I’m not sure if GreenCamper and I both have DVMs that measure high, or if there is some sort of change between the prototypes and the production units in terms of current draw. The video from Lexel below shows the early phenomenon with the tint ramping. In this case, the current during the “mixed” tints is lower than either end.
The measurements in this thread approach that of Lexel’s, but with only 5 of the 7 channels enabled. Are the 7135s wacky? or is there some other loss happening? 5x0.35A=1.75A, not 2.2 or 2.4 like we are measuring now.
I agree with atobe, the resistance would be more like 1.2 ohms or less, not 1.2 megaohm
Mystery solved. Its the tint ramping. with factory default, current at max level is 1.77A on my meter on either end of tint ramping. So with either just 4 WW leds, or 4 NW leds, I get 1.77A, which is what I would expect for 5 7135 chips each drawing their max current. But at the default mixing, I get 2.1A (this go around). I’m guessing there are small peaks and valleys of current draw throughout the tint mixing scheme.
Moral of the story, if you want max runtime, choose full WW or NW, else just enjoy your lantern.
As far as run time tests go, one also needs to specify the setting of the tint, which realistically can only be all WW, all NW, or default tint mixing.
That may be a part of it DBSAR, but it does appear that if tint mixing is in the middle of the range, that the current increases. In these cases none of the channels is at 100%, but some hybrid to keep the brightness constant. I recall seeing graphs that ToyKeeper posted some time ago regarding plotting total current versus tint mix, and it seemed reasonably flat. Not sure if running at the max level causes a peak for some reason, or ???.
Glad that is cleared up! My initial runtime test wasn’t with default ramping so I’ll reset the firmware and redo the test again. I’ll update the OP once I have the results.
I noticed that too, as there was less amp draw at the far-end of each of the tints then in the middle range, (3700K ~ 4200K) when all eight LEDs were running.
My runtime test 1 (no graphs, sorry) was 4 hours and 38 minutes with 4 recycle cells pulled from the recycle bin at work that measured between 2950 and 3050 mAhr Zanflare C4 charger. My charger is a bit optimistic for capacity measurements, or so it has been reported. I stopped the test when the lantern made a noticeable step down. I was starting to think they light was dimmer, then within a minute or so the light stepped down to about half, similar to the graph in the OP. I measured the cells all at nearly 3.0V when I stopped the test. Like GreenCamper, I didn’t want to run the cells down any lower, at least for this test. I can’t be certain what the tint was when I started the test, but I think it was the default. 2 hours into the test I paused it to make the observations noted above regarding tint mixing and current. Then I continued the test at the factory default tint mix.
I’m charging the cells back up in the lantern overnight, and will run a full NW tint run test tomorrow.
Nice measurement, thanks. But I don’t agree with the conclusion, at least before I see output measurements (even intensity would do).
The reason is that higher current likely means higher output. And if you want to prolong battery life - run it in the most efficient mode with just enough output to do what you need to do. If at level X tint-mixed variant is brighter than the end of the spectrum you should turn it down a bit to get a fair comparison of what is better for maximum battery life.
I suspect that the best might be cool simply because cool LEDs are more efficient. Though by tint mixing you’re doubling the number of LEDs and in effect halving current per emitter. This is good for efficiency…
So actually my guess would be that the most efficient for a stock LT1 would be either cool or (more likely) a farly-cool mix.
One could unsolder some 7135 from the warm side to make all mixes more coolish. I think that it would improve the peak efficiency of the lantern.
With my LT1 using a non-production firmware load (with SOS added, helping ToyKeeper test this for stability), 5 WW and 5 NW 7135 chips enabled, running the lantern at level 150/150, the longest runtimes appear to occur when the lantern is at either 100% NW or 100% WW in the tint mixing. I tested the firmware default above for approximately 3 of the nearly 5 hours of run time. For the remainder of the time I beleive it was also at the firmware default, but may have been just “somewhere close”, as I didn’t return to firmware default prior to the test.
Hows that?
BTW, I don’t believe my firmware version has anything different in the code regarding the way the channels are driven at the various tint mixing settings. I will repeat the 100% NW tint mixing later today, after I sleep more (hopefully) and my batteries finish charging. Unfortunately I have no way to measure light intensity.
Fyi, I’m using an Adafruit 4162 VEML7700 light sensor and Raspberry Pi Zero to measure the light output, along with a custom Python script to generate a csv file to import into Excel to create the runtime charts. Cost about $20 total and works very well :+1:
I tested my Fenix UC35 v2 flashlight with it on turbo and the runtime chart pretty much matches the one from Fenix.
I been getting 1.78 to 1.95 amps with 5 channels on the production model with 5 AMCs default, and with 7 i have been measuring between 2.2 and 2.46 amps with all 7 AMCs enabled, on the Sofirn 3000 mAh cells with the pedal-to-the-metal, (maximum) with three different meters i tested with. i noticed the amps change a little depending on the tint-range setting, and seems to draw a little less amps at the each end of the tint ramp range.
Thanks for the tip on the Adafruit 4162 VEML7700, I poked around a bit but need to do a bit of learning before I could do something with it. It sounds like a fun project, so if there’s a “how to” somewhere that would be super helpful. If you have recommendations PM me with them.
I finished my second run test. 100% NW tint. Unfortunately I missed the first step down, it was somewhere between 6 hours and 6h17m. I saw the second step down at 6h17m, and the third at 6h20m. The last step down during my test was at 6:46. I measured the cells after the step down at 6:17, and they were 2.93V. I measured them after I ended the test at 6h46m, and they were 2.89V. The firmware is trying to keep the light alive by stepping down the levels. I did notice the 4th step down was not to the lowest of the five levels in the stepped ramp.
Anyway, its rather obvious that mixed tints result in more lumens and shorter run time. 4h38m vs ~6h17m. My statement of more lumens is an assumption I can not prove, but I suppose Green Lantern could run this graph for a few minutes at one tint, then change the tint to near the middle and there should be some step. As a reference, as noted previously I was using 4 cells poached from the recycle bin that measured between 2950 and 3050 mAhr in my Zanflare C4.
Out of curiousity I decided to measure the current at 100% WW, 100% NW, and default tint at the five defalut stepped levels.
|Level|( WW )|( NW )|( Default )|
|1| (0.016) )| (0.016) )| (0.016 )|
|2| (0.158) )| (0.158) )| (0.178 )|
|3| (0.490 )| (0.492) )| (0.492 )|
|4| (1.01) )| (1.01) )| (1.27) |
|5| (1.80) )| (1.80) )| (2.23) |
For levels 1 and 3, the current is the same for each tint mix. But for levels 2, 4 and 5 the default tint mix current is higher. Used a button top 30Q cell for the measurements.
Yes TK adjusted the tint mixing code to get about a flat output curve after my test showing it dropped in the middle
Seems it is a bit too much with over 20% more current while original was about 25% too low
A slight warning about using ambient light sensors as light meters: usually their spectral response is utter cr*p (I could not even find the spectral response of this Adafruit device), so they are perfectly fine for runtime tests (because you use one light source that has the same light spectrum throughout the test, and all light meters have good linearity), but if they are used for comparing output of different light sources that vary in tint and CRI, large errors will occur.