Output is about 900lm with active cooling. Could you also measure output with out cooling?
Test 5 looks great! And yes, a version minus fan cooling and plus hand cooling seems most realistic to me. And given the nature of this light (hotrod) I would not mind the thermal limit to be a bit over 50 degrees for some higher regulated output.
What kind of stable output are we looking at in test5 btw? My guess would be 600-ish lumen?
Thanks for the fine review.
I prefer the size, tint, beam and UI of the Nichia D4 much more than any of the Manker, Astrolux or similar lights popular here. I rarely use them because the UI just isnāt simple enough for simple me.
Double-click the D4 and you will immediately see more light than you think possible from anything this size. Within seconds it will be too hot to hold.
But use the ramping, and, ah, you can get whatever level is perfect for you.
The Nichia tint and beam are just right for walking the dogs in the dark. The beam is floody and the tint seems neither too yellow or white ā but just right.
I am using an unprotected 3500 mAh 18650 I purchased from ZebraLight. Lack of a clip is a nit as is the absence of more knurling. I added the deep carry pocket clip available from Richard for $1.10.
The 18650 body is small but large enough to comfortably hold. I passed on the 18350 body.
Darn good light ā eager to see whatās next.
This active-cooling test was just a sanity check to make sure the regulation doesnāt do anything dumb. The 45 C limit was arbitrary. The thermal ceiling will be user-configurable, so you can change its stable output level by changing the temperature limit.
Most likely, the calibration procedure will be:
- Fast-click 10 times from off, then hold the button on 11th click.
- Light comes on at a low mode.
- To disable thermal regulation (or set it to a very high temperature), let go of the button now. Otherwise, keep holding.
- Light goes up to maximum.
- Keep holding until the light feels hot.
- Let go of the button to set the thermal ceiling.
Attiny85 chips use un-calibrated thermal sensors, so a factory default value will be inconsistent. It seems best for the default to err on the safe (low) side, with a config option so the user can set it to whatever they like.
Done. Had been posted there already, though. 
Is there any word about 18500 tubes becoming available.?
Mt. Electronics for U.S. should be getting his order of Aspire 18350 soon,
Iām not sure. I donāt trust the sensor on my phone to give me numbers which are accurate in an absolute senseā¦ only relative output.
So, 600-ish seems like a decent estimate; I donāt have a better guess.
Just a thought that occurred to me. Letās say 600 lumens is what the light can output comfortably and with stability:
1 click from off = low
2 clicks from off = 600 ish lumen
3 clicks from off = FET turbo
Easy access to a nice high mode of 600 lumens, and slightly more effort is needed for the FET turbo, which is reaaally a burst mode. Just a thought, and certainly not a request.
hIKARInoob 3 clicks ā¦??.. :+1: 
Edit: Beacon mode, would be nice if the light emitted was at a higher level or user defined like the M43, I know you donāt own one, but by holding down on switch will increase beacon light output and a much better brighter flash, maybe a slight bit more speed in between flashes also, counting the flashes in battery level is so perfect, but canāt for the life of me figure out the flashes for temp check, 10 degree Celsius per flash, starts when very hot hot at 4 flashes but as it cools will then start a second group of flashes like the bat check, so not understanding what each group represents, or why I am wondering about it actually, when itās hot it is real hot and when itās cool I can put it back in my pocket, no need to watch temp flashes I suppose, just there so had to look at it and try to understand.
Yeahā¦ I mean this light is just stupid bright that it gets too hot in, what 30 seconds or less? So this output is purely for wow or burst use. And that is perfectly fine, but personally I consider this output for my taste āoutside regular use of flashlightā, so I wouldnāt mind if this output is only activated with additional effort so to speak. Especially because this is an EDC light. Maybe the idea is comparable to turbo and turbo max for the X7.
Surely people still prefer a single click to turbo for this light; to each his/her own of course.
Reminds me of my smartphone telling me that it is raining today. And Iām like thankfully I have windows in my house, so I can look outside to see what the weather is likeā¦ Not really high techā¦ but it gets the job doneā¦
Only issue I think with the 3 click idea is that it would then cause TK to have to adjust all the other clicks since 3 click now is bat check, but nice idea I think, could be double click from off to high then double click again to fet turbo ā¦?
Ha Ha Ha, to funny, I know what your saying, phone tells me stupid things like that and makes me wonder why do I own this thing, my old flip phone worked just fine, until the phone company said I had to buy a new phone since my flip was so old it was the equivalent of the rotary dial phones, I liked them toā¦ ha ha,
I am not sure those phones are that smart but they are capable of scaring me sometimes, like how did this phone know that, and it wants me to turn on photo recognition, ha I donāt think so, making me paranoid nowā¦
Oh no no no, I donāt want anyone change anything because I said something: itās just me thinking loudā¦
M43ās UI1 is also great for this light when it comes to press and hold for momentary turbo (no ramping then). That would work for me as well.
Iām trying not to change much of Hankās design, so I donāt think the click patterns will change. However, it might be feasible to make double-click toggle turbo instead of being a one-way trip.
Additionally, with mode memory you can set the brightness to whatever you want, and then it is a simple one-click on/off at that level.
Mode memory is not saved across battery changes. This means it needs a default value. The default could be moon, 10 lm, level 65 (100% on 7135 chip, 0% on FET), or something else. Iām thinking the 100% 7135 mode might be a good default, since itād be nice to have a shortcut to the highest regulated level with no PWM. I measured it at 133 lm on mine. This level would be accessible at any time by loosening the tailcap, tightening it again, then clicking the button once. However, 10 lm might be nice too.
Sound like a plan?
BTW, the difficult thing about reading battery and temperature values is that ā0ā digits donāt get a blink. So, a temperature of 21 gets āblink blink pause blinkā, while a value of 20 gets only āblink blinkā. I intend to fix this by making zeroes give a very short blink instead of nothing.
That thermal regulation looks really good! Great job TK!
Yes TK the temp reading was impossible for me to figure out even after taking a actual temp reading then attempting to figure blinks to the actual reading.
First mistake I think I made was an attempt to read the blinks from the point of extremely high temps and seemed rather pointless because of how fast this little light really does cool down, it cools very quickly and will give you blinks all over the place, it doesnāt seem to have set temp levels to give an accurate blink reading but seemed to me that would have been thought of but mine at times would either just kept blinking steadily when very hot but when cooled quite a bit there was continuous blinks of sets of four and light was just warm, a cold light is 9 blinks so not sure what that translates to assuming in Celsius, 9 Ā°C or 48.2 Ā°Fā¦ doesnāt seem logical to me. ?
Edit: The Beacon mode is rather low,
I think the beacon light output must be set at 20% of max output and that seems way low to be very effective, it is also very slow and should be more like a flash if thatās possible, present light level is a good default level but it would be nice to have the ability to ramp up the beacons output to max if need be by just holding down the switch while in beacon mode like it is in the M43, Edit: Delete wrong infoā¦
Glad your on the upgrade team TK but also means I will be buying more D4 lights with the improvements or drivers for change outs if available and then Hank will probably have the other D1S light about ready I hopeā¦ Thanks for listening TK
Yes, I love it! (133 lm)
You make the best UI:s :+1:
Itās funny that people want the beacon mode to be brighter. I find it way too bright.
Maybe it can use the memorized ramp level instead. Thatās an easy change.
Anyway, Hank sent sort of a ācare packageā. It consists of a bunch of parts which (mostly) were either prototypes or failed QC, but can be assembled into working lights for testing and development. There are sharp edges and dents and other issues on nearly every piece. These are the kind of parts you wonāt get in production units:
There is one XP-L HI and one XP-G2, unknown tints. Iāll get measurements after theyāre assembled and working:
Of particular interest is that these show how the switch is assembled. It appears to be very compact, just a membrane switch and a rubber boot, with a ring press-fit over the boot to keep pressure on the parts:
Since I know someone will ask, here are the dimensions:
- Boot hole depth: 1.9 mm
- Boot hole diameter: 13.7 mm
- Membrane switch thickness: 1.3 mm
- Membrane switch PCB: 6.0 mm x 6.0 mm
- Switch hole: ~1.0 mm (?) x 6.0 mm x 6.0 mm (plus cutout corners)
ā¦
On a totally different note, I updated my thermal regulation simulator today to try to make it match what I measured on the D4. Itās fairly close, but I think the actual D4 gets less-hot and reaches its stable temperature sooner than the simulation. And for some reason I had to use attn=6 here instead of attn=4, but otherwise the parameters are the same as I used on the actual hardware. Hereās how it looks when simulating the āregulation test 5ā from earlier:
The most interesting parts are a bit easier to see if I give the simulator a smaller ābatteryā:
There is, um, probably a bit too much info packed into each graphā¦ but I think what it shows is pretty close to what I measured on actual hardware. So it seems like a fair model for testing changes without waiting an hour for each test.
My opinion about UI is
From off
Long click min
Click - saved ramped mod
Two Click - max mod
And some idea about ramping. Now there are 150 levels and it is impossible to tell which one is selected now. Idea is to make 2 mode of ramping one with 150lv and another with only 10lv. Mods are easy to distinguish and have nice spacing. The ramping time is the same ,but mods change every 0.25c