Sorry, earlier post had more details. Yes - reused same exact parts. Didn't track the LED to the side, but this board always had one brighter than the other. It's pretty well known at this point that LED to LED variations will result in uneven brightness with them wired in parallel like this. You need one resistor per LED to get more even results.
As far as root cause, I cant say for certain -- may be some issue on the PCB that was dependent on the reflow/solder work, or could have been some contaminant that was somewhat conductive under low heat or movement, etc., or might be the quality of the reflow. Dunno for sure the exact cause, but I'd say it wasn't the parts and don't think it was the wiring or driver side, so in this area of the switch PCB and solder.
I just touched the tip of my iron the both ends of both LEDs and behold: there was light again! 30 minutes later, no more light. Really weird. The voltage across the LEDs is also lower now than before the fix.
Edit: Took some measurements. When the LEDs work, Vf is 2.23V. 30 minutes later, it is back down to 1.5-1.6V and there is no more light. It’s like there’s a bridge which slowly rebuilds itself while the LEDs are on. I’m leaning towards “faulty” LEDs.
Yes - I had the same result, works for a bit then fails again. When they are in a failed state, check for a short from grnd to V+ coming in to the resistor - I swore the DMM indicated a short, repeatable.
So the voltage from the MCU is a steady 2.2v, but somewhere a slight short develops around the switch area which causes the voltage to drop down to 1.6v?
That’s interesting.
No, there’s about 4V coming to the switch assembly (0.2V drop from the diode is about right). The voltage across the 15K resistor is 4V minus the LED Vf. I measured as high as 2.4V on the resistor when there is no light, and 1.7V when there is light.
That means higher current when there is no light. Which makes sense if the current is going around the LED somehow. Next thing I’ll try is removing or moving one of the LEDs which seems awfully close to the switch.
I don’t know very much about the chemistry involved in making an LED, but it seems like these little switch LEDs are possibly acting funny when they’re not fed enough current. Maybe they start to leak current past their internal parts. IDK. I’ll leave the technical stuff to others. Lol
I think it is (first production Narsil), the Q8 has the latest version of Narsil and Tom worked really hard getting it done and tested.
I have an earlier variant that I flashed and put in a little FandyFire Rook that uses 4 clicks to lock out… the Emisar D4 and D1 take 6 clicks to lock out! 4 clicks to the D4/D1 (thinking I was locking it out) puts it in a momentary switch tactical mode. I’m presuming the Q8 will also have the 6 click lockout? Or will it be 4 like the earlier versions?
Edit: One thing that confused me was that when I clicked the D4/D1 6 times for lockout (after doing some reading/research) the lights blink 4 times to indicate lockout. Unlocking they blink twice. Seems somehow not intuitive for the number of blinks to not match the required clicks. But now I know.
I'm all confused which thread I'm in, but when the LED's failed to light up. I detected a short from grnd to V+ in to the board, to the resistor. Verified it several times. I think the full parts removal and reflow fixed it all up. The light has been fine now for 27 hours and still running.
OTSM is a nice evolution of the clicky power switch, but I doubt we can write off e-switches yet. For example, a reverse clicky power switch, even with OTSM, cannot keep the LED (s) lighted while the button is held. It’s fantastic at “click to turn on or off”, but it cannot do the other half of the Q8’s core interface — “hold to ramp”. The single most defining feature of the Q8’s UI isn’t really possible with a power disconnect switch.
The switch types can try to emulate each other, but I doubt they’ll ever by fully interchangeable. I implemented ramping on a clicky switch, but it’s not the same.
It’s probably worth mentioning that the Q8 doesn’t use a triple-channel driver.
OR it means that there is additional high resistance adding on in series with the fixed resistor. Voltage across resistor, plus voltage across LED must add up to about battery voltage, less a little bit (i.e. the voltage of the MCU pin, which I anticipate will be roughly battery voltage minus about 0.3V.)
If Vled+Vr doesn’t add up to the full amount, additional voltage drop must be occurring elsewhere (poor joint, internal PCB problem etc.)
And for one of the most important thing. The Q8 draws over 20A in turbo. A clicky switch does not like high currents like that. With E-switch you can easily turn on and off 30A current because the electric components of the driver do the switching, but a clicky will melt down.
In Narsil’s ramping UI, 4 clicks from off activates lockout.
In Narsil’s mode set UI, lockout is mapped to 2 short clicks and a long click.
The Emisar 6-click thing (and 4/2 blink thing) seemed strange to me too, but I didn’t change it… generally tried to minimize changes, especially ones which didn’t match the project’s specifications. However, I’ll send you some firmware showing how I prefer to do it instead.
Although most shorts change a connection from near-infinite resistance to infinitesimal resistance, it sounds like this may be more of a middle ground.
Even a partial short would cause problems on such a sensitive circuit; the smallest trace of conductive material in the wrong place could alter the voltage enough to interfere with the switch LEDs. For example, even an oily fingerprint on the PCB could potentially decrease resistance enough to cause intermittent misbehavior on a particularly sensitive circuit. Or if there was metal dust in the air during assembly, some could stick to the PCB in the wrong places.
