That’s part of why I measured and posted the numbers. I wanted to find out what the curve looked like. And it’s actually multiple curves depending on whether you look at power or brightness. For example, “moon” was 2.35% of the maximum PWM value, 0.144% of the maximum power, and 0.041% of the maximum lumens.
I assume part of the power is lost due to overhead of running the MCU, part is lost as heat in various components (extra voltage burning off in the 7135 chips, etc), part is lost during ramp-up and ramp-down it the PWM cycle, and so on. And these are all pretty significant factors at a moon level. The MCU and ramping get less important at higher levels, but then we run into other issues like thermal sag and the emitter simply being less efficient at maximum power.
Each level has different ratios between PWM, current, and lumens… and the ratios change as the battery discharges or the light heats up… so, it’s definitely not simple. The data here is at least 4-dimensional, if not more. But calling it a single value like “2%” or “25%” or whatever is generally close enough for practical purposes, so I find it to be an acceptable simplification.
I saw similar results - within 1-2% for sony vtc4 vtc5 samsung 25r and efest 35a (i do lots of testing before publishing a result)… PF, PD, and BD were slightly lower (5-10%)
I didnt bother with regular protected panasonic 3400
We don't care about ever using 25A, all we care about is resistance and loss of voltage over the discharge time. We do want high amp capability, but 15A is about the most extreme. Just so happens higher rated amp capacity cells almost always mean lower resistance. Others may care about extremely high amps of 25 to 60, but we don't -- we want a cell with near 0 resistance. There is a great deal of problems getting an industry wide spec on resistance, so, amps is as close as we can get. Of course HKJ is the ultimate source for resistance, because he tests the cells consistently, so the comparative #'s, all from HKJ, do have a real meaning. Unfortunately, he doesn't always get the cells we are most interested in - he gets many cells, some totally useless, from vendors and manufacturers that want an independent test performed. Also, he tests cells that are simply re-labeled, re-branded ones, then doesn't always get the detail info we need of the underlying true cell.
It seems like the better cells we want (low resistance) are all rated 20A and above. Sometimes a 20A cell is better than a 30A rated cell - there's no magic formula. I'm looking more at 50A and 60A rated cells because I'm hoping to find something that beats a factory fresh Samsung 20R with higher capacity. Now, of course, the sells are diluting the waters, claiming a cell can do 60A, but in reality it does 60A cyclically, not constant, so once again, we are getting screwed from getting real specs.
I wonder if we will see a high current capable high voltage cell… a 4.3 or 4.35v cell with low internal resistance… the XML2 and XPL would benefit from the higher voltage
Using Samsung IRC 26C cell I’m trying to check calibration of new IS sphere and new LX1330B meter. I’m seeing around 2490 lumen with cells fresh off the grill.
Anyone measure theirs with standard cells?
BTW, I’m using Foot Candle readout x 10.76 then divide by 10 to match my only ANSI light of 172 lumens. Math, sphere, light meter may be all over the map.
On mine, I’m just using a simple linear correction to convert lightbox lux readings into lumens. I picked the correction factor which most closely matched the modes on my ZL SC52 (which seems to have extremely good self-consistency, but the entire scale could be skewed).
If I ever find time, I hope to build an actual sphere instead of a lightbox, and calibrate it according to lots of data points like what selfbuilt did. But what I have now seems to at least get within ~5% of most of the other numbers people throw around on BLF (with the exception of EagleTac products), so it seems good enough for now.
ToyKeeper, thanks for the link. One of the conversions given to me was in error so I have give all another go. One thing I liked about the sphere is moving the light around barely changed the meter reading but have not tried p-traps or cartons for same effect. BTW, have you used sunlight for a dirty meter calibration? Check here, also same info at Wikipedia. Mine read 2,400 (FC) high taken at 11 am in New Orleans.
Dale, heck yeah! I have been bugging a few for a BLF calibrated light source. Tom E had a good idea, just buy more lights at a known value (grin)
Anyway, at least this is good for relative measurements and easier to see light sag in first 30 seconds or so while I try to get it dialed in.
Actually, the key is have one guy do all the crazy ANSI light tests, dupe the lightbox, then distribute 1-2 known consistent lights around to cal up the unit to your own light meter and light box unit.
Can’t you calibrate off a quality house-hold fluorescent/incan bulb? That way you will have no battery sag, no voltage/amp drop, and the power to the bulb will be consistent for a long time. I would guess that a quality bulb would have very little difference in output across batches. Get an accurate sphere to calibrate off a bunch of bulbs and then send those out.
A bulb would only provide one data point, especially if plugged into a typical on/off-only switch. Chances are, the calibration is more of a curve which needs a lot of different data points before a good fit can be calculated.
Household voltage varies considerably from location to location and even depending on time of day and grid load. My night time voltage after most people go to bed in spring or fall with minimal AC and heating use in the area can be as high as 130 volts while nominal is 115V to 120V in most areas. My house drop is directly off of a pole transformer with only about 40 feet of line to my fuse panel. Motors love it but 120V incandescent lights have shortened lives.
Unless you have a AC incandescent on a variac and an accurate AC voltmeter there will be too much variation in output I would expect.
Does a sphere need to be calibrated for different color temperatures due to the particular lining of some spheres having slightly different reflectivity for different color temperatures? Seems to me I have read that some white paints/pigments have very poor reflectivity in the infrared.
Color temperature is most likely an issue. I can’t even get my lux meter to read values from my wide-spectrum modded SRK. It just displays an error.
OTOH, it reads a red XP-E2 just fine.
In any case, I hear that teflon tape has good wide-spectrum reflectivity. The publiclab spectrometer wiki suggests bouncing light off of that if the light itself is too bright for a spectrometer. It’d be rather a pain to coat the inside of a sphere with it though, considering its non-stick nature. Perhaps bleached white paper would be a better idea. At least, it sure does respond to UV. Or simply a nice white paint of some sort. Not sure what kind works best though.
I calibrated my sphere using around 10 ANSI rated lights… nitecore, olight, thrunite, zebralight… I was surprised at how close a number of them where when I backed out a “correction factor”… I would say my sphere is accurate +/- 3% after calibration….
worst case scenario - I have tested a boatload of lights and they are relative to each other if nothing else.
The lights you test are relative to each other. You can fine tune lights in your possession with a degree of accuracy and know where you’re going with them.
Tom and I as well as Richard can all feel fairly confident that our numbers are very close, because the 3 of us have similar systems made by the same very nice people. This doesn’t mean that I can build a light and test it in my box, go to Toms and get an exact same reading, or at Richards either. Why? Because quite honestly it’s difficult to replicate that first reading with almost any light. The led get’s broke in, the cells get aged, I don’t know exactly why but when I build a light, test it, write down the numbers….later when it gets tested again it’s almost never the same.
So if recurrent tests aren’t the same in my own measuring system, how could I expect them to be consistent across the board with anyone else’s? The thing is to realize it’s a ballpark figure. It lets you see what the changes you’ve made to a light bring to the table and gives an idea where to go from there. And that’s really about all it’s good for. That and maintaining a light. Or keeping an eye on the characteristics of the cells inside it.
When I step outside to find that pesky coyote that’s yapping, I fire up a 600Kcd light and start looking for him…it’s only going to be 600Kcd in those first moments. It’s going to be dropping, I know this, but the eye doesn’t pay attention to that as you’re looking for the details within the exposed area itself. Looking for or tracking that coyote, not paying acute attention to exactly how bright the light is at 330 yds. So the actual use of a light is much different from all the number crunching and wall hunting. And, after all, what do we build them for anyway?
My wife is leaning into our addiction. I’ve given her a Solarforce P1D that she carries. She likes the Hi-Lo-SOS interface. It’s only making about 560 lumens and I’m wanting to mod it up, but she won’t let me. She likes it right where it is.
Walking out to her car Friday night there was another Hospital employee in the parking lot, 11 at night, with Pop-A-Lock trying to get into his car. They were using completely inadequate lights and the guys windows were tinted quite dark, so they were having a great deal of trouble seeing to get the car unlocked. My wife asked if they needed a light. “No thanks, we have one and another standing by” With that, she hit Hi on her P1D and fired up the inside of the car through the tint. They were amazed, relieved, impressed, somewhat astonished at this “girl” (she’s 5’ tall and looks like a kid) had such a light.
She giggles about it even now telling the story. Men! lol
.