Celebrating the independent kiwi spirit of invention.

Caving Headlamp – Extending Runtime

By Ian Mander 24-27 January 2010, updated 26-30 October, 5 November 2010, 25 June 2011, 6 April 2012.

The caving trips I've been doing in the last couple of years have been getting longer and longer, but because of thick mud I don't want to have to change the battery during a trip if at all possible. This is OK if I use the medium mode of my present headlamp a fair bit - at around 30 lumens it's enough for small passages, but on my last trip I found I used the high setting of my headlamp, at 330 mA or 75 lumens, most of the time. This is partly because I wanted a reasonable amount of light without changing modes often - the greater amount of light is needed for larger passages and especially for moving quickly - and partly because the lens of my headlamp got covered with mud, so the output dropped to less than the medium setting with a clean lens.

Of course, the down side of running it on high is that the 3xAA Eneloop cells were getting very tired by the end of a 7+ hour trip. (A fellow caver on that trip with a similar headlamp but much younger eyes used less than half his battery capacity. Perhaps his reflector is more effective too.) It would be nice to have a longer runtime.


Solution 1: Bigger battery

One easy solution is simply put a larger battery on the back of my helmet, for example wire another 3xAA Eneloop cells in parallel with the existing three. This would give 12 hours runtime but I don't think I want that much extra weight, and too much weight at the back could make the helmet unbalanced.

I'd probably be happy with four cells, though. Many headlamps do come with four cells, and 4xAA suits most four slot NiMH chargers. The extra voltage means that buck drivers could be used, which will allow the extra energy in the fourth cell to be used, giving about 8 hours runtime with 4xAA Eneloop cells.

Update June 2011: As an option to consider for the future, using a couple of protected 18650 Li-ion cells would give roughly double the runtime of 4xAA NiMH cells at roughly the same weight. Unfortunately 18650 holders are a little hard to find.

Update April 2012: 18650 holders can now be bought at DX (see options here), so Li-ion cells are now a more likely option. I have also located a source of high capacity Eneloops - Sanyo XX cells (US$21.99 at Intl-Outdoor Store). These will give about 10 hours runtime, and I use them in my present caving light.

Accepted - change from 3xAA NiMH to 4xAA NiMH, and use buck drivers with either Sanyo XX or Li-ion cells.


Solution 2: More LEDs each run at lower current

LEDs are more efficient when run at lower currents, and weigh a lot less than Eneloops, so a more attractive solution might be to add more LEDs instead of more cells, and share the current between them. Thus a slightly lower total current would be needed to produce the same lumens.

For 75 lumens, three Cree XR-E P4s would draw about 280 mA instead of 330 mA. This would give an extra hour; 7 hours runtime instead of 6 - not much improvement, and so probably not worthwhile just for that.

However, more LEDs would allow different optics on the separate LEDs, which would be nice if suitable switching was included. More LEDs also increase the maximum illumination from the headlamp - something that was essential when 5 mm LEDs were the only white LED available (at a price about the same as the Cree XR-E P4 now).

Accepted, change from single LED to multiple LEDs, but more because of the different optics that could be used and that it greatly increases the maximum illumination possible than because of the small benefit for runtime.


Solution 3: A single more efficient LED

A single Cree XP-G run at just 180 mA would produce about 75 lumens. Compared to a Cree XR-E P4 at 330 mA, this change by itself would extend the working duration from 6 hours to 11 hours runtime. That's more like it - that sounds excellent.

Single XP-G Example Modes
Mode Lumens Current Duration
(assuming linear driver, 3xAA)
Low 10*
reading, caving rest breaks
30 mA 2.75 days
Medium-Low 35
small passages
80 mA 25 hours
Medium 75*
general caving
180 mA 11 hours
Medium-High 138
demanding caving
330 mA 6 hours
High 180*
demanding caving
445 mA 4.5 hours
Turbo 350*
large chambers
1 A 2 hours

* Just happens to be 50% (or ~56% for the 350 lumens) more lumens than the equivalent mode in the Fenix HP10.

FWIW a single XR-E P4 at 1 amp provides about 180 lumens. For the XP-G R5 to produce that at just 445 mA is quite a significant improvement in efficiency. Its efficiency is so good that a $3.90 headlamp from a discount variety shop could be modified with one (and an inexpensive optic) and even with just the stock driver would provide almost 40 lumens and 8 hours of runtime from 3xAAA cells. (The future is here! But that's another project; now completed, but with a different driver arrangement.)

Accepted - change from Cree XR-E to Cree XP-G.


Solution 4: Use multiple more efficient LEDs (or multiple dice in one LED)

A multi-die LED like the Cree MC-E has four closely spaced dice in the one package, and even in the entry-level K bin each die is more efficient than the P4 bin. The down side is that only one optic could be used. A K bin can produce 77 lumens at 60 mA per die, or 240 mA total (when run in parallel). That would mean 8 hours runtime instead of 6 hours. Getting better, but not quite there yet - the 11 hours offered by a single XP-G is more attractive, and doesn't have beam pattern issues from using one optic or reflector to focus four dice.

At this point I'll pause to compare various LEDs' price performance. For about the same money (at time of writing, Jan 2010) I can get either two MC-Es (K bin, total 8 dice), or four XP-Gs (R5 bin), or 8 XP-Es (Q5 bin). Which gives best bang for buck (lumens for dollar)? The calculated lumens from various configurations, with the LEDs or emitters in parallel are in this table.

Current 1x MC-E K
2x XP-G R5
4x XP-E Q5
lumens (approx.)
4x XP-E R2
350 mA 113 151 117 128
1 A 302 400 313 340
2 A 538 702 567 612
2.8 A 685 882 739 790
3.0 A - 920 778 830
4.0 A - - 949 1008

It's obvious the XP-G R5 is the most efficient, and it holds its ground against twice as many XP-Es each running at half the current (and therefore is more efficient than a single XP-E at the same total current) because the XP-G has a large die (and therefore high efficiency) and exhibits less fade than other LEDs at high current. Since the XP-G is so small they should be among the most practical to include in a headlamp; four of them on 10 mm boards (round or square) could be put in roughly the same footprint as a single MC-E on a 20 mm star.

Accepted, change from single XR-E to multiple XP-Gs.



Standardise on 4xAA 2500mAh cells and multiple Cree XP-G LEDs. The higher battery voltage and lower Vf offered by the XP-G also allows the easy use of buck drivers instead of linear drivers. The total gains from the various changes will (in theory) push the 6 hour runtime to over 18 hours runtime.

As a future project, using a couple of 18650 Li-ion cells will extend this to 29 hours runtime. Three cells would give 43 hours runtime.


Celebrating the independent kiwi spirit of invention.

Return to ianman HOME | Back to Aqualab Home | Return to TOP
Inventions: Super Soaker Backpack | Air Cannon | Car Interior Lighting | LED Torch

* This would have been an ad.

When you buy stuff from Asian sellers:
Please don't buy stuff from a country in the middle of intimidating its neighbours.