[Review] Sofirn SD05 Diving 21700 Flashlight

Yep, just the light. Delivering to Germany you have to pay tax and customs for everything over round-a-about 26 EUR so as a kit the torch was a lot more expensive. I already ordered a Sofirn 21700 battery separately but it hasn’t arrived yet.
But it seems to be the old trustfire battery, so i can calm down. :smiley:

This light will not allow current over 7.1A so you don't have to use a protected cell.

This is an awesome light the perfect size and i think its dang near as bright as both of my Sofirn C8F’s and my sp33 , by the way my ring does not rattle at all.

^THIS The most important feature of a dive light. The Sofirn is inexpensive, so it it works for quite some time before the stuff rusts out……I don’t think you get high quality from China at that price. Jus sayin…mines inbound as we speak.

Sharks Bay, Egypt

Daytime shot, depth - 10-12 m

An actual dive shot, nice! Looks like fun.

Hi gchart! I got my SD05 a couple of days ago, and decided to piggy-back on your topic a bit. I bought it mainly as a diving light (I scuba dive every now and then), and while I'm very impressed with build quality and light output, the timer + temperature is a major flaw IMHO. Thermal throttling thankfully is almost a non issue underwater, but the timer makes it super annoying to reset the light every 3 minutes, which tends to be quite difficult in drysuit gloves. I decided to tear the light down (driver board comes off easily once you unsolder the wires from the front) to see if re-programming the driver would be possible, as that would also give me a chance to adjust the levels (I find the difference between Low and Medium to be too subtle, and it's a shame it's missing a strobe mode). Unfortunately it's not a simple ATTiny based driver, but something a lot more complex. Haven't seen any driver pictures on the forums yet, so here they are:

https://imgur.com/a/CEJYJYj (higher resolution in the imgur album)

As far as I can tell, the main MCU is Microchip F1824, and clearly visible there's a boost converter, namely TPS61088. The modes are selected thanks to Hall effect sensors (3 of them) that respond to the magnetic ring.

Have you seen anyone writing firmware for such MCUs here? I can reverse engineer the schematics for the driver board, but I'm lost as far as programming is concerned. I feel like building a new driver based on an ATTiny would be much easier, although I'm not sure how to implement the Hall effect sensors in the firmware.

Would you be interested in helping to improve this light a bit? I've seen your designs (2 of my Convoys are running your tailswitch light boards) and I feel like you're the right person to ask.

Would love to hear your thoughts!

Interesting… I wanted to take over apart (the first one I received appears to have a defective driver, it doesn’t turn on), but it seems like the battery tube is glued to the head. Was yours not glued?

There’s a couple people here that write firmware for PIC, but not many (tterev3 comes to mind). I could probably pick up the skills, but it’s not going to happen at the moment.

The best I could probably do at the moment would be to create an adapter board and piggyback a attiny816 in place of the PIC. Would take some probing around to confirm the function of each pin.

Battery tube has double orings and a normal thread towards the head, it seems like there was some sort of thread locker on there as well, but if you grip it tight it’s going to give in eventually. After that you just need to desolder the wires from the LED and you can take the driver out.

I’ll be happy to help in any way I can - I’ll start mapping the pins and maybe reconstruct the wiring diagram. It really is a great light, but this timer/ATR fiasco really grinds my gears. I’m open to any suggestions, so if you want me to do something, let me know!

Agreed. Timer step down in a dive light is one of the dumbest things I’ve heard of. I generally really like Sofirn stuff, but that was a big oversight on this light.

Mapping pins would be helpful. I’ll have another go at getting my light open.

Happy Holidays!

Thanks for the board shots aabatteries, they helped me figure out a few things. I replaced the current limit resistor with a 100k one as per their data sheet. Not quite sure it made any difference but ill add a pic once I have it apart again.

The big difference was adding a 4000k xhp70.2 in place of the XHP50. Efficiency is higher per Texas Aces (Thanks TA!) output tests, and it fits perfectly with no plastic ring.

Beam shots and such:

XHP70.2 Modded Sofirn SD-05 XHP70.2 Modded Sofirn SD-05 - Album on Imgur

Shot of the reflector:

Shot at ISO 100, 1/20 exposure on OP7 Pro at 2.5 feet.

Now trying to change duty cycle by adjusting feedback. I think timer is dictated by a capacitor if im not mistaken.

I’ve just picked up an SD05 in December and bought a Samsung 50E 21700 to run it with. I took it out mountain biking the other night to test its use as a bar mounted light and didn’t notice any stepdown behavior at all. I am confused as no one has mentioned that Sofirn revised this timed stepdown (that I’ve seen). It was about 24ºF outside and I purposely rode with the light set to high for a good straight 7+ minutes (thought it was longer, but my video shows just over 7 minutes) looking to spot the stepdown. I kept my helmet mounted light off to be sure to see the stepdown too. After that 7 minute segment I then switched the light to medium and saw a noticable drop in output. I then switched back to high and believe the output went right back up to where it was. I will say that high doesn’t look like 2,500 lumens to me comparing it side by side with +/- 1,500 and 2,000 lumen bike lights (which I’ve measured myself in my DIY sphere), but maybe it’s just “optical illusion” and I’m not a good judge of output by eye. I’ve not done any tailcap current testing or sphere testing (yet), so perhaps I need to in order to figure out what is going on (let ’er run over 3 minutes with and without a fan blowing to see if it steps down and how much if it does). Does anybody have any information on Sofirn getting rid of this stepdown?

Also, am I correct that this light is using the 6V version of the XHP50.2 with a boost driver providing regulated output? Am I correct that there (so far) has only been one version of this SD05? I think I was confusing the SD05 with the SP33 thinking the light had been revised to the 3V XHP50.2 with an FET driver.

I don’t particularly light the emitter tint (looks too cool to me and not what I’d call neutral white at all - not even the cooler side of neutral white), so what do you think about swapping in a /- 4000k XHP70.2 like shown above? What should I expect after the emitter swap as far as output lumens and runtime and also heat? Since this isn’t just to be a “wow light” but potentially as a bar light during/-2hr mountain bike rides (mixing medium & high mode use), does this sound wise or not?

Here is a beamshot pic from the other night with the light on high (taken before the ride even started, so on a fresh battery). White balance is slightly off, so it appears slightly cooler than it looked in reality:

To compare, here is a Sofirn SP40 (4000k version) on Turbo (decent full battery, but not a high performance cell):

-Garry

@ garrybunk;

Quote: “I took it out mountain biking the other night to test its use as a bar mounted light and didn’t notice any stepdown behavior at all. I am confused as no one has mentioned that Sofirn revised this timed stepdown (that I’ve seen).”

That would be wishful thinking. There is no mention other than timed step-down on their site. It is by far the biggest gripe we enthusiasts have, this timed throttling for an underwater light - albeit in ambient air a thermal roll back would suffice.

As for your battery, a Sam 50E has great capacity but not much current capability; about 10 amps. It’s internal resistance is a bit high at 40 mOhms. This would, perhaps (?), account for your lesser luminosity output and the non stepped down.

Prior to purchase I could never find any official CCT for this light. There were a few members that did mention it as cool white. This does suit me fine as throwers are better illuminators in the higher temps, the rods are more excited versus the cones’ color rendering at night. But then we all have our preferences…

I went back to revisit the light yesterday. After reading your comments I went ahead and placed the torch on my windowsill with my camera pointed towards it, and started recording with a fixed exposure (so I could see the drop in brightness level). After 7 minutes of recording nothing happened - there was no change in the output (!?). I was running it on a semi-charged (~3.6V) batt. The torch didn’t heat up a lot, it was just mildly warm to touch. I’m starting to wonder if they actually made a revision of the board, or changed the firmware at some point. I got mine on Dec 2019.

@gchart can you confirm that your version has the timed step-down included? If yes, then can you tear it down and post pictures of the board? Maybe there actually is a revision, but then I would be absolutely baffled as to why they did not communicate this in any way.

Anyone else who has this light, and confirmed the timed step-down - please chime in, and let’s figure this out!

Also, @FerociousBear, can you give some more details about the mods you made? I would love to do the same emitter swap, which one did you choose exactly? And which resistor did you change? Thanks!

I think you should fully charge the battery and try that test again. Maybe even put a fan blowing across it to be sure temperature regulation doesn’t kick in. But thanks for chiming in.

-Garry

I just checked tailcap current on mine and with the Samsung 50E @ 3.98v it shows 7.4A for just a split second and then drops to about 2.86A. Medium shows 1.51A. So I’d say you’re test needs done with a full battery. I’m going to fully charge my battery and test again too. (But I definitely had a full battery when starting out the other night.)

-Garry

I recorded a few tests, and unfortunately the step down is still here. It’s just so gradual that you won’t notice, and still if you drop down to ‘M’ you still see some more output drop, and obviously resetting to ‘H’ makes it much brighter.

Here’s a short album with the recordings: https://photos.app.goo.gl/1wigwdVPTGCu8CYaA

I’m thinking it might be easier to create a whole new driver board, but the Hall switch part has me worried as I have not seen any projects/firmware that would work with them easily.

Yeah, I believe you are right - that the stepdown is still there, but is barely noticeable. I read somewhere (included manual?) that the light makes multiple gradual steps down from high after the 3 minute timer. I also think that the max output you get from High is dropping so fast because of the high current draw from the battery (and in my case probably dropping faster than other cells). So in other words, you’re not hitting that “2500” lumen level after you’ve used the light a few minutes - so when you click back up to high after some stepdown you may only jump back up to say 1600 lumens. I’d like to do some testing of this “high runtime” and am thinking to rig up a DMM measuring tailcap current while the light runs in my DIY sphere with a timer running and my action cam running timelapse. Of course the test will really only be accurate for anyone else running the same cell. I believe high mode is really only usable for just a few minutes from start and that this is mainly a medium/low mode light. They probably should have just made modes Low-Med-High-Turbo and noted that Turbo was only usable for a few minutes on a fresh cell. Anyway, I was counting on using Medium mode for MTB riding anyway - a nice regulated 900 lumens for 2.5 hours.

Last night on my fully charged 50E I was seeing high mode hold at +/- 7.5A for about 15 seconds (that’s as long as I tested it) and not just immediately drop to medium level, so I know the last test was due to the battery level.

-Garry

Okay, I’ve done a runtime/thermal analysis of the first few minutes of operation of this light on a freshly charged unprotected Samsung 50E cell. In the video below you’ll see I am monitoring tailcap amps, head temperature, room temperature, runtime, and lux in my DIY sphere (not yet converted to lumens). Note that I did have a small fan blowing across it to simulate use during cycling (little air movement). I did turn the fan off and let the light run on high toward the end of the video. At the end of the video, after a short rest, the cell measured 4.06v which I thought was really good. Disregard the idiotic narrator’s comments about the 4 minute stepdown instead of it being 3 minutes. I, . . . ahem . . . I mean he, forgot that switching the light from high to medium and back to high had reset the 3 minute timer.

Video Link: https://youtu.be/1wagnsYqzFw

Things to note:

  1. Yes, a 3 minute timer is present. It then makes a slow gradual stepdown to a medium level taking 40 seconds to do so (though switching the switch to medium mode will drop output slightly more - from +/- 1.50A to 1.04A).

2) On a freshly charged cell at the start of the test I saw +/- 7.0A, climb to 7.17A, and then drop to 2.70A where it remained for the duration of the 3 minute timer. That runtime from 7.0A to 7.17A was only 13 seconds in my case!

3) Turning the switch from high to medium and back to high only bring its back up to the 2.70A level.

4) The output in high after stepdown, the stepdown level itself, and the “actual medium” all appear to be very well regulated output levels.

5) This light has plenty of mass to handle the heat at the 2.70A level without a fan running, reaching 99ºF after 5 minutes and 40 seconds in my case.

I will be doing further evaluation of this light during a night ride tomorrow. I expect to have more videos and photos. I will be creating a thread over at MTBR’s Lights & Night Riding forum and take the majority of my posting over there.

-Garry

Anyone else notice that lumens on Medium change when stepping up and stepping down?,

Low to Medium 760 lumens
High to Medium 920 lumens

This is manually changing modes, not from thermal or timer step down