How accurate is a ceiling bounce test to measure difference in output?

I’m finishing my build of a 2D Maglite with 4x XM-L2 in direct drive… I decided to test it against my brightest LED light, the Nitecore TM26 (rated 3800 lumens).

I have the inexpensive LX1330B light meter, and did the test by placing both lights and the meter in the ground under a white ceiling about 2.8 meters high…

These are the numbers I got:
Mag: 283 Lux
TM26: 197 Lux

The maglite produces roughly 50% more lux, but does it mean 50% more output?

Points to consider:
-The maglite has 20mm medium clarclo optics and the beam is very floody, it produces a large smooth hotspot while the TM26 beam is defined. Could the floody beam mean that less light is being bounced back?
-At 1M the throw is very similar when both have fresh cells.
-The mag draws 7.6A from 2x KK 26650, the TM26 10.5A from a single Samsung 25R (It could be more but my cheap meter is rated 10A).

I’m asking because even though the meter says so, visually I cannot tell that the mag is considerably brighter, but could this be just a case of “smaller, more defined hotspot leads to believe there is more output”?

I think a ceiling bounce can be a good way to gauge output. However you’d need several results from known outputs to offer any sort of calibration and to check correlation. I suspect however it is most useful, when comparing output modes on a single light, rather than light to light.

The reason I say this is, beam profile seems to have a large affect on the readings. Floody lights seem to read higher when I tried this. For example, my Maglite MagCharger went from 251 lux down to 45 lux by simply going from a defocused flood beam to a tight spot.

Now while I’m sure lumen output will be affected by focus position, I don’t think it’s as high as this ceiling bounce test suggests. It also suggests that floody beams/larger hot spots will read higher than smaller beams and more tightly focused lights.

Will, one thing in particular I try to do is to keep the size of both beams (spill) approximately the same size. When light bounces off walls as well as ceiling, lux readings increase.

I can be useful to guage relative output… but the light meter needs to be catching only indirect light… if its focused on the hotspot on the ceiling the results wont be useful.

I have done many ceiling bounce measurements and it works reasonable ok (like within 20% or so) in output comparisons. There's a few things to consider that will lead to errors.

1) reflection from a ceiling is never perfect 'lambertian', however matt it is there's some direct reflection as well. So with your luxmeter right next to your light source, pointing straight up makes the luxmeter receive that direct reflection and you thus overestimate the amount of light compared to if you point the light/luxmeter to the ceiling at an angle.

2) not only ceiling reflections, also most luxmeters are angle-sensitive, the same amount of light coming from an angle reads lower. This results in that floody lights are underestimated compared to throwy lights because a larger portion of the light enters the luxmeter at an angle.

3) another reason for under-estimating the light from an angle is that the edge of a beam projected on the ceiling comes from further away compared to the center of the beam

like jmpaul says, indrirect ceiling bounce readings suffer from less errors (the 'integrating bathroom' method :-) ), but is still notperfect, there's a reason for using a sphere instead of a bathroom.

I wanted to make a standard set-up pointing the light and luxmeter to a corner of the ceiling, where three white surfaces meet, reducing the errors of 'straight up' ceiling bounce measurment, but instead I have build an integrating sphere. (In fact I am even building a second -better- sphere at the moment).

Thanks everyone, I will try the integrating bathroom method.

I can’t remember who uses it, but it was a set of several PVC angles with light source and meter on each end, I could build something similar to that…

Several people who sound serious about measuring lumens use that PVC-pipe method, and they claim it is not much beam profile dependant, but by design the integrating properties should be seriously worse than a sphere. But compared to a bathroom it is more convenient probably...

So I guess my "integrating bathroom" measurements in my diffusion testing thread weren't very reliable (my suspicion due to the low reading of the DC Fix). I'm one that has accumulated parts to build an integrating PVC pipe (I just bought (3) 3" 90degree elbows, an end cap, and some thick foam.

-Garry

integrating bathroom measurements and also pvc-pipe measurents are fine, it is just not dead accurate. And dead accurate light output measuring is in practice not important, it is just a frustrating holy grail for nerds (like me :-( )

How i should place the sensor and flashlight in the room? Its still a bit unclear to me, i have just a room but i dont need dead accurate data. Like how much space between flashlight and ceiling and where i should place the luxmetersensor, for example 1m away from it and 1m beneath it.

I’ve swapped the led’s in my S2+ triple from xp-g2 s3 3d to xpl v6 3d and i did some basic “ceiling” bounce test. I placed the light metter and the flashlights i tested under my desk and tried to get an average reading from known sources, and since i don’t have much lights that are ANSI/NEMA tested i used the Nitecore MH20 and Olight S1 Baton, both on their highest setting, 1000 and 500 lumens respectively.

So after some math i came up with a 3.31 lumens per 1 lux on the lux metter. Next i fired up my S2+ on turbo and the light metter was showing 1080lux at startup and about 950lux after 30 seconds. Those numbers are very good, very unreal to me, it might turn out that my s2+ is still making over 3k lumens after 30 seconds on turbo at which point it is very hot.

My question is, are those numbers pure bulls*it or they might be considered as fairly accurate. Not trying to achieve lab. accuracy but i just wanna know if my S2+ is pushing the magical 3k lumens or not :slight_smile:

Quick test with the DMM says ~11A at the tail with 25R @ 4.19V.

How accurate is a ceiling bounce test to measure difference in output?

IMHO can be pretty accurate.

If you start with a good setup (room, position of the flashlight) you can even do without a luxmeter to see if there are major differences.
You can add a camera with fixed settings pointed on a very far detail of the room. This will help a bit.
A luxmeter is recommended if you want numbers and digits about a light.

I agree that it can be pretty accurate. My strategy is to point the luxmeter up and have as much of the beam on the ceiling right above above the luxmeter. This is opposite to some of the advice above. My reasoning is: I don’t expect the room to perform the function of an integrating sphere; there are too many absorbing things so the “sphere multiplier” most likely won’t be close to the 10-25 recommended for accurate integration.

Instead, if I concentrate the light from the flashlight on a small spot on the ceiling above the luxmeter, and assume the ceiling is a lambertian reflector, then there should be some reflected lux at the location of the luxmeter that is proportional to the total lumens hitting the ceiling. It is important that the spot be small and in a consistent spot relative (to keep the apparent area of the spot the same) to the luxmeter. So if one is not careful to keep the spot size small and consistent, one will get smaller readings for a flood light with a wide beam because light at the edge of the beam is not directly above the luxmeter and less light from this area will reach the luxmeter (because the apparent area is less). To keep things more consistent for a floodier light, I move the light closer to the ceiling, so the spot becomes smaller.

I did not consider the possibility of a significantly non-Lambertian reflecting ceiling, as was mentioned by djozz earlier. I tested this with my ceiling and a couple lights. While keeping the spot directly above the luxmeter, I found that when the angle would cause a specular reflection towards the luxmeter, the reading was higher than otherwise, but only by approximately 5%. Of course, this will depend on one’s particular ceiling surface.

EasyB: according to the way you do it, there should be a difference in lux measured for the same ouput but with different pattern beam.
a thrower light will have more light in the spot and the reflection of the spot can increase the reading.

For wider beam lights I move the light closer to the ceiling to keep the spot size small.

In fact, this strategy could be adapted to actually calculate the flashlight output in lumens, not just a number proportional to the output. To do this the ceiling should be a lambertian reflector, and the surfaces other than the ceiling should be black (non-reflecting) so the only light getting to the luxmeter should be the diffusely scattered light from the spot on the ceiling. The spot should be small; if all the light from the flashlight is in a spot of 20 degrees (full angle), the error should be within a couple percent.

Here is the method: because the ceiling is Lambertian, the reflected light is distributed in angle in a known pattern. If one integrates this distribution over the half-hemisphere, one gets the total light (lumens) reflected by the ceiling. But the ceiling is not a perfect reflector; it will absorb some light. Doing this integration, we get:

(reflectivity) * (flashlight output) = I_0 * pi

The reflectivity is the fraction of light that is reflected. A white ceiling might be 0.8 or so. pi is 3.142. I_0 is the luminous intensity (candela) at normal incidence, where the luxmeter is located. Then one can calculate the lux at the luxmeter if one knows the distance between the ceiling and luxmeter.

(flashlight output in lumens) = (measured lux) * (height^2) * pi/(reflectivity)

For example, in my setup, the luxmeter is located 2.27m below the ceiling. I measured a triple dedomed XPL V6 1A with Cute 3 optics and got 265 lux on the luxmeter. Assuming a ceiling reflectivity of 0.8, the calculated flashlight output is 5359 lumens. The output is overestimated because my walls are not black.

Probably the easiest way to use this method is to do it outside on a dark night. This would get rid of reflections from walls. Suspend a piece of white styrofoam board horizontally ~2m above the ground using a tripod and put the luxmeter on the ground, maybe on a dark sheet. Then point the light up at the styrofoam so the whole beam is on the board.

An alternative way to get rid of the secondary reflections from the walls is to subtract them out. I just tried this with my supfire L6 with 3V XHP70 with a mostly full 26650 (13.7A with full battery). I taped a piece of white styrofoam board to the ceiling. I got a lux reading of 290lux. Then I held a sheet of cardboard above the luxmeter (~1m up) to block the light coming directly from the ceiling spot. The remaining light is just from secondary reflections from walls. This reading was ~85lux. So 290lux-85lux is 205 lux just from the ceiling spot. Assuming the styrofoam has a reflectivity of 0.95, the calculated flashlight output is 3491 lumens, which seems close to a realistic value.