The new clip looks pretty promising. White anodizing is interesting too, and very uncommon, though personally I think I’ll stick with my green one. I should maybe upgrade that at some point; it has been getting neglected.
About the Ti/Cu model and heat… yow. Zero to finger-burning in 7 seconds really sounds like it’s begging for less-crazy emitters. I’ve done some things to improve thermal regulation recently, but there’s basically nothing the firmware can do about heat in just 7 seconds except maybe something totally draconian. It could limit turbo to 3 seconds or something, before dropping to 50% power.
It should have better thermal regulation than older builds, since I made some significant improvements over the past few days. But it’s still not going to properly handle 25A of 219c full turbo in such a small host… because that’s getting dangerously close to a flash bulb instead of a flashlight.
Way back when I was trying anodizing for myself I read up a lot on the process. They said that the anodization process makes honeycomb like “tubes” in the surface of the metal and the dye goes into these tubes and is sealed in during the final stage. The most common is black as the physical size of the black dye molecules fits into these tubes easiest. White dye is too large, they said you’ll never see white anodization because of the physical disparity of the tube size compared to the molecules of white dye.
It was also stated and confirmed recently that the anodization process aids thermal dispertion of aluminum, the dye process does nothing for this thermal dispertion, including black.
Yeah, I’ve always wondered how Hank made the Meteor white. Did that mystery ever get resolved?
Yes. But imgur’s copy is out of date. I don’t know who uploaded it there, but the only places to get up-to-date info on it are the FSM thread, my site, and the firmware repository, which all have the most recent copies.
FWIW, the improved thermal regulation probably won’t step down from turbo as fast as the original D4 firmware, but it’s much smoother and more configurable and less prone to oscillating. And the UI has a whole bunch of other features the original D4 doesn’t have.
FWIW, contactcr has been helping me test the recent changes. Here’s the result using a D4-219c with Anduril from yesterday. Thermal ceiling was set to 45C, and it had air blowing past it.
Without any form of cooling though, just tail-standing in still air, it hit the floor and stayed there unless the temperature limit was higher:
Since then, I’ve lowered the “elbow” level where it switches from a fast drop to a slow drop. So it should work a little better in both cases… but still. it’s important to have at least some form of cooling when running a D4 at turbo. Hold it in your hand, point a fan at it, dip it in water, whatever… but it needs something.
Due to the introduction of an elbow level value, telling the code approximately where a safe-ish level is, I added some new Anduril build targets for D1 and D1S. They’re not massively overpowered like the D4, so they don’t need to drop quickly when at high levels. Their “elbow” levels are about 75% and 90% power, respectively.
I tested the FW3A yesterday too. Having a 7x7135 medium-power circuit between the 1x7135 and FET really seems to help. It’s almost completely flat after the initial drop, and sustains a decent amount of output. It has an odd plateau first though, before the driver can really feel the heat.
Just got done rebuilding my Ti D4. The Nichia 219C’s were just way too hot so I replaced them. About to find out if I was successful… charging up a brand new Aspire 1100mAh 18350…
Oh, almost forgot… while the new Noctigon quad board is nice and thick, they missed their intended mark when it comes to screwing it down… the head has threaded holes on each side so they did mean to make it secure but there isn’t a relief in the masking on the MCPCB so the screws short ground and cause the light to come on without the switch. So I used a T-PAD board and screwed it down snug, clipped 2 legs off the Optic so they sit on the screws. Poifek!
Edit: With the same Efest 18350 I used before it now makes 3957 lumens at start ON THE 18350! I’m able to hold it for 37 seconds, hot but the driver is stepping it down so it’s mo bettah!
Edit II: Oh yeah, I used XP-L2 5000K emitters, sliced and diced to fit under the dome and tame the irregularities…
Edit III: And just in case anyone was wondering, with the Sony VTC5A from yesterday it now makes 5568.3 lumens at start. Will be running the 18350 though…
Dale, when you test those, can you please tell if there is a significant shift of the tint (yellow, green), please?
I’ve been thinking of modding some XP-L2 emitters to make them “HI”, but never seen the final result of how they end…
I’m just curious if Hank (Intl-Outdoor) accepts reflashing of the Emisar D4 (or the other D1/D2S) to the latest version of Anduril if ordered from his website? Even if it has a nominal additional cost? I got a D4 with the v2.0 firmware. I’m not skilled enough to reflash the D4 to another firmware… I’d like to reorder another D4 and would like it to have Anduril pre-installed…
(I think Mtn-Electronics has this option of reflashing firmware, but shipping costs around $20 from Mtn-Electronics compared to the $2 (for tracking #, since Intl-Outdoor won’t allow shipping to my country without a tracking number) of Intl-Outdoor…
Not that I’m aware of. I don’t think he takes orders for custom firmware.
If there is a RampingIOS V3 though, it’ll be FSM-based and incorporate some of Anduril’s improvements. Even if, for no other reason, it’s easier to clone the old UI in a new toolkit than it is to make significant changes to the old code.
contactcr talked me into ordering some Samsung LH351D to try in this light, so those are on the way. Should be a nicer beam profile with better overall color management and could end up making even more lumens. These will be my very first Samsung emitters so I really have no idea what to expect. I got the 80CRI 5000K variants…
I think I will make a jig to hold the original MCPCB and use a 4-flute mill bit in a drill press to clear the trace from the screw holes, use the original thicker copper board for the Samsungs when they get here.
Edit: Cleared the trace, made some compromises so it’s not as neat as it perhaps could have been… my end mill bits are out in the shop as is my more rigid drill press, I used a diamond rotary tool bit that was in Old Lumens kit and only went so far as the smaller drill press on the back porch. It’s wicked hot, temperatures are down as we have rain impending but humidity is ridiculous so the heat index is way up there anyway! lol (It’s “only” 95 today)
Oh, um, I told Hank he should use LH351D, maybe even two 4000K and two 5000K to get the otherwise-unavailable 4500K tint people like. I have no idea if he’ll do it, but at least now he knows they exist.
I have noticed a couple of interesting things with this build compared to the code delivered with the D4. I believe the delivered D4 code flashed twice when the battery is attached, this build flashes once. I also noticed there was a bit of a delay after battery install until you could power the light on with the delivered code, that is removed in anduril. I also find I can get the light in a mode where if I turn on and hold the light starts in moonlight, but then jumps to probably full 7135 current when I release. Not sure what that is, I have not yet gone back to read the code description to see what I am doing there. Finally, when I disconnect the battery, it takes many seconds to minutes before reconnecting the battery will generate a flash. I also found that just clicking the switch with the battery disconnected will reset the driver and then reconnecting the battery generates the flash.
I believe I read somewhere that anduril is more efficient, and my DVM observations seems to support that. I decided to monitor the current during start up and such, I found in moonlight mode the current is 1.5mA, less than the 5mA I had measured with the delivered code. I also found the light would work with my DVM attached to measure current, which was not the case with the delivered code.
I have not fooled around much yet in the higher lumen modes to check out the thermal improvements, but all in all I am liking the new code in my D4. Nice work ToyKeeper!
There are quite a few differences… the code is completely different, a from-scratch new firmware.
About the moon before going to a higher level, that’s a UI hint which didn’t exist in RampingIOS. Basically, there are a few regular-mode actions available from off:
Hold to go to moon, then start ramping up.
Click for memorized level.
Double-click for ceiling.
But as far as the light sees, here’s how those events look:
Hold:
Button down.
Timeout expires.
Click:
Button down.
Button up.
Timeout expires.
Double click:
Button down.
Button up.
Button down.
Button up.
Timeout expires.
IIRC, RampingIOS responded this way:
Button down: Do nothing.
Timeout expires: Moon.
Or…
Button down: Do nothing.
Button up: Memorized level.
Timeout expires: Do nothing.
But Anduril responds a little differently:
Button down: Moon, immediately.
Timeout expires: State change to “on” state. Nothing visible changes. If user keeps holding, start ramping up.
And…
Button down: Moon, immediately.
Button up: Memorized level.
Timeout expires: State change to “on” state. Nothing visible changes.
Or for a double click:
Button down: Moon, immediately.
Button up: Memorized level.
Button down: Turn off LED, in case this is going to be 3 clicks or more.
Button up: Do nothing.
Timeout expires: Turn on at ceiling level and change state to “on”.
Basically, Anduril responds a little quicker by giving a UI hint which says “hey, there’s moon here, if you keep holding”. It turns on as soon as the button is pressed instead of waiting. And then it also does some other things without waiting for timeouts to signal the end of an input sequence. But it doesn’t know for sure what the user is doing until that timeout happens.
Thanks for explaining the details TK, I definitely appreciate it.
FWIW, if anyone is taking notes, sunset takes about 64 minutes on my light to completely set. The next challenge for TK is to implement the green flash. May not be too hard with the right emitter choice . . .
