Sofirn SP36, impressions of a sample

That’s not going to happen with a tiny85 MCU, but it could be possible on bigger chips with more pins and more ROM. I also considered using the LEDs themselves to read data from a computer screen, but the data transfer rate is painfully slow.

Instead, I’ve mostly just tried to make the interface easier to use and easier to configure. And I’ve been thinking about putting in a factory reset / auto-calibrate function people can use as a panic button. Basically, hold down the button while connecting the battery, keep holding it for a few seconds while it flashes out a warning, and then it would reset everything to defaults.

Mostly though, I don’t think a computer-based configuration interface is necessary. It’d probably be harder to get the thing installed and connected than it would be to just configure the light using its built-in button. It would also take significantly more work to write a cross-platform config app than it does to write the firmware.

Guess I missed that part . That alone adds $15ish to the cost of the light. So maybe $50 to $55 is reasonable.

I think there might be another reason for the 6000 lumens figure.

@djozz, do you know what kind of springs they are using in the prototype SP36?

I’m asking this since your sample seems to be using the same spring array as in the BLF Q8, and it seems normal for you to get “only” 5000 lumens from the light.

The prototype Sofirn SP70 is using BeCu springs, and that allows it to get 7200 lumens from it.

Maybe they are basing off their lumen values with higher conductivity springs?

Or maybe they are just overrating the light.

I did some thermal testing on this SP36, and I think it should run fine with the normal regulation algorithm. For context, Anduril has two methods available, and the choice of which one depends on the light’s power-to-mass ratio:

  • One algorithm does normal PID regulation across the entire output range.
  • The other (optional) algorithm takes over when the light is at very high levels. When it gets the slightest hint of overheating, it quickly drops to a hardcoded sane level, then goes back to the normal PID regulation. It won’t attempt to regulate back up above that “sane” level.

When I tried the more-paranoid one on the SP36, it ran for 2.5 minutes at full turbo (~5000 lm), then quickly dropped to ~3000 lm and didn’t change much. I ended the test at 5 minutes.

When I tried the regular algorithm on the SP36, it was still a bit warm from the first test so it responded a little sooner, at about 1.5 minutes, but it then dropped extremely slowly until it finally settled. It took 15 minutes to reach a steady state of about 1600 lumens. Then it gradually went back up to ~2700 lm, as cell voltage got closer to emitter voltage, because that produces less heat. Then my phone battery died and the test was aborted. The SP36 cell voltage afterward was 3.46 V.

The results are graphed here at two different time scales:

In both tests, the light never got painfully hot… just warm. It was gently fan-cooled, to simulate having a hand available for heat sinking. The temperature limit was set at the default of 45 C, which is relatively conservative. Users may want to raise this to a higher level, to get higher sustained output.

Also of note: none of the brightness adjustments are abrupt. They happen very gradually, and can’t easily be seen by eye. The only exception was the big sudden drop at 2.5 minutes in the first test, which is fast enough to see, but still smooth. But I don’t think that algorithm will be used on the production model, because it’s unnecessary on this light.

FWIW, here is the firmware used in the blue line: http://toykeeper.net/torches/fsm/anduril.2018-11-09.SOFIRN_SP36.hex

Wow the thermal performance of this flashlight is amazing! Even at $75, it’s still a bargain price for what you get. The USB C recharge alone for multi cell flashlight makes this worth $25 extra. But of course the lower the better.

So the batch being sold right now is NarsilM mot Anduril? How does thermal regulation work for Narsil? Curious how it compares with Anduril

Either it uses a timed stepdown (recommended) or it uses a temperature-based stepdown. The temperature based one checks the current temperature every N seconds, and instantly drops by a large amount if that temperature is above a preset limit. Last I checked, the default value of N was 45 seconds. It does not attempt to regulate back up if it gets cold.

The algorithm is something like this:

while on:
    wait_45_seconds()
    t = read_thermal_sensor()
    if t > 50_degrees_C:
        set_power(current_power / 2)

I don’t want USB recharge so if that makes the flashlight more expensive then too bad…
I was hoping for like 40 bucks because sometimes that was price of Q8

11.11 Sale.

SP36 light $54.95 without battery

We rated 6000 lumens according to tests result in lab.

Sofirn is the SP36 being sold with the NarsilM or Anduril firmware? I’m waiting for the Anduril.

Also, if you are sending your lights out to a third party lab to perform the lumen testing, I recommend you buy a “4.5” Texas Ace Lumen Tube”:Redirect Notice and a set of Maukka calibration lights. That way you can do in house testing and will save you time and money in the long run.

This light needs some bling. It’s so blah, just a tube with a few grooves around the switch. Just look at how sexy the ROT66 looks compared to the SP36. The ROT66 has the cool rectangular pattern in the body tube, the slight widening of the finned area around the switch, and the shiny stainless tail and bezel. The Sp36 looks like one of those $29 fly by night ultrauranusglarefire lights on Aliexpress.

I like Sofirn. One of my top 10 lights ( I have a LOT of lights) is my Sofirn C8F. That’s a good looking light, with great performance, and tremendous value. I’m just not feeling it with the SP36.

Photo below was pilfered from Toykeeper’s post earlier in this thread. I don’t have a ROT66 yet, but the more pictures I see of that light, the more I want one.

The Q8 finning is done better too. I would prefer the SP36 to have deeper and more fins with stainless steel bezel to match the Q8.

I went ahead and ordered one this morning along with 2 SP33’s.

I totally agree that the SP36 does not look exciting. And that the ROT66 has that extra special look that the SP36 does not have.

Having the sample for a while it does feel like a light to use because of the size and the performance. And I really like that bit of throw that the ROT66 lacks. And the USB-C charging. And the price is ok too.

So I think that what the SP36 lacks in exciting looks has quite some compensation.

What I do still hope for is a nice neutral led option

Maybe you could concider it a sleeper. Like a dull bland Chevy Malibu with an LS 427. Not pretty to look at but performs better than the average car.

Beauty is in the eye of the beholder. I have often heard that the ROT66 looks ugly, while the classic tube-like designs look timeless and elegant. It really depends on personal taste. As for my taste though, the one I think looks nicest in that picture is the D4S.

Regardless, I think the Q8 does look a little nicer than the SP36 in some ways. The steel bezel is a nice touch, the fins are more consistent, the logo lines up more accurately, and it is less prone to rolling.

About the shallow fins on the SP36, I did not find those to be an issue while doing thermal tests. It seems to handle heat just fine. It actually performs well above my expectations, both in terms of heat management and throwiness. By eye, it appears to throw about as well as a Q8, and it took a full 20 minutes to reach the bottom of its thermal regulation curve.

Typically on FET-driven lights, the output makes a S-shaped curve as the cell drains. Steady temperature, S-shaped brightness. I didn’t get to complete the test because my sensor’s battery died, but it was definitely making the expected S shape, and appeared to stop right at the point where it should turn downward for the last time. The output during the entire test was higher than I expected for a light its size.

Anyway, the SP36 charger works really well on mine, and it provides the unique benefit of charging three cells in parallel. I normally use only a 2-bay charger, so it’s a pain to recharge 3-cell lights. But now I have a way to charge those without breaking the set and taking two sessions. More importantly, I can do a partial charge and still keep the cells synchronized. Even if I didn’t use it as a flashlight, I think I’ll probably still use it as a charger.

I hope someone can compare the heat of SP36 and Q8 in different modes within 30 minutes. It is interesting to know how practicality suffered from compactness.

Both ToyKeeper and I were pleasantly surprised how close the thermal performance of the SP36 was to the Q8, despite the smaller size and weight, and the absence of decent fins. This observation should be sufficient to get the correct idea about the SP36: a good thermal design for its form factor.
But if the Q8 and SP36 were tested side-by-side I’m sure they will differ.

djozz could you ask Sofirn for GB price like Q8 ? The list price is quit steep.

Barry has told me that it is unavoidable that the SP36 wil be a bit more expensive as the Q8, so the current $50 for the SP36 is certainly not far from a possible group buy price (if a group buy is going to happen).

And personally I think that 50 dollar for what you get (if no design flaws will be found after all, which is always possible with new designs) is not steep, it is pretty much what I expected.

That said, I suggested Barry a price of 45 dollar when he asked me what I thought was a good price that was higher than the Q8 price.

I’m glad to see this light available now.

I wish it would have been a tad cheaper, but then i bought more expensive lights with less output and no USB charging… (thinking of the L6 or S70 for instance). This one i really like. Smaller soda can shape, almost as bright as the Q8 with the convenience of on board charging. What else? :+1: