center detent. Steering is good for seeing around corners, in part to make up for having too narrow a beam. The lack of a detent is good in that the detent is never in the wrong plcae, but bad in that the light has to be centered every time you re-mount it and it is more clumsy to center after steering it around a corner. In general, the less you have to fiddle with the light, the more hands you have left over to brake, shift, and steer — things you often want to do near corners.

Another good/bad attribute is the flashing mode gives some people nausea. It did not bother me but did bother some friends. The good part is the flashing light is very visible.

That's a pretty short list of annoyances. On the whole, I was pretty happy with the light.

Ultimately — after about 2 years use — the high running time was down to about half an hour. I went to look for the latest and greatest version. Despite two years and LED improvements of nearly 3x the lumens/Watt, modest NiMH battery price reductions and far lower prices for LiIon batteries, I was surprised to find the current model is unchanged. Same price, same lumens, same running time, same weight. (And same beam pattern, same handlebar mount, ...)

Also worth noting, you can buy a flashlight with modern LEDs, Li-Ion batteries, and a charger, with brighter output and longer running times, available retail, for much less than the price of the Vega.

I bought a USE Joystick MaXx, which is slightly brighter on high (nominally 3x, but in practice not so much), has a three-hour (1.5x) nominal running time on high, is one third the weight (85g with clamp vs. 240 for the Vega), significantly smaller (smaller diameter and lacking the Vega's mounting "shoe"), more expensive (US$250 vs. US$180), and has fairly comparable features: it is one-piece, easy on/off, has a plug-in charger, adjustable brightness, etc.

Here are beam shots of several lights. From left-to-right, a mini-Maglight(tm) flashlight with Teralux(tm) (140 lm nominal, 2+h run time nominal); unmodified L&M Vega on high (85 lm nominal, 2h run time nominal); U.S.E. Joystick MaXx on medium (120 lm nominal, 10h run time nominal); and Fenix P3D flashlight on medium (50 lm nominal, 6h run time measured). Note especially the relatively sharp focus of the unmodified L&M here compared to the Joystick on medium, and compare to side-by-side shots of the modified Vega compared to the Joystick on high, shown at the bottom of this page.

It seemed interesting and possibly useful to take apart the Vega, replace the batteries, and update the LED. Here are some notes on that.

Disassembly

Note: disassembly and modification doubtless voids the warranty. Mine had expired, as had the batteries, so there was not much left to lose. Opening and modifying was easy for me, but your milage may vary.

The bezel (lens holder) is attached with a left-hand thread. It was not even very tight, and just unscrews. The bezel is also the "heat sink" (heat dissipator) for the LED, so there is some thermally-conductive goo between it and the plate on which the LED is mounted.

The rest of the unit pries out of the case easily. In particular, there is no extra sealing or attachment around the charging plug or power switch.

(A rubber plug is supposed to cover the charging socket, but in practice it is hard to seat and falls out of place easily. It has also not given problems with water in the light, so for casual riding it is an okay design.)

The Vega uses a 3W Luxeon and on high overdrives it to 4W for brighter light but lower efficiency and shorter LED service life. The back of the Star is labeled

TY0KW 007 1657
1005LXHLLW3C
Luxeon

The LED is held to the heat spreader by two Phillips-head screws and is connected to the PCB by two soldered tabs (flat metal bars). These need to be unsoldered to replace the battery, and seem to be most easily unsoldered at the LED (which you might want to do anyway to replace the LED). It is possible to deduce which tab is + and which is -, but you might want to go ahead and mark them before unsoldering.

The battery pack is more complicated to remove, with red, white, and black wires. Red is +, black is -, white is connected via a small unknown component (perhaps a capacitor). The wires are short, pass through a small hole in an insulator/load spreader, and are all stuffed in place.

The battery leads are most easily unsoldered and resoldered by heating the other side of the PCB. They are clearly marked on the PCB.

     

Battery

The battery pack is 4x AA NiMH tab-type cells, glued together in a 2x2 square, 116g. Nominal capacity is 4.8 V and 2000 mAh (nominal 9.6 W-h). For charger compatability it should be replaced only with NiMH cells, not NiCd or Lithium-Ion. I did not trace the circuitry enough to know if the charge smarts lie with the PCB or the charger and thus do not know if a different charger could be used.

     

Replacement cells were Tenergy brand NiMH 2300mAh cells with spot-welded tabs.

I looked for and did not find an assembled tab welded battery pack, so bought cells with tabs which I could then connect. This is a problem: with all tabs pointing in, there is not enough room to press the cells together. However, pointing them other directions makes the central connection difficult.

Also, gluing the battery pack is desirable. There may be room in the case for a larger pack, but glue is not good at bridging gaps. Ultimately, I folded the tabs over and soldered them at the ends, but this is lumpy and slightly longer than the original.

During soldering, some care is needed to avoid melting the plastic that insulates the (live, negative) sidewall of the battery from the (live, positive) tab at the end. I placed a piece of thick corrugated aluminum foil underneath the tab I was soldering. It seemed to protect the plastic without cooling the tabs excessively. (The corrugated aluminum foil is backing material from a Rema tube patch, available from bicycle stores...)

Rather than recreate the heat-shrinked solder joint at the end of the existing pack, I simply clipped it off the old pack and soldered it to the tab of the new pack.

     

Re-soldering the pack to the PCB was not difficult but not entirely straightforward. The wires were already tinned, so I placed the wire at the back of the PCB, pushed it against the hole, heated the front, and pushed the wires through. This was a little bit awkward but not as bad as it appeared. The last wire took several tries and was most easily pushed with a pair of needle-nose pliers.

I do not know if it is safe to run the converter without an LED (load) attached, so some care is desirable when handling until the LED is re-soldered.

LED

The Vega uses a 3W Luxeon Star (marked: TY0KW 007 1657, 1005LXHLLW3C, Luxeon). According to Light&Motion, it gives the following nominal ratings:

The Star is held by solder tabs, two screws, and thermal compound to conduct heat to the heat spreader plate.

I replaced the Luxeon Star with a Cree XR-E Q5, available for US$9 retail including postage. At present (2008/01) the Cree XR-E Q5 has the highest nominal lumens/W of readily available volume shipping LEDs, though other LEDs are close. The Cree XR-E Q5 has ratings:

Note the Vega drives the LED at 4W, but both Luxeon and Cree XR-E are only rated up to 3W. The usual problems with overdriving are higher heat output and worse lumen maintenance. I assume Light&Motion has already solved the heat problem, though a Luxeon may be more heat-tolerant than the Cree XR-E (so this assumption may be false).

For lumen maintenance, an industrial target might be along the lines of "10,000 hours or more at 90% of original lumens". In contrast, a bike light at one hour per day over 4 years will barely pass 1,000 hours, by which time it will probably be worth switching to a new LED anyway (135 lumens at 1W have been demonstrated in the lab and will probably be affordable products in under 4 years; that's nearly 30% more lumens at fixed power or 30% longer running time at reduced power). Finally, even if lm/W falls off by, say, 50% over 4 years, that's still more lumens than the existing LED (though beam pattern may be an issue, see below).

The Cree XR-E is not rated at 4W. Extrapolating out to 4W gives an idea of the light output. Even if efficiency falls off to, say, 55 lm/W that would give:

Note this is only 5 lumens or about 2% more lumens at 4W than at 3W! However, (a) 55 lm/W is just a guess; and (b) no 3W setting is offered.

Removing the old LED was a simple matter of unsoldering the leads (which I had re-soldered to the old LED to verify the battery change was successfull), and unscrewing two Phillips-head screws. The Luxeon LED came off easily, with thermal compound removed just by wiping it vigorously with a rag.

The Cree XR-E star has matching notches around the rim, but they are slightly smaller than the Luxeon's, and slightly smaller than the plastic insulating posts on the Vega. Filing the notches slightly wider (maybe 0.2mm, I did not measure) was sufficient for the star to drop in, they did not need to be filed deeper (which would have been harder and/or required a file I did not have).

I applied Arctic Silver brand heat transfer compound between the star and the Vega's heat spreader. The heat spreader is electrically isolated and does not require a non-conductive compound.

The Cree XR-E star may also be slightly thinner than the Luxeon. The screws bottomed on the insulator before the star was pressed firmly on the heat spreader. However, further tightening yielded an apparently firm pressure. If I do this again I will first check clearance without thermal compound and maybe file a 1-2 strokes off the insulating posts. (Be sure to clear all debris from the heat transfer area.)

Re-soldering should have been straightforward. My goal was to put solder on the pads of the star (tin them), then solder the tabs (already tinned). It was remarkably hard to get solder to adhere to the pads, given I was trying to avoid overheating the emitter. I'm not sure if the issue is the pad material or the pad is thermally well-sinked so it is that hard to get it up to temperature. Ultimately I gave a few tries with longer and longer contact times (and cooling off periods in between) until getting an apparently good solder puddle on the tabs.

Soldering the tabs is somewhat complicated as they need to be held in place during soldering. I used an awl and pressed directly in the solder area. The awl has a sharp point and did not obviously drain a lot of heat during soldering.

I also applied a ring of Arctic Silver compound to the aluminum spreader and bezel where they contact. Installing and then removing the bezel suggests the thermal compound was distributed well.

Use

I have not yet performed run-time tests.

With the Cree XR-E LED, the beam pattern is bad. The original optic was designed for a different LED. I had hoped the SSC would be similar "enough". Instead, the beam has an irregular dark spot in the center; uneven brightness; and a series of concentric and gradually darker circles. The hot spot of the beam appears to be much broader than with the Luxeon's — good for seeing around corners and for being seen, bad for wasting light where you don't want it and for putting light back in your eyes (from things you don't want to see) thus interfering with light from things you do want to see.

The first photograph gives a rough idea of the beam sizes. The Vega's "hot spot" is much larger and has halos which are also much larger. The second picture gives a vague impression of the dark spot in the middle. It is less regular than appears in the pictures.

     

In riding tests, the large amount of near spatter gives an impression of great brightness. In practice, it is no easier (though no harder) to see far objects on the 2h/∼220lm setting than with the Joystick's 3h/240lm setting. and slightly easier to see far objects at the 4h/170lm setting than the Joystick's 10h/120lm setting. (In fairness, the Vega also weighs 3x as much.) The Vega is far superior for seeing around corners, both because it has a broad beam and because it can be steered.

It would probably be useful to experiment with Luxeon Rebel (nominal 100 lm/W) and Seoul Semiconductor Golden Dragon (nominal 100 lm/W) stars to see if either gives a better beam.

I obviously do not have the original for A/B comparison, but on high (4W) the hot spot is only a bit dimmer than the Joystick MaXx (nominal 240lm on high) and a much larger puddle. Also, where the Joystick's beam fades away to dim, the Vega+SSC has relatively large/bright rings at a farily large angle. Again, good for being seen, but bad for wasting light and for making it hard to see other things.

In riding, the Vega is sufficiently bright that it fares well against the Joystick. The Vega on high gives the impression of being far brighter than Joystick on high. It gives a wider beam which is good for seeing around corners and being seen. It is neither harder nor easier to see distant objects. The Joystick also has 1/3 the weight and 50% longer running time on high.

On medium, the Vega is much brighter than the Joystick on medium, but the nominal running time of the Joystick is 2.5x as long.