Oh I get that. But over 30%? That’s a lot. I have an xhp50.3 hi light that’s putting out about what a test on here showed. But this threads test shows the sft-40 as being some kind of super emitter. I had no idea it produced so much output yet I’ve never seen a light clear 2000L with it
Do not take everone’s word for what light output they claim, you can take most lumen claims around with a grain of salt, 20/30% off is really common. That is because measuring light output (lumens) accurately is not trivial , it is hard, lots of parameters need controlled very tightly and people do not realise that.
With a simple home-made contraption and some flashlights with manufacturer-listed output people get ballpark numbers within 20/30%, several people on BLF have made that and it is really what most people, even hobbyists, are happy with. Some BLF-members I trust to be a bit more accurate because they have checked calibration light sources, they may be within 15% (i.e. me, and I know Marco from 1lumen has checked light sources too, and some others)
But to get within that last 5% accuracy is either very expensive (for many thousands of dollars you buy an official integrating sphere plus spectrometer and calibration light sources, and learn how to operate the setup correctly, BLF-member maukka did that) or if you brew a device yourself which is surprisingly difficult to get accurate, you need a very controlled setup (directionally insensitive light collecting geometry, spectrally neutral reflectivities, spectrally correct luxmeters, correctly calibrated light sources) that almost no-one manages to get really right (that official setup is expensive for a reason). People often think they did a good job but end up with 20% variation from the real output easily, because of a flawed setup/wrong operation/dubious calibration sources.
I also do not not think that Hank (Emisar/Noctigon) or Simon (Convoy) have very exact lumen measurents. Maybe Sofirn.
2 Thanks
I understand that some setups aren’t the most accurate. But even with ceiling bounce app and a paper lined box, I get repeatable results that match official numbers and I cross reference 1Lumen, zero air, and grizz numbers and they close enough. My numbers for a 519a match your test of that emitter after counting light loss from reflector and lens. Same with koef3s test of the xhp 50.3. The only thing that doesn’t match is your test of this led and all the lights that use it. Not saying your test is in any way wrong. It’s just showing significantly higher numbers than resistance losses and light losses can explain. Even fet driven lights seem to have a hard time breaking 2000L
Edit. I think it’s the high vf that’s causing the “problems”. Looks to be about 3.8v at high current. A single lithium ion is going to have a hard time doing that at 8+A. But you’re using a power supply. Ok so this guy really needs a beefy buck driver to really make it sing. Not complaining that 2000L isn’t a lot because that is a ton of output
Beefy buck driver + beefy battery + beefy or bypassed springs + beefy wires going to the LED PCB, to get that peak output. With a power supply you can always subtract the voltage drop across the wires and crank up the voltage. Best case scenario is to have a 2S buck driver, but that’s too much effort to get 300 more lumens out of this poor thing.
I do not check the voltage coming from the power supply during my led tests, the voltage is measured very close to the led itself, just 30mm of thick wire and a short piece of PCB trace between the voltage clamp and the led itself. The PS-voltage will be higher.
I believe that a well beefed-up flashlight (as said: bypassed springs, driver with thick traces, short thick led-wires, ) adds very little voltage loss between battery and led, but light loss is a different story: I have done some measurements in the past of added light losses inside a flashlight compared to bare leds and it is quite a lot, in the measured example it is 17% ( Light loss in a S2+ triple mod, added similar data for a reflector light in post#8, added data in post#20 ).
I agree that Luminus may have altered the voltage of their SFT40 leds in later batches. No promises, but I have bought some 3000K SFT40 leds from Simon last week and I may do a quick led test on one of those. For the voltage 3000K 95CRI or 6000K 70CRI should not matter.
Agreed. That’s only worth it for the SBT90.2. I am perfectly happy with an SFT 40 at a measured 1700 lm
Yeah, I don’t imagine the VF would matter that much between 3000k and 5-6500k. But like the SST 20 it could very well be a completely different LED. I’m waiting for a few to come in, so I can swap out a couple. It should make a pretty good LED.
I’m pushing the SFT-40 with a 12 amp driver I dropped into my DM 1.12 I have no way of testing lumen output but it’s much brighter now than with the 9 amp
2 Thanks
Do you have a way of measuring it? I would be interested to see some actual Output numbers With the same cell
2 Thanks
Yeah, like take 2 measurements with simple phone luxmeter, with the only variable being the amperage. There you can calculate percentual brightness difference.
3 Thanks
Or even a current reading. I’d be curious to know the actual current draw
1 Thank
I will try that out keep u posted
Would you know how to get a current draw using a multimeter I dont have a current meter
With the leads in the correct spot on your meter, you take one lead and hold it to the battery and touch the other lead to the bare aluminum part of the battery tube.
Many/most meters can only measure current up to 10A, so you may blow the fuse. Also, testing current with the leads isn’t accurate. The resistance is too high, as they’re usually quite small gauge wire. I’ve had some luck making my own leads out of a short length 10awg and high current banana plugs, but a clamp meter is still a better option
Thanks I’ll give it a shot…probably should just break down and buy the proper tool for the job
How did you like the SFT40 at 3000K?
I love it! I put one in a Sofirn SC18 (which is an amazing flashlight for very cheap btw, if these were available 6 years ago, it saved me a some modding) and it is wonderful (I think I measured 800 lumen of beautiful light for this flashlight).
I’m still thinking of doing a voltage/output test of the bare led, comparing to a cool SST40 (not SFT40, for a reason), waiting for some time and motivation.
1 Thank
Nice! Thanks. I’m thinking of picking this one up with the SFT40 at 3000K. I almost always prefer orange peel, regardless of purpose. Is there a compelling reason to stay with smooth here?
(Simon pairs it with a smooth reflector here but he’ll often grant requests.)
You can’t with a regular multimeter, its resistance is too high, itll drop too much voltage at that current. You could measure voltage drop across a shunt resistor and calculate current that way but it’ll still affect the measurement.