Astrolux S43S review (4x 219C, 18350/18650)

Two things, or perhaps 3:

1. Difference in LED bins. There can be as much as a 7% brightness difference between bins.

2. Spring compression/battery size. It can make a huge improvement in conductivity, especially since these are copper alloy springs. If a coil touches another one, there will be a non negligible compression bypass, meaning resistance will be lowered, and current will be upped. A button top can help with this due to the additionnal height.

3. Most importantly, maukka has a ANSI calibrated light setup. Most of us don’t, like flashaholics.

A lot of people have uncalibrated light measurement devices without a proper shape for accurate integration. Their results are still useful, but can be off by quite a bit. The amount of inaccuracy in most cases is unknown.

Flashaholics is known to report significantly higher numbers than reality, and uses a very odd measurement device so the readings can’t be scaled down by a consistent ratio to get accurate numbers.

Maukka has a calibrated integrating sphere, and reports some of the most accurate numbers available. When in doubt, trust maukka’s numbers.

I want a light with good thermal throttling.

My Emisar D4 is retarded - if I set the thermal limit so that it behaves well (neither gets too hot nor steps down excessively) in the medium lumen range then it burns my fingers if I accidentally ramp too high.
If I set the thermal sufficiently limit low so that turbo can be safely used, then the light behaves too conservatively the medium lumen range - stepping down too fast and too much.

I was hoping that due to:
a) the large(ish) mass of copper (and generally a light that seems way bigger than it needs to)
b) user-configurable ceiling (remove the possibility of accidentally going too high)
c) the fact that its 2018 and thermal throttling should improve over time due to lessons learned etc

that S43S would be right for me. But according to review, it is not. It gets hot enough to burn, then steps down to 95 lumen and never steps back up. Basically the worst of both worlds, somehow even worse than the D4.

uff…after the turbo limit it never goes back to the maximum? there are no circles?it can be used for maximum just once?

You can see the behaviour here:
https://i.imgur.com/zA5LUr7.png
From turbo it takes 50s to step down. Some people are saying this is great - for me this is very bad if it reaches burning temperatures during this period (which he reports it does).
From there it goes down rapidly, eventually settling at 95 lumen (after 2 min) and never goes back up.
From top of ramp it fares not much better, eventually settling on 95 lumen after 3 min.

Of course you can manually go back up (to turbo) but for me this is beside the point.

okok.thanks)

As varbos said, turbo is always available, it’s just direct drive so the output depends on the battery voltage under load. With a 30Q at 50% state of charge, it’s about 1600 lumens.

With high discharge batteries, the S43 and S43S will actually shut down from overheating. This happens shorty after they’ve stepped down to the ~95 lumen level.

What firmware is this in your D4? If stock try Anduril, it has a lot updates regarding thermal management.

I have been planning to try it someday. I think putting a ramp ceiling would help a lot. The problem currently is that towards the top of the ramp it is just too capable of rapidly generating huge amounts of thermal energy. So I need to configure a very low thermal limit…which then kicks in prematurely during normal mid-ramp usage.

Thermal mgt has never been a top priority for me, as many of you may know. Of course it's really about the amps&volts (i.e. watts/power), causing the heat problems. Two things that don't go together: EDC size pocket burning power and true thermal regulation control. Zebra claims a true PID, so in theory if they do it well, should be wayyy better thermally controlled. Dr Jones also developed a pretty good PID based algorithm for thermal control, but not sure if it's used in a high powered driver (high performance FET). Of course Zebra and Dr Jones are commercially selling lights and drivers/firmware, very much unlike me.

I'm sure Anduril would do better, but not sure how well in this particular light, and again, depends on what you want for max output vs. max temp, how automated/manual you want the control to be, etc. If Anduril still is using the built-in Attiny85 temp sensor, it's got some major limitations. For the S43S, copper makes a great heat absorber, taking the heat away from the LEDs, but that's also part of the problem with copper - the heat is retained and held longer.

Oh boy, I think I see why the S43S is so inefficient. Measure the lumens of the stock S43S setup on the 0.35 amp mode (from mode sets), max 7135, then remove the black bezel and lens, and re-measure lumens - it's ok, the plastic optics stays in place. Mine (219C version) goes from 102 lumens to about 132 lumens -- that's a 30% jump!!

Yikes! Then compare a D4 to the S43S head from head view, here's the differences:

  • D4 is wider overall
  • S43S LEDs are closer to each other
  • S43S optics have a wider cup then the D4's
  • The D4's optic cups are all inside the bezel, no clipping, while the S43S's cups are all clipped off, covered by the bezel's edge

Besides the above, would Astrolux use better optics and AR lens than Hank would? No I don't think so. Hank's optics are Carclo optics, high efficiency, what optic does Astrolux use? It's a non-standard size, they don't say, so who knows. Needless to say, I'm sure there's additional loses coming from the lens and optics. But the big drop is from the optics being overlapped by the bezel and partially blocking the cups.

Unfortunately since the optics are a smaller diameter than standard quads, there's no easy way to mod/upgrade the S43S, least I know of. If the height would work out, might be better using a triple board and optics - this should at least clear the bezel.

Well, the copper is at least purty , mostly - mine has some nasty discoloration on the top edge above the switch by the bezel .

That also means the potential lumen numbers are off by 30%!

Maybe the 2600 lumen numbers weren’t so inaccurate after all.

Some cheaper AR coatings cause losses too.

S43S optic is 1.70 mm thick while the Carclo is 0.89 mm thick. S43S has 2 thin posts that easily break (yes, I broke one trying to remove it), Carclo has 4 thick posts (never broke one). I dunno why I spend the money on these quad Astrolux's - I got a dead S42 in pieces, and now this one. The cheap price combined with shiny copper I guess, too tempting.

Well, went ahead and ordered the 18650 tubes and replacement optics... Ugh, for losing roughly 25% in output for no good reason at all, except a poor optics/bezel design. Maybe I can strip the ano off the bezel, dremel/sand down the bezel to remove the lip and epoxy down the optics, while tossing the lens altogether.

Well, I hope someone somewhere can tell Banggood/Astrolux their optic/bezel arrangement used in their S42, and now the latest S43S is greatly lacking. I guess they never got the message on the S42 before.

And heating up the head in the process.

Sorry, I’m just finding this thread.

I should point out that the output specs for the S43 and S43S are listed incorrectly on Banggood.

The aluminum head S43 XP-G3 should be around 2100 lumen and the 219C should be around 1600 lumen. Same as the previous S42 version. I don’t know why Banggood says both led versions do the same output. Everyone should know the 219C are not as efficient as the XP-G3.

The Copper head S43S replaces the single driver and tail spring for a driver button and twin tail springs. This allows for even higher amp draws and more output for the copper version.

It does not surprise me that Maukka was able to measure 2100-1900 lumen from the 219C version. Well above the 1600 lumen spec. In M4D M4X’s video he measured 15A draw from a 30Q at 4.0 volts. I’m not sure if he had the 219C version or the XP-G3 version, but this copper head definitely draws more amps than the aluminum version. Then again, you can bump up the aluminum versions output by bypassing both springs.

It is a real shame that the bezel may be loosing us some output for no good reason.

I don’t have data sheets for Nichia emitters, but with Cree they show there can be a +–7% variance per bin. So as much as a 14% brightness difference between bins.

Where are you hearing about 2600 lumen? I did not see it mentioned in this thread anywhere.

I'm confused about 2600 and 1600 lumens specs - where? I see 2023 lumens and 2100 lumens on Banggood - that's all I could find on lumens specs. The Astrolux website: https://www.astroluxlite.com looks pretty worthless.

I hope someone else can verify what I saw by removing the bezel and lens - wondering if I did it right or something wrong with my setup to do this kind of measurement. The higher 102 lumens I measured vs. maukka's 89 is understandable - I'm still in the industry std (non ANSI) world of lumens.

Least it gives me some direction of what/where to mod to try and improve it. Thinking of sanding down the inner part of the bezel - black to raw alum would probably help as well. Tricky because there is an o-ring being held by the lip on the bezel.

And I hope the lo ball estimates aren't a justification for a poor optics design -- how can you trust any spec that has one number for 2 differing LED options? This is pathetic, we all know it and they just continue to do it - why not? 5 star reviews just keep rolling in and we keep buying them, myself included . Where's that head slap?

The 2600 lumen number was measured by WalkintotheLite.

I’m just saying that if their numbers were measured with a better lens and less bezel, then the numbers might benefit a lot from these mods.