The thing is, in our application, i.e. running LEDs at very low currents, in the 100 uA region, we are pretty much “off data sheet”.
Typically datasheet Vf is specified at currents of 10s of milliamps, but below that the transfer characteristics can be highly non-linear.
So yes, a green LED with a datasheet Vf of say 3.3 V may well emit our desired levels of light at much lower voltages, but it may not be a region that the manufacturer tests for, nevermind commits to as a guaranteed characteristic parameter, and might be highly variable from batch to batch, even within a batch…
For others’ interest I came across this tutorial which explains the basics:
The most reliable way to use an LED to consistently produce these low light levels might be to drive it at the lowest current level that is fully characterised (much brighter), then PWM it down to the desired level. If you have an MCU driving it already, and sufficient free internal resources (timer-counters etc.), maybe it could be done with just a little more firmware, and a resistor value change.
In an alternate reality, they would immediately issue a new absolute shipping day after they discovered a problem (preferably in a few days). They would then diagnose the problem, get delivery after finding they needed a replacement supply of parts, quickly rework the switch boards, and then deliver the2000 remaining lights to the carrier on the exact date previously stated. And send a repair team to each site that has received one of the first 500 and bring all of those up to spec (should only take an hour or so if there are the required 500 teams (with slack for hurricanes and other flight delays). It would be polite if each owner would arrange for good Chinese take-out while they were there.
In the reality we are forced to live in, they are proceeding exactly like they should and would/should/could absolutely not issue a new shipping date without confirming that the problem is resolved and all updates to the 2000 lights done. Then work out a protocol for replacing the problem part(s) in delivered lights. With the direction testing is leaning, this may have to replace the switch boards. The fault could be a problem with the Q8 electronics not providing enough voltage to the LEDs. And if this is the issue, it is further likely that just about everyone with bad switch boards could have the same problem at a later date based on the charge of batteries and maybe burn in drift.
Indications of ongoing testing (going on as we type) indicates a possible problem with the quality/Vf variability/incompatible Vf specs for the color and type.
Both of mine arrived today; they were ordered within minutes of each other the first day they went on sale.
One is perfect so far. The other has a couple issues we've seen, and (I believe) a bran-fresh one (lucky me ;) ).
First issue is asymmetrical LEDs in the switch. Here's a pic of both, and while it's not a great picture it shows a smaller bright spot on the right side of the right light's switch:
The second issue can be seen in this picture. There's an obvious flaw in the tube/anodizing. This came out of an absolutely unscathed box, so I don't believe it occurred in shipping:
Here's a close up; not sure what caused it tbh but it's not a simple dent.
Interestingly, the light with the flaws pictured above also has a battery tube that's not correctly aligned with the head when fully tight.
Frustration is understandable given the anticipation of this exciting light. Maybe I’m a persimist, but I recognize the monumental task of getting a complicated piece of electronics manufactured half-way around the world. The barriers are enormous. It seems to me that off the shelf, stock flashlights have a higher than average failure rate compared to other commodities.
In fact, I buy most of my lights from modders like Vinh Nguyen, not just because they are bright, but any problems have been fixed in the upgrade process. Not to mention, I’m dealing with someone I consider a friend and warranty work is a fast, state-side turnaround.
So, why go for this given my pessimism? Because this is a great flashlight design, backed by a hard-working team who did their best. And let’s be honest, where pray tell can you find a better bang for the buck in a flashlight. Wasn’t it worth taking a chance? And the outcome seemingly has only one insignificant glitch, which has nothing to do with the reason we bought it.
This post is to say thanks to the team!!!
Thanks for mentioning tube alignment ‘feralcomprehension’. :+1:
I gotta go back & check that. Nothing jumped right out at me…. but never can tell, I was paying attention to other stuff.
Instead of shaking your fist at the empty sky demanding to know when it will rain again, this might be a good time to do something else. Go for a walk, talk to a friend, laugh at cat videos, rock out to loud music, argue with trolls on 4chan, do some cleaning, whatever works for you. A delayed flashlight isn’t worth raising one’s blood pressure.
I’m anxious too, with a project or two waiting on a Q8 hardware, but there are plenty of other things to do in the meantime.
I’m not sure why people are expecting the tubes to be aligned ? Do you expect them to be individually indexed, or selectively fitted, or so precisely machined that every one is perfect, and interchangeable ? Do you know of any SRK type design which does so (certainly not the classic SRKs.)
It is a feature of the design. Even if somehow all the above was done, after a bit of use and wear, the indexing would shift anyway.
The alternative is I suppose a bland cylindrical tube.
There are reasons why people around here laugh at data sheets.
Not an option. With the MCU doing PWM, it would be about 0.1mA for the indicator LED and about 5mA for the MCU. It can’t do PWM while it’s asleep, so it would drop the efficiency and standby time to about 1/40th of what it is now. It would be pretty much the same as leaving the light on in moon mode all the time.
Edit: I did find a potentially-relevant trick on my old Ferrero Rocher light though. I was able to get three brightness levels (high, low, off) from an indicator LED in sleep mode by messing with the pin mode. High and off were fully within spec, but for “low” mode I had to go off the data sheet and do things the MCU didn’t really support. I suspect it might not work on a Q8 though, because it probably relied on the voltage divider resistors which newer drivers don’t have any more.
Better they stop production and figure out the issue and change whatever needs changing than them continue putting out hit or miss products. As for a definite time that is impossible to guess at until the solution is decided upon, parts if necessary ontained, etc.etc.
Received mine yesterday. Real fine light. Looks and feels great. My wife digs it too. I immediately bypassed the springs and drilled the driver mounting holes to clear the mounting screw holes. Neither were necessary, just my preference. The light is the real deal. I measured 6K lumens in partially depleted 30Q's. The light get hot fast like it should. Beam pattern is a great mix of flood and throw.
This is my first time using Narsil. Love the ramping and natural feel of it. I got lost a couple times in the UI, but that is probably just operator error. The indicator light is very bright for my taste (Think I will swap in a less harsh red or amber emitter). The indicator emitters turns off when the light it on. Is that by design?
Very happy with the light and I totally understand the frustration of those awaiting to receive theirs. Big congrats to The Miller and team. First light I don't feel the need to do any major mods on. Hang in there guys.
