Andúril 2 coming to Sofirn - The general Sofirn development thread

Nobody should be complaining about short run time because it is normal. All small high powered lights have short turbo. If anyone knows of one that has a decent long turbo run time, please let us know so we can analyze how it works.

I am not so pleased by how fast my Astrolux S43S gets hot. It is small with a copper head so you would think it has a decent turbo run time. Unfortunately, it also draws 20A and the emitters and lens are not very efficient. It’s not that bright, yet it heats up super fast.

So small lights need good efficiency (light output to amperage draw), they need a limited amperage draw (nothing crazy high), then they need good control over the temperature (such as calibrated sensors combined with a smooth ramp down). All of this usually results in an expensive light.

I believe Zebralight is known to be really good at building small, yet high powered lights. They are very sophisticated and expensive, which is why I don’t own any.

Liquid cooling is not practical at all. It would be much larger than air cooling. So forget that idea.

I did a search and found your post with pics of this light. Wow that LuckySun F3X looks amazing. I like it alot more than the C8 series with those very functional cooling fins! I wish Sofirn can make one with fins like that and with USB-C recharge!

Love the optional extension tube

In general there are no silver bullets - small but super-bright will be very hot.

1. Liquid cooling - if you add a pump it’s going to work much worse than a fan and adds a lot of complexity as well. Passive liquid cooling, i.e. heat pipes may work, but there’s an easier improvement that I’d suggest trying first:

2. Unibody construction. Improving heat transfer across the light will improve both turbo and sustained performance.

3. For storing heat I suggest trying paraffin wax. It’s lightweight and as it melts it takes away a lot of heat. If you can make a sealed container and good thermal transfer into that container you’ll get much better turbo times.
If you heat 1g of aluminum from 20C to 60C, it can store ~54 Ws of heat.
If you heat 1g of paraffin wax from 20C to 60C, it can store ~360 Ws of heat.
Some lights released recently went with super-thick shelves to improve Turbo times. As you said - that made them large and heavy. I don’t think that using aluminium as heat storage is a good option….but it’s certainly a cheap one.

4. Consider thermally insulating the points where user touches the flashlight body. This way you make the LED hotter and a bit less efficient, so you harm sustained performance. But you can increase body temperature without burning user hands. This improves heat capacity and therefore - turbo times can be longer.

5. Also, I can join others in saying that the more efficient a light is the better. A 1000 lm light that does 50lm/W needs to dissipate or store roughly 17W of heat. A 1000 lm light that does 100 lm/W needs to dissipate or store 7W of heat. Double efficacy again, at 200 lm/W that’s just 2W.
So:

  • efficient driver
    • buck
    • boost
    • buck-boost
  • many big LEDs (note: this can make the LED a flooder. If you don’t want a pure flooder consider flat LEDs, these are more efficint for throw)
  • efficient optics
    • multi-layer AR coating is better than single-layer which in turn is way better than uncoated optics
      • some multi-layer AR coatings are better than others but they are more expensive as well
    • review available reflector coatings - there’s a huge room for improvement though cost may be prohibitive
    • TIR+glass lens is a bit less efficient than TIR alone
    • some TIR lenses are better than others
    • take care about your bezel design, every bezel reduces output but some are better than others
  • low electrical losses
    • quality springs, reasonably fat wires, e-switches…quite easy to do well and overkill doesn’t help much

Note that good cooling improves efficiency as well, but for a regular flashlight this is a small effect. As a practical example please consider Emisar D4 with Nichia 219C and XP-L HI. With the XP-L HI it has significantly longer turbo time even though it’s brighter.

6. I’d like to also add that compact designs are better…if you remove some features that use space you can re-invest that space for better cooling. Same with implementing features in a more compact way. Your light will have the same size and weight but it will perform better. What options to reduce size would I recommend?

  • clicky switches are less compact than e-switches
    • though quite a few users love the ergonomy of tail switch. Tail e-switch is an option then.
  • Zebralight-style light engine
  • DQG-style tail caps offer super-low electrical resistance and take less size than the more common spring-equipped ones

Storing heat through wax? Wrong objective.

Myself I tend to disregard turbo times, I tend to need only several seconds anyway. I’d much prefer manufacturers to focus on lm/g and lm/mm³ of their flashlights with lm being either turbo (disregarding turbo times) or sustained.
But I’m unusual in this regard and from what I’ve seen most users do care about turbo times. So it makes a lot of sense for manufacturer to optimise that.
To optimise turbo times without lowering output there are only 3 options:

  • improve efficiency
  • improve heat shedding
  • store the heat

The first option doesn’t get you very far. The second is pretty much meaningless for improving turbo times of powerful lights. The last one? I don’t like that objective but I think it makes a lot of sense for manufacturers.

Heat is indeed an issue for small lights. Playing with my D4 made me aware of that.

Still, with the popularity of smaller 18650 lights (DQG, Emisar FW3A, Zebralight and others) I think that if you could find a way to make a light in this size range, you would find a good deal of customers.

I wonder if the tailcaps could be reduced in size. The one on my SC31 is a good deal smaller and lighter than my other 18650 Sofirn lights. A shallower reflector, no charging circuit, shorter springs (to be used with unprotected flattops only) and perhaps using a triple carcio optic could reduce the length in a measurable way. You could also use a driver that doesn’t allow the highest levels for safety and longer turbo runtimes if you choose.

A light that is shorter in length (using some of these ideas) but larger in diameter like the SP33 might have enough mass to handle the increased heat too.

I realize that it’s easier said than done, but with the solid reputation Sofirn has established I don’t see how a more compact design like this wouldn’t be a success. Thanks for your consideration.

@JasonWW, Agro is talking about phase change materials.

The advantage of using aluminium is mostly good heat dissapation, but can’t handle heat too well.

Using a material that changes of phase can handle a lot of heat.

It goes from solid to a viscous liquid. It can be very useful to hold heat then release it.

Our goal is not to hold on to heat. It’s to get it away from the led and driver as quickly as possible.

Trying to store heat inside the flashlight would be like having a pill inside an 18650 tube flashlight holding the led and driver. Then we isolate that pill so the heat doesn’t reach the outer body of the flashlight. We might think it’s a miracle. The light gets super bright and doesn’t even get warm. Great, right? Unfortunately, the led and driver are frying themselves inside because all the heat is stored up in the pill.

Technically it works, but it’s short term. If the pill, led and driver get too hot they will destroy themselves. The advantage here is that they can handle higher temps than our hand can. So we can extend turbo run time maybe 50% or something. But that is the only advantage (at the expense of stressing the internals more)

Storing the heat inside is a disadvantage when it comes to sustained output. That heat has to get out somehow. It needs to be exposed to the air and our hand to cool the light. Otherwise we have a light that can do a longer Turbo run, but then needs to be reduced to maybe 50 lumen or maybe shut down completely for 10 minutes or so to let the heat slowly go down inside.

There is a good reason nobody builds flashlights like that. Instead, everyone builds lights that try to get the heat out to the exterior as quickly as possible. It heats up fast, but also cools down fast and can actually sustain maybe 400-500 lumen (still talking about an 18650 tube light)

A light designed to store the heat could not run 400-500 lumen continous. It would fry itself internally and might only do 100 lumen continous due to the poor heat path.

My above example might be kind of extreme because I wanted to show both the advantages and disadvantages of storing heat inside. Maybe there is a way to build something in between. Like a hybrid. I don’t know. To me it seems like any type of design that tries to store heat instead if get rid of it just has too many disadvantages.

One way to view this differently is using a phase change material with a vapor chamber heatsink.

The material would store a lot of heat without rising in temperature much, and the vapor chamber would be directly connected to the fins.

That’s the advantage of using phase change materials: the temperature doesn’t actually change until it changes of phase. That’s the magic that makes it so amazing.

This would not work well at all with small lights.

However, with lights like the BLF Q8, SP70, and BLF GT sized lights, this would work extremely well since there’s a lot of surface area to dissipate the heat outside of the light, even at 50-100W+ and 200W+ power levels.

The design of these lights allow for very efficient heat dissipation. They don’t have enough thermal mass to sustain that, since they aren’t being actively cooled.
Phase change material change that.

Heck, even some eBike battery packs using this design with aluminium heatsink. They use phase change materials to cool down very high performance packs, and aluminium heatsinks to get rid of the heat quickly.

Maybe I should do an experiment related to this…

I’m serious on this, but I’m actually going to buy some paraffin wax, and fill the BLF Q8 driver cavity with paraffin wax to not only waterproof the driver completely, but to also see if it helps with sustained thermals.

Chances are you’ll bust the Q8 wide open. Paraffin wax expands a lot with increasing temperature. That’s how the thermostat in your car works.

And anyone who ever tried pouring hot wax into containers for making candles is familiar with the “sinkhole” that forms when it cools.

I… wouldn’t do that.

BUT…… if you decide to do it, we want pictures & report; success or failure. . :innocent:

Do the fins really help a lot for heat dissapation? I found Surefire lights don’t have fins, streamlight lights have but not many and not that deep. Maybe I’m wrong.

Anyone do the experiment to compare with mass fins and without fins?

Yes, but you need a bit of airflow

I wouldn’t rely heavily on the cooling fins. their efficiency strongly depends on external conditions - air movement.
As an example “no air movement” ~24°C- there is not much difference between Olight X7 (~47°C) and Haikelite mt03(~45°C) in the ~2000Lm mode, while Haikelite has noticeably more area cooling fins.
But it is better to conduct an additional experiment, without wind and with blowing and compare. for example, if anyone has old Convoy C8 and new updated C8+(add fins) in the same configuration, or Sofirn C8F 18650 and C8F 21700.

You seem to confuse heat (thermal energy) with temperature. These are related but different.
For sustained operation, LED temperature = surface temperate + thermal resistance * power dissipation

Adding mass in form of wax, (that is: one that doesn’t contribute to thermal conductivity of the light) would slow down temperature rise before the peak is reached and have no effect once the temperature peak is reached.
Adding mass in form of alu (that is: one that improves thermal conductivity, but for well designed lights just marginally) slows down heat rise and (marginally) improves efficiency at peak.
Adding insulation layer reduces sustained performance. When the layer is too thick it’s indeed an issue. However if done right it’s not bad at all. If you start with a good host that never heats up the LED to more than say 80 C and skin to 50C and insulate flashlight skin in a way that increases LED temperature to 100 C at Turbo w/out chaning skin temp you end up with a light that:

  • is still reliable
  • has way longer Turbo
    • temperature delta near LED is 33% larger, improving heat storage there by 33%
    • temperature delta near skin is 66% larger, improving heat storage there by 66%
    • temperature delta at not-the-hottest points of the skin and other thermally-distant points is more than 66% larger
  • assuming that Turbo power is 4 times larger than sustained, runs the LED 5 C hotter during sustained operation which has marginal effect on efficiency.

Note that by making the insulation smart:

  • high emissivity
  • with thickness that varies to equalize surface temperature at touchable points
  • which is not present or has marginal thickness on untouchable points, maximising temperature there

you can actually improve heat shedding and increase sustained power dissipation. You still take the small efficiency hit but nevertheless sustained output should be higher.

Yes its a big difference with fins. There is a custom C8 head from kiriba-ru in this forum named “C8TT triple heatsink head”: Low-cost copper pills, spacers, optics, drop-ins (Updated 23/10/2022))
Here are some tests on it:
What did you mod today? - #6022 by contactcr
Convoy C8 triple with Kiriba machined head

Sorry, I’m not understanding most of what your describing. A lot of it sounds theoretical from a book. I’m not so book smart, I’m more old school, learn by doing.

About 20 years ago I started building active and passive air cooling systems for PC’s. Then I started brazing together pieces of copper and brass to make water blocks for water cooling systems, also for PC’s. Although I have experience with phase-change system’s using compressors and freon, it’s only from working on refrigeration products, not PC’s. I’m familiar with Peltier style cooling as well, but have never tried it myself.

Keep in mind that cooling a PC CPU is very similar to cooling a flashlight LED. Both have a small, but intense source of heat.

The specific type of phase-change cooling that you’re referring to is something I have no first-hand experience dealing with. I can only use my previous experience to guess what is going to happen. Maybe you are right and wax will work great. I’m just saying that, based on my previous experience, I don’t think wax is going to work well.

If you experiment with it, let me know how it goes. Good or bad, I’m sure there is something to be learned. :+1:

There is a plenty of similar flashlights, which can do about 800-1000 lumens, and they have a mediocre (at best) emitter and a simple UI (which is not matched enough for use cases), and that’s all.

We need to find things to differentiate this specific flashlight beyond that it haves built-in USB charging.

I think, rost333 told many useful ideas, let me add a few things on top of that.

My personal observation was, that when I started to show a few of my flashlights to muggles, for the first flashlight, they didn’t care about different CCTs and high CRI, but when I showed them different CCTs, the most popular combination for emitters were in the CCT range of 4000K-5000K (a few older people were felt themselves very good with 3000K as well). And, when I showed the first high CRI flashlight to them, they acted like - “really, it shows colors seemingly better” - “it is not that dull and harsh cold white, which I can get from most of the cheap flashlights” - “wow, this is beautiful, I want one, too”. So, if measured lumen output does not differ too much, high CRI still can be a win, when it comes to human perception.

So then, I would suggest Luminus SST-20 4000K 95CRI, if a more defined hotspot is desired
and I would suggest to get Samsung LH351D, which is more efficient (I don’t have exact numbers, maybe by 15-20%), but noticeably floodier - and there is a little tradeoff in CRI, but it is still 90. It is also available in 5000K and 6000K.

Taking a look at the UI, the current UI has these:
single click to ON/OFF
long click to cycle modes: 10 lm (Low) - 200 lm (Medium) - 610 lm (High)
double click to full power strobe

Jump between Low and Medium is too big. Low is not enough low.

I would suggest something like this (output number suggestions for SST-20 4000K 95 CRI):
from OFF/ON: single click to ON/OFF - resume to last memorized output level (Moonlight is also memorized, but not Turbo and Bike Strobe)
from ON: long click to cycle output levels (change level every 0.5 seconds) : Eco (3 lm) - Low (25 lm) - Medium (150 lm) - High (500 lm ~ 66% of current in Turbo mode, for continuous use)
from OFF: long click to enter Moonlight (0.3 lm) - from here, long click to start output level cycling from Eco level
from OFF/ON: double click to Turbo (~700 lm) - timed stepdown (3 minutes) to High, single click to last memorized output level - double click will return to Turbo any time for another 3 minutes
from OFF/ON: triple click to Strobe - single click to last memorized output level - another triple click to Bike Strobe (how it exactly works, needs to be defined later, but having this mode should be definitely useful, as this is an EDC light)

Having this UI, this flashlight would have 6 well spaced output levels, which don’t need to be cycled in full length, as two of them are ‘hidden’, so fairly quick output level selection is still possible.