DEL may have had posts showing PWM's on the Q8 driver in the big Q8 thread - again, DEL is the EE who designed the original Q8 driver which is the basis for all the FET+1 designs we are seeing in the china lights, either officially (ex: S43S) or unofficially (ex: Amutorch AX3). We got a 47 ohm inline and a 100K pull to grnd on the FET gate - dunno if that explains it or not.
eas: is the soft ramp a good thing or bad thing? Boy, wish I had a memory - think I lost mine but forget when... 
IF the shape of the light measurement mirrors the pulses of current through the FET then it would appear that it is spending most of its time as a resistor, rather than a switch. I’m not an EE, but I’m pretty sure that’s not good.
The flip-side is that too steep a ramp can generate lots of high-frequency noise, which is not good either. It can cause interference in nearby devices (probably not an issue with metal enclosed flashlights) and make problems for MCUs.
very, very detailed. as usually )
thanx
So, got this light in this week with the 18650 tube. Is it normal for the direction of the tube to matter?
I had a heck of a time getting this light to work even with a fully charged Q30. Then on a lark I flipped the tube around - and it works now! Is this a thing with this light? I didn’t have any more time to fiddle with this last night, but is there something I should check? Thanks folks.
Sometimes yes. Whether it matters is very dependant on the light and tube.
In the case of the smooth S43 long tube, the ends look identical at first glance, but one end is about 1mm longer than the other.
Thanks Jason! It apparently matters with the short tubes as well. Same thing for the special edition Red and Green S43’s also. Flip them around and they don’t work.
One threaded end is a bit shorter than the other (as you observed). If the wrong end is screwed into the head the body of the tube doesn’t appear to reach the back face of the electronics and won’t close the circuit.
Red and greens? Wow, didn't think they shipped, but also someone received them. Ohhh, just checked, my green one shipped today, ordered Dec 25th.
Would love to show you, but I can’t seem to post pics using links to iCloud Shared Albums. Will Flickr work? Nope, can’t get the embed tool to work with Flickr either. 
imgur works well
I received the red version, i have to say the anodisation looks darker than on the picture, and it’s really cool, i did not think i would like this version so much.
And it’s so much lighter than copper…
Hhmm, that does sound nice. I am a huge copper fanatic, but I try to suppress my urges to buy anything and everything in copper - it works mostly... 
Mine does something weird…If I unscrew/screw tailcap, turn on the light in fet+7135 mode(2 blinks), it goes to lower mode automaticly in +–5sec and switch blinks 3 times slowly
If I do factory reset(hold button at firmware check), the light stays at the same mode.After I disconnect battery again, light does it again?
Either LVP or temp regulation is kicking in right away. You sure the battery is good?
Battery is a samsung 30Q and has 4V.I also tested it with another fully charged 30Q battery and flashlight has still the same problem, so I think the voltage doesn’t seem the problem…
Even if I set a new temp(to hot to hold), the light steps down at 5sec.
The only thing that works, is doing factory reset.
But after I disconnect power, light does it again
Ok - just went thru the code and only thing that would cause 3 slow blinks (1/2 sec between) is temperature step down. LVP blinks twice slowly (1/2 sec as well).
So it thinks it's over temperature. From what you describe, it's acting as if the temp setting is not able to be saved to EEPROM memory, and that there's bogus data in the EEPROM memory location where the over temperature value is stored.
I'm not sure what we could really do bout this - this is all in the MCU, so I would say replace the MCU, but obviously that's easier said than done.
Can you change other configuration settings and they stick? Another words they stay set after power resets? If so, then other parts of EEPROM memory seem to work ok.
You might try switching the temperature stepdown to a timed setting (1 minute). See if that fixes the problem. It might, but it’s not very practical.
I would do a factory reset, then use the light as long as you can without reseting the battery. Obviously you will eventually have to cut all power. Do the factory reset again. It might take several tries before it stores the info permanently.
Your issue sounds strange. Where did you buy it from? You might need to report it as defective and get a replacement.
Wonder if it's a fake Attiny85 that not quite to spec...
I think this is the first time I’ve heard of this particular issue, so it seems unlikely it would be a fake MCU.
NarsilM does have a known bug that pops up on the very rare occasion where it won’t store a change from the user menu. Usually you have to make the change a few additional times and eventually it will stick.
The thing with it activating the temperature step down super early sounds kind of familiar. Let me try to find some more info.
I know ToyKeeper has good details about the memory bug. Let me send her a PM.
Just checked - Anduril uses the API: eeprom_update_byte() while I use eeprom_write_byte(). These are routines provided by the Atmel library. Dunno maybe she researched it and found the "update" version had some advantage over "write".
Not sure if this is related or not. If it's a bug with the wear leveling method, then the temperature setting would have nothing to do with it.
Temperature disabled only will be memorized, even after unscrewing tailcap.
I’ve discovered another strange thing: Blinkie modes 2 and 3 are the same, both gives me all strobes…