Andúril 2 coming to Sofirn - The general Sofirn development thread

I am interested in the Q8 with 4xXHP50.2 (3V).

That’ll get hot fast. I wonder what the highest sustainable output will be

~2000-2500Lm

Hopefully the same as the original Q8 which was 1500-2000lm.

The original Q8 can hold ~1500Lm for no more than 20 minutes, after which the temperature will already be over 50°C

The highest sustainable output is mostly determined by the host itself, not the LEDs used. There is some variation with different LED types, but the differences are small. So the sustainable level would probably be very close to the current versions.

Here’s what I measured in a turbo runtime test:

The direct-drive-style power circuit makes output sag over time so, to maintain temperature, the driver keeps bumping the output up. As the battery voltage gets closer to the LED forward voltage, it becomes more efficient and can sustain higher brightness levels at the same temperature. It’s not the prettiest runtime graph, but it’s normal for this type of light.

The HK04 is a host with quad XHP50.2 3V's, only 3 cells instead of 4, overall slightly smaller than a Q8 but similar form factor.

My HK04 w/spring bypasses does 14,500 at start, 12,000 and 30 secs (maukka calibrated), and with raising the the temp threshold to 85C or so, it's hot to hold at 30 secs.

Knowing Sofirn they’ll probably cheap out somewhere limiting max potential

No idea. Really there's no such thing as constant lumens with a driver like this. The HK04 uses a FET + 1, but the "+ 1" is actually not a 7135 but what I suspect is another FET with resistors in the circuit. This is similar in design to TA's high amp drivers.

Unfortunately I have no loomen toob to test with, but my M6 (4x18650, 3xXHP50.2) holds full ramp (16x AMC7135) until the voltage drops too low if I set the thermal limit to 65C. The entire head gradually heats until it’s somewhat painful to hold, though. But that takes a while - for 20 minutes of “bright” it’s just a decent handwarmer.

Used ‘ceilingbounce’ app (with HK04 4x XHP50.2 pointing at the ceiling)

HK04 is set to “ramp ceiling” at the start of the test, estimate the ramp ceiling output to be around 3000+ lumens.

(I had a small USB-plug in fan pointed at the HK04 head the entire time though)
(I set the HK04 temperature threshold to 65 degrees Celsius, ambient temperature at my place would be somewhere around 30 degrees Celsius)

At 10 minutes mark, brightness still good, dropped to around 90% of starting brightness.

About 16 minutes later (I stopped the test), the brightness has dropped to around 80% of the starting point. (so probably around 2500 lumens)

The flat lines are interesting to see from a driver like this. The ramp ceiling is PWM'ing the FET's so output should slowly drop, even with the additional cooling of the fan. What you are showing is steps down - could this be from the resolution of the measuring device? It appears to be taking course measurements.

It’s almost certainly due to poor analog gain staging causing low resolution on the measurement device.

Like, let’s say the phone’s sensor has a range of 0 to 10,000 lux. But the light is ceiling-bounced with the phone aimed up at the ceiling. The incoming signal is pretty dim. So instead of 10,000 steps, the brightest level is only … 37 lux. That’s what it appears to be from the graph, anyway. So it’ll look pretty coarse no matter what the light does, since there are only 37 possible levels.

To fix this, aim the light directly at the phone sensor, and adjust the positioning until it measures around 80% of what the sensor can handle. ON my phone, this means adjusting until turbo reads out as about 8,000. Then run a test, and the graph should look much more smooth.


Edit: Or it could just be the light having a strange output pattern. The graph suggests the peak value was 501, not 37. So if that’s correct, the sensor should have been able to see in-between steps.

Also did similar ceilingbounce test with SP36 BLF Anduril.

Batteries were around 4.0-4.10v in each case.

Brightness set to Ramp Ceiling.
Small USB fan pointed at the flashlight head.
Ceiling bounce to ceiling, with smartphone running ‘ceilingbounce’ app, facing the ceiling.
Anduril temperature limit set to 65 degrees Celsius.
Room ambient temperature would be somewhere like 30 degrees Celsius here.

BLF SP36 (LH351D 5000k), “old” Anduril version (no 2-level lockout)
(my estimate of lumens in ‘ramp ceiling’ would be somewhere like 1700-2100 lumens)

BLF SP36 (LH351D 4000k), Anduril version 2020-03-18
(I would estimate the lumens in “ramp ceiling” to be somewhere 1400-1800 lumens)

Seems to look basically similar for older and newer Anduril firmware, at least when set to ramp ceiling.

(comment: it’s good that Sofirn is now updating the firmware installed on their Anduril and Narsil flashlights).

I’ve noted report that the SP36 (XP-L2) which used to run NarsilM v1.2 is now installed with NarsilM v1.3
(not sure what’s the difference between v1.2 and v1.3 though)

I wonder why the Sofirn AT35 never went ahead. it’s shown in the OP

Here's the NarsilM rev history of v1.2 and v1.3:


Vers 1.3 2018-08-11:
- make SaveConfig() safer by disabling ints, ensure 0xff's get written in old position, using library routine first
eeprom writes and not using eeprom erase
- limit mode set choices to just those sets that are compiled in, if it's 12 or 8 sets for example

Vers 1.3 2018-07-29:
- added the TWOCHAN_3C_PINS option, so a standard TA triple board can be used in FET+1 mode (no 7135 bank needed)

Vers 1.3 2018-01-28:
- added support for TA's driver for the new Astrolux S42. It uses the single 7135 on pin #3 and the FET on pin#5, with the Indicator LED on pin #7.

Vers 1.3 2017-10-16:
tweaks made to battery voltage level readings:
- The single cell mode using the internal V ref and no resistors appears to be more accurate now, rounded to the blinked out value.
- for using R1/R2, it should also be more accurate now - still in testing


----------------------------------------------------------------
Vers 1.2 2017-10-16
- changed momentary/tactical mode so fast clicks have no effect, and only way to exit is to power cycle

Vers 1.2 2017-10-15
- if strobes are disabled, a 2X click from 2X turbo should restore the previous level you were at
- BUG FIXED: 4X clicks in modes operation engages lockout - not supposed to
- operation change: make click&hold in MODES or STROBES wrap from 1st mode to last
- BUG FIXED: in LVP switch LED control: the LED sometimes is left on after an LVP drop, and is not
blinking the way it should be. The "bug" is that I'm trying to control the switch LEDs from
multiple places so it's getting turned off quickly after turned on, and left on when it should
be left off. It's a timing thing, so has sort of a random pattern - sometimes left on, sometimes
left off. The 8 sec LVP blink need to be qualified better, and should not be calling Setlevel()
as is because it wants to control the LED
- BUG FIXED: temp stepdown should not happen right away from turn ON. Delay it by 15 seconds
- BUG FIXED: for temperature stepdown, in moon mode, a temp stepdown is actually done because moon
mode is marked as special level 255, which is considered as a high level of output instead of very
low. The stepdown results in the light switching much brighter. This should only happen id the
temperature threshold is set to a low temp, or the light is still hot and not cooled down when moon
mode is chosen. It can also result in an immediate jump to the stepdown level as soon as the light
is turned on in moon mode. Might appear as a bright flash when ramping first starts.

Something like olight maradur 2 with both flood and throw.but not as big.with one single 21700 would be new and great

Probably too much cost involved for Sofirn’s price range. Sofirn only does so well with sales because the prices are so affordable.

But if Sofirn does endup making something like it budget friendly would,nt it be great.just thinking

In general I absolutely agree with Funtastic here… Sofirn offers VERY good value in a light (in general) that usually follows another great design from a higher-end maker, but for much less.

BUT… they did surprise us with the LT1 for sure! They have been friendly to our market by often accepting design advice from BLF (they take it often just like Simon at Convoy does too). And for the dollar spent… it’s just real hard to beat a Sorfin light. They’ve also delivered some other decent “newish” lights over the years (Q8 and C8F are other great designs!). Sofirn also stepped up to Anduril this last year where some other manufacturers aren’t keeping up.

I feel Sofirn is now at a crossroads where they will either keep innovating, or start to fall off in quality. I don’t think anything is wrong there today, but they (like ALL light makers in China) have been pushing price up and when you do this you HAVE to keep quality UP and innovation moving FORWARD. Else, you end up in a “death spiral” trying to sell the good old design that did so well… last year. Like any market, you innovate at cutting costs or cutting edge (or some of both). Sofirn seems to be firmly in the middle there and I hope they “creep up” with some new stuff like they did with the IF25a and SC31 Pro in Q3 and Q4 this year.

So fair to say that for their market, they have done very well the last 3 years or so. And as Funtastic is a dealer he’s been watching that market for a while and is probably correct in limiting our expectations from the Sofirn brand. But can’t hurt to ask- right! Who knows? They have surprised us a few times and time will indeed tell :smiley: