Sofirn sp36 blf anduril runtimes and lumens per level

Hi all, just recently got a sp36 blf anduril. I’ve searched but can’t find runtimes or lumens per level etc. Has anyone tested it in stock trim using 7 step mode at least? Id like to know what lumens and runtimes im operating at somehow. Thanks.

Seems like you have a task ahead! :partying_face:
Please report on BLF

I assume there might at least be a general formula dividing total lumens from 1 to 5500. There’s 150 levels in anduril and that would be something like 36lm per level if they are even all the way through. But it would even be harder to figure starting from 1 lumen etc.

From eyeballing with other lights knowing their claimed lumen on levels, I’ve roughly come up with - moon .5, 8, 65, 240, 410, 1200, 2700 and obviously 5650 turbo.

I’m mostly just impressed that your first post was 6 years after making an account. :slight_smile:

About the lumen levels, the ramp isn’t linear. It’s meant to be perceptually linear, but that means the actual power levels and lumen levels are pretty far from a straight line.

An ideal “perceptually linear” ramp is usually shaped like y = x^3 (ish), instead of just y = x. Fit the cubic curve between ~0.3 lm and ~5500 lm, and scale it to fit a range of 1 to 150, and it should give you a pretty fair estimate of the ideal lumen output at each of the 150 ramp levels.

But there are a bunch of things which make the actual curve different. For example, the power control chips don’t have a linear response curve, so PWM levels don’t scale directly with lumen levels. And the actual ramp isn’t really a single curve… it has three or four different segments which were calibrated manually by eye. The lowest levels in particular follow a different shape due to limited hardware resolution, so it’s probably best to ignore those and put a false value in at the bottom to make the rest of the curve more accurate. Maybe 1 or 2 lumens instead of 0.3.

In any case, even after correcting for all that and getting actual lumen estimates from the ramp calculator, there’s a good chance those values aren’t very accurate. So it’s better to measure the light’s actual performance instead.

Moving forward with estimates though, the bin/level_calc.py tool can help. And the anduril/steps.py tool can be used to determine exactly which of the 150 levels are used for the stepped ramp.

The raw PWM values can be found in the source code, but this may or may not be very helpful. It at least shows the approximate shape of the ramp. However, in this case, the SP36 config file simply inherits its ramp values from the BLF Q8. And the BLF Q8 file is old enough that I hadn’t given the ramp calculator the ability to change its curve shape yet, so the lumen values given to the script are pretty far from reality. Also, it’s old enough that it doesn’t specify the actual stepped ramp parameters; it just uses the defaults. So that makes things a bit inconvenient too.

The default stepped ramp is level 20 to 120 in 7 steps:

$ ./steps.py 20 120 7
1: 20
2: 36
3: 53
4: 70
5: 86
6: 103
7: 120

The bottom power channel maxes out at level 65 and ~150 lm, so here’s a ramp which places it at that level:

$ ./level_calc.py 7.67 2 150 7135 1 1.5 150 FET 1 10 5500
...
20: visually 1.31 (8.02 lm): 12.15/255, 0.00/255
...
36: visually 1.53 (25.92 lm): 42.77/255, 0.00/255
...
53: visually 1.76 (76.08 lm): 128.57/255, 0.00/255
...
70: visually 1.99 (195.56 lm): 255.00/255, 2.69/255
...
86: visually 2.21 (432.42 lm): 255.00/255, 13.96/255
...
103: visually 2.44 (926.35 lm): 255.00/255, 37.45/255
...
120: visually 2.67 (1852.37 lm): 255.00/255, 81.50/255
...
150: visually 3.07 (5500.00 lm): 0.00/255, 255.00/255
...

This is just made up on the spot though, and only matches two points on the curve. I ended up with a curve shape of x^7.67 and a “moon” level of 1.5 lm to fit those two points.

More accurate values definitely require measuring the actual hardware instead of just throwing numbers around.

Hehe 6 yrs. Well, I had to think about it for a while lol.

Anyways thanks, though much of it sounds like Chinese over my head to me lol. But I do have a better idea now.

I guess my “eyeballin-it o meter” wasn’t to horribly far off I suppose.

So given those rough ballpark numbers and using 3x18650 is there a rough runtime per those 7 levels you might guess? Thanks again!

About runtime, it spans a huge range, and the calculator again won’t be very accurate. However, at moon it should use about 1.5 mA to 2.0 mA of power, and at level 65 it should use about 350 mA, and at turbo it’s probably about 15 A to 25 A… ish.

The runtimes with 3x3000mAh cells then are essentially just 9000 mAh / the mA of the level being estimated, to get the number of hours of runtime. For example, 9000 mAh / 350 mA = 25.7 hours at level 65.

Wow, this stuff is pretty neat!

I treated myself to one on 11/11 as I had my eye on one for a while. I’m a little confused about the batteries I should get for it though.

I normally opt for the safe route - Protected NCRs ~3500mahs. I’ve heard reports that to get the max out of this light I’d need high drain batteries such as the Samsung 30q’s but I really don’t like using unprotected batteries for muggle-proof safety.

Do you think that the NCRs can perform adequately or will I not manage to reach peak output?

Cheers!

Keeppower sells a protected 30q

You can see you dont lose alot at 7 amps per cell

I think that NCR18650B’s may wear faster than normal in this flashlight because they will run at the top of their rated amperage (also under max ramp setting because many levels are PWM-ed max current). High drain batteries will have an easier life, and you get more light on top of that.

If you are the muggle in question, you can use unprotected high drain batteries no problem: since you ask questions on a flashlight forum you are already way ahead of what I see muggles do with my flashlights. And the driver and Anduril firmware takes care of the batteries pretty well. Just do not insert one of the batteries upside down or you will create a nasty smell and a toasted spring that needs replacing. Inserting them all upside down is fine though, but you will have no light then.

If someone else is the muggle, something is wrong, they should not be allowed near high drain multi-thousand lumen flashlights. :expressionless:

I admit, the muggle is I! :person_facepalming:

I have a small (ahem) collection of lithium powered flashlights which I usually manage feel safe with due to investing only in the protected NCRs.

Thanks @ Bearbreeder for the suggestion very good idea. I think I’ll go with the 30Qs though as Keepower are hard to get around here and I feel a little safer with the Sammys as my fears are spontaneous combustion rather than improper use so much.

@Djozz, thanks, battery health is actually very important to me so your suggestion has given me the confidence to order 3 30Qs. Will the SP36 with Anduril drop modes / switch off when voltage drops below the 3.0 mark? Do you know what kind of internal protection the flashlight will offer that may compensate for the lack of battery protection? I.e. what would the protected cell offer that the SP36 BLF not offer in terms of protection?

Thanks for all your help! :beer:

I hope you will order button-top versions of the 30Q, or else manage to solder a small blob of solder on top of the plus-pole, because else the batteries make no contact with the driver, the +poles need to stick out a tiny bit.

Hey Djozz, thanks, I can’t find 30qs that are button top, but I’ve found the Samsung INR18650-35E. I think they may have a discharge rate of 13 amps instead of the 15A the 30Qs offer. I think that’s twice as much amperage than the NCRs if I’m not mistaken.

Do you think the 35E will do?

Cheers!

35E’s are very nice batteries, and suit the SP36 well I think.

Much appreciated Djozz, thank you so much! Just placing the order so I’m ready to receive the SP36 BLF accordingly!! :partying_face:

The 35E’s only have 8 amps of Continuous Discharge current capacity as seen here: Samsung 35E

Whereas the 30Q’s have 15 amps of Continuous Discharge current capacity as seen here: Samsung 30Q

Thanks 71k5, isn’t that enough amperage for the sp36 on max? I would have thought that since they are in parallel that would be a total of of 24 amps, right?

That is not exactly how it works. Rated amperage is the maximum that the battery can deliver that is still healthy for the battery, it has little to do with what the device actually gets because that is much more dependant on the resistances in the circuitry than on the batteries. My guess is that the SP36 with 35E batteries run at about 13A. With 30Q’s because of lower internal resistance 16A (when batteries are still full). So the light output on 35E’s is marginally less but you get longer runtime back for it.

what output is this light stable at?