Is there any room in firmware to make it a toggle?
You guys sure are efficient! DTP went from 0 to 60 in less than a day.
As far as LVP goes, I believe it’s not safe to charge cell from 0V. It sure works for some people, but there was thread here about forcing 18650s to vent and reviving dead cells was one way to do it. Compromise could be some complicated keypress combo that will turn on the light after LVP kicks in. Something that can only be done on purpose and for those who understand the risks (they exist?) of killing cells.
I tried my EC2 off 2 near-dead Ni-MH cells (2,5v unloaded), it works on low without switching modes.
Thanks Tom E :+1:
Yes this sounds logical
When protection kicks in you read 0V but the li ion cell still is in safe condition to charge
An unprotected cell is not safe to use after being depleted too low
A moon from 2,5v toggled on/off (so extra setting “low voltage cut off / low voltage moon 0/1 default 0) for those using protected cells or are going to cave and prefer some light NOW rather then protect cells seems versatile
But of course it is up to Tom 
No it kind of got lost but the DTP ledboard with options was discussed with the addition of two sets of solder pads for two footprints. But in a thread like this long things can get lost too easy. Your pushing for thermisistors brought it back and TA stepped in to make it visually clear in record speed! Thanks for you both!
+1
Thanks for all the kinds words from everyone. Now it’s my turn. I have been on a LOT of forums over the years as everything from annoying noob to senior admin.
In all that time I got so disenchanted with the internet I basically stopped “internetting” lol.
I can without question say that BLF is the best and nicest forum I have ever had the pleasure of participating in. People here are genuinely nice and they don’t twist the meaning of every other post starting fights where there wasn’t one.
Before BLF I was convenience that people were incapable of getting along for more then a dozen posts on the internet.
I am glad to have been proven wrong.
What I mean by that is the width of the trace from the ground wire to the outside ring.
if I flip the LED pads the trace has to go between the LED’s entirely + clearance so it doesn’t short. This led to a smaller trace then I would like to see.
By flipping the LED pads I am able to run the trace right through the ground pins of the LED’s and I don’t have to clearance around them. This lead to a noticeably larger total trace size to the outer ground plane.
Enough to make a difference in the final light? Not sure. Enough to annoy me? Indeed.
Speaking of traces, I think we should also request at least a 2oz copper pour on both the MCPCB and the driver. When dealing with this much current there is basically no such thing as too much copper.
Now a few other things that I missed above, like was stated, the screws do not directly contact the MCPCB, in fact it should not touch anything except the bottom shelf and the plastic centering rights.
TomE the driver is looking good!
Far as low voltage goes, as long as it by defualt has a low voltage warning/shut off I do think an emergency override would be good as well.
My family has the hardest time understanding that you are not supposed to run a lithium flashlight down to nothing (partly because they are so bright even at low voltages they don’t think they are low). LVP has saved many a cell from them.
While I have had success bringing sub 1v 18650’s back from the dead there is also no question that they were hurt in the experience.
So something like a bunch of fast clicks or a really long press and hold after lvp kicks in to override it would be ok, just so long as it was hard enough that family members would not do it just because they are too lazy to swap batteries lol.
I don’t know at which level the low voltage shutoff is currently done, but there is a point where the mcu shuts down anyway due to low voltage. So we won’t be able to use the cells down to 0 volts.
About LVP, I think that below the protection level, the light should not remain on for long without human interaction. So either you have to hold the switch or do the override sequence again every few minutes, etc. The idea being that you shouldn’t be able to kill your batteries by forgetting the light. You’d have to actively work at it.
I think the idea is elegant: once LVP kicked in, the light only works with the switch continuously pressed in (like say a cheap laser pointer), and at very low output. annoying enough to not do that for hours.
I like the idea too, but it doesn’t solve if the button is pressed in a backpack or something. What about after LVP, a triple-click or a double-click-and hold buys you 5-10min of light before shutting off again?
This sounds reasonable to me. In an actual survival situation it would be verging on dangerous to have to hold the button in non-stop to have light since you would not be able to use both hands.
the sequence to activate the light after LVP does need to be stupid proof, something that you would try out of frustration so that anyone could figure it out if they needed it. a bunch of clicks, and a hold or something else imprecise.
First of all thank you to everyone for volunteering your time for this cool project.
I like the idea of thermal management rather than using a timed stepdown. I share Tom’s concern over calibration. AFAIK Thermistors are not known to be particularly uniform. I would like to propose a different possible solution: the DS1820 temperature sensor. This has a couple of advantages that I can see.
- It is a Digital device, therefore it does not require an Analog pin on the ATTiny.
- It is also pretty dang accurate (0.5 deg C) right outta the package with no need for calibration.
- It is available in a TO-92 package (standard small transistor) and should fit in the same space as the set of pads for the voltage divider currently planned for thermal guesstimating.
Downsides:
- Cost - DigiKey shows $3.00 in quantities of 100. This may be enough to kill the idea outright.
- Code space - I think Tom has said that he is starting to get ‘close’ to filling memory space. I don’t know how much space the OneWire library consumes to interface with this.
- Wiring the interface - The device needs Ground (no problem) Power* (not PWM’d) and a data line in order to sense the temp and communicate it back to the CPU. This means another couple of wires that need to go to the LED board. On the other hand, I think the Thermistor needs the center node of the voltage divider to come back to the CPU anyway, which I don’t think has been addressed in the above design.
*It is possible to run the DS18B20 in ‘parasitic mode’ where it steals power from the data line.
Thoughts?
Since overiding LVP is desired for emergency situations, I don’t believe this solution is reasonable. If you need your hands free (which often is required) you have to put down your light. In case of total darkness you perhaps won’t find it again if it shuts down meanwhile. Pressing the button permantly for low level light is also no solution, you don’t want to do it for hours (and the above mentioned problem remains).
I would rather suggest a configuration toggle for this where the light switches to a low level when voltage goes down and blanks out for some milliseconds maybe every 10 seconds or so to remind that it is in emergency mode.
I don’t know much about firmware, but I think a ‘toggle’ would probably be a lot simpler to code as well.
At 80%, so got 20% of 8K to spare right now.
Just read a bunch of previous posts - yikes! Lot to consider... But I do like the idea of basically, when LVP is in effect, allowing a dbl-click, or something, to keep the light in some lo state for a few mins at a time, forcing you to re-activate it periodically -- sounds like a great and unique feature. I would say some level above moon would be good - a PWM on the 7135 only of course. Maybe a PWM level between 5 and 10, where 5 is a fairly bright moon mode. Think a good value to use would have to be determined based on the cells level - PWM value of 5 may turn out to be too low with cells at 2.0-2.5v, for example.
So what do you do think suits Narsil best for the LVP Tom?
That’s a legitimate concern but at the other end of the spectrum, anything which requires configuration leaves you at a risk of not being able to use it if you don’t have it preconfigured and don’t remember how to configure it.
Oopsie, k, instead of turning off dead cold once it times out in 5 mins (or so), could periodically blink (locator type of function) for 30 secs or so...
So from an LVP shutdown state, dbl-click to activate in a low mode for 5 mins, after 5 mins blinks every 2 secs for 30 secs, then goes out. dbl-click repeats this.
Should definitely have room for this, hopefully temp. monitoring too? 