[Modding] Astrolux MF01 --> OSH park NarsilM

Nice! I’ve been on the fence about this mod ever since I noticed that one bank of leds is a tiny bit dimmer than the other two when in low mode.

It looks quite do-able and I love ramping on my other lights (Q8 and D4).

Any challenges or bumps in the road to look out for? How was it taking off the bezel to get at the emitters? Wasn’t there a warning from Astrolux about removing the bezel and damaging the leds?

So in other words making a moonlight with very low duty cycle PWM should increase LED life while the same moonlight with constant current will make it wear much more rapidly (altough with little more lumens per watt)…
I wonder if 20+ kHz range PWM can have any negative effect on LED though…

PWM increases the LED life in 2 ways the low current age is a lot less plus the LED is only 1/100 of the time on
so that would increase the lifetime by a couple hundred times at least

new version of MF01 needs to get a MCPCB modification as well

you can ignore the cables I put in there, basically you can wire it like you want, just make sure the red circles and the black circles are connected towards each other

the cables of me are in the 3 120° sections just different approaches

you cut so that 2 LEDs are together in series, while to board had originally 6 in series, so 2 cuts per 120° section

Wouaaa just 2 small wires. The current is lower?? What the new configuration? Old was 2S3P right?

new uses a boost driver to 20V LED voltage

Are you sure about this? Doesn’t this also mean that the new version with boost driver is able to maintain turbo output with depleting cells?

absolutely sure its 6S on the output of the driver

It wont be able to deliver Turbo much longer, the light pulls 120W from the cells and at like 30% discharge of the cells, it will fall out of Max brightness
The 2S driver can go to DD if the Buck regulator falls out of regulation while the boost driver gets more and more inefficient with dropping cell voltage and higher input current

Thanks! Freeme confirmed this as well. Really interesting development here. Ok I understand less cell voltage means more current is required for 120W with boost driver. But how how do you know the figures that with 30% discharge it will fall out of max brightness? Secondly, how many lumens would you loose? Just 5, or 25? Isn’t an actual test required to know these values?
I do know from my own ceiling bounce test using the M43 as a reference that with 3.75V for the 30Q in the MF01, I get less output (about 6000 lumens) than my M43. I’m curious then how many lumens the new boost driven MF01 will pump out… Output as a function of cell voltage is what I’m after.

This depends mainly how efficient the driver can boost without getting saturation in the inductors or other hardware limitations

keeping an eye on this one, great thread.

Basically this new MF01 needs to be tested again. M43 can push turbo with cell voltage as low as about 3.5V with GAs (my own test), and the TM38 can push turbo with cell voltage as low as 3.3V TM28 4x XHP35 HI Review Video. New TM From Nightcore
Of course different settings, but we’re talking boost drivers here. Curious, curious, curious.
Seriously, if this new boost version can push 10000 Nichia lumens with cell voltage less than 3.7V, I’m probably getting it (again)! :sunglasses:

LED lifetime in ultra-low/moon mode is in any case academic. Whether very low constant current, or PWMed at much higher current.

They will last forever, in our usage.

People who need to understand the detail are making e.g. streetlamps or warehouse lighting that must last 25 years plus, on for maybe 12 hours/day on average in the northern latitudes, where I live. Cree e.g. provide copious data for professional engineers to deeply understand, run the LEDs at their most efficient current level (lumens/watt), understand how the silicone domes can degrade due to environmental contamination, even incorrect choice of e.g. o-ring seals than can outgass degrading chemicals, this is more and more a precisely understood engineering subject, and the likes of Cree etc. are continuing to refine their products for their real markets, which is not us.

Worry more about our over-driven turbo etc. modes, where heat and current density kills.We are usually pushing these things way beyond datasheet parameters.

Except for things like the MF01, where the extravagant use of e.g. 18 decent LEDs to kick out e.g. 10,000 lumens i.e. 555 per LED is very modest indeed. This is one of the last “mega torches” that I would have any concern about over-stressing the LEDs, assuming good thermal protection, which by reports it seems to have, maybe even over-cautious.

10 minutes in turbo getting the torch roasting hot (who knows what LED junction temperatures, upper limit is melting point of solder, where the none-lead stuff is better) might be equivalent to decades in moonlight.

Interesting to discuss, academically, but of no practical relevance, IMHO.

Indeed. Everything has changed. And I have high hopes. Even if not perfect, I like the direction they have taken, it has so much potential beyond simply connecting a cell to an LED via a FET, or a buck driver.

yeah in reality you get at least 10000h to 50% in CC moonlight
what often people underestimate is thermal shock during reflow, it can make an LED fail after 30-50h

Yes, flow (not re-flow) is very relevant. Manufacturers will publish a profile for soldering, based on commercial equipment, starting with a pre-heat at 100+ C to bring everything up to temperature and dry out the solvents from the solder paste, then a sharp ramp up to melt the solder, even as high as 250C, then a sharp ramp down again to the safe 100+ temperature, then a gentle cooling. The best commercial equipment will even do all this in an inert atmosphere . The pre-heat phase and subsequent cool-down is very important, but not something easily done with DIY equipment, but it is possible to get quite close, if interested.

Pre-heat, and slow cool down reduces thermal shock immensely.

I was just getting ready to put the cells on the charger and noticed they where at 3.68v so i took it out and tested it.Started out at 9670 with 9220@30s.

Hold on… let me repeat what I just read here… 3.68V… You put them in the MF01 second gen (with one pair of wires to led board)… And you get 9670 lumens at startup, and 9220 lumens at 30 seconds? Cuz if this is so… then I’m so getting some more… :smiley:

(P.S. With 3.7V I get about 6000 lumens from my first gen MF01 219C)

Yes once my 1st batch MF01 drops below 3.7v the output is around 6,000 lumens. I am rather interested in the new configuration.

Wow, thanks a lot JIGHEADWORM for measuring. I believe level4/high mode is reduced from ~5500 lumens to ~3000 lumens as well. This means the light should run almost continuously on this level without stepping down as well. My concern is that another member here (patmurris) only measures about 8000 lumens instead of ~10000. So far things look really good though. :+1: