Integrating sphere multipliers and reference lights

nkresho · 2018-05-09T14:34:05.000Z

Good Morning Everyone

I’ve been having tons of fun searching up and down to sort this concept out…

On my third sphere design…

First two were PVC traps, and very similar to one another. One was BaSO4 coated and the other was not. I did not get consistent numbers as the lumens climbed. I suspect that calibrating under 1000 lumens, and using a multiplier derived there, loses it’s accuracy when the light output is increased substantially. For example, I calibrated using three brand-name lights. An olight M2R warrior, olight S10r baton iii, and a nitecore p12. All going off manufacturer’s ratings, at 30s timed, and figuring a multiplier with that. Using that same multiplier on a higher powered light seems to give me substantially lower numbers than expected. By higher powered, I am referencing MT09R, L6 XHP70.2 FET, D4, Q8, and the likes.

I’ve even calibrated using diffusing sheets, smiliar to TA’s sphere design, and removed the multi alltogether, reading lux equal to expected lumens. Still reads lower than expected with higher power levels.

At this point, i’ve given up on my plumbing trap designs…

This third sphere, I went all out and trying to make it as consistent as possible. I am wondering if it is reasonable to expect a single multiplier to be applicable to all lights and all power levels (ML, L, M, H, Turbo, etc.).

3rd sphere build:
17” bouncy ball from walmart
paper mache about 10 layers of 1-2” wide strips of newsprint
1 outer layer of fiberglass fabric, cut into strips similar to the paper mache (using fiberglass resin for hardness)
cut the ball in half
added 4 sets of tabs on the outside to align all 4 sides when reassembling
added 4 L brackets to clamp the ball back together and secure with bolts and wingnuts
two ports 90 degrees off (one for input and one for output)
painted internally with 3-4 layers of untinted white paint, then 3 layers of 50% BaSO4 with the white paint
all centering rings also painted with BaSO4

So when I use my meter (Dr. meter) and read the 30s measurements from my reference lights, I get a large range of multipliers for each light. Somewhere in the range of 0.26 to 0.17. I’ll post my detailed numbers when i get home tonight. Typing this on a break at work at the moment.

I’ll snap pics of the new setup as well for critique, and post them here.

My main question is, can I use one multiplier to measure all lights? High and low outputs? Ranging from moonlight mode to 20k+ lumens? Is this a reasonble expectation? Do i need to derive a log scale to vary the multi based on the lux value?

As a side note, I haven’t calculated the port to internal sphere surface ratio. I am guessing this might have a little to do with the variations too, as the lens diameter of the mt09r is significantly larger than the p12 and olights. I haven’t really delved into what this should do to my multi. In that case, i’m getting 907 square inches internal (from 8.5” inner radius), minus the ports. Gotta calculate the port diameters tonight too.

Preemptive thanks for the ideas and discussion.

djozz · 2018-05-09T19:58:08.000Z

If everything is perfect, a single multiplier should work for any light source.

But there is a lot of stuff that makes it less perfect, here’s some:
-as you already mentioned, different flashlight shapes create different reflectivity inside the sphere, if uncorrected that will give an error. Your sphere is pretty large so this error is not huge.
-your luxmeter will read different for different colour temperatures and CRI. This error can easily be 10% or more for a cheap luxmeter like the Dr. Meter (LX1330B). A non-perfect coating can also add to these ‘spectrum’ errors.
-do not trust the specified output numbers of flashlights even of major brands, errors can be easily 10%.

About power levels, I have not seen clues yet that different power levels need a different multiplier, as far as I noticed even cheap multimeters have a pretty lineair response: if the light level increases 10 times they also read 10 times.

One thing that you do not mention and is quite critical for a sphere to function well, is a baffle between the input and output port: an obstacle inside the sphere that makes sure that no light from the entrance port can fall directly on the exit port. It is quite essential to make the sphere integrate the light. I hope you have one, but in case you don’t, here is how they look in one of my spheres (there are three baffles in the picture because there is an extra exit/entrance port in this sphere, you would need only one)

nkresho · 2018-05-10T00:16:37.000Z

Thanks for the input. Yes, I added baffles between the input and output, as well as the direct reflection path between the area immediately in front of the input port and the output. Two ping pong balls, also painted in BaSO4.

I’ll snap a few pics shortly.

Glad to get confirmation that a single multi is possible. This is starting to get a little frustrating.

Coating definitely has some imperfections. It’s not sanded yet, and I plan to do so after a few more coats of the BaSO4. It’s definitely coated thoroughly though, just a bit bumpy.

Nearly all the lights I own are neutral white variants. 5000k to 5700 or so. From the olights to my self - built ones. The nightcore is cooler though. Might be a factor with the variance there.

edit: here are a few pics:

New sphere

nkresho · 2018-05-11T15:29:21.000Z

How about another idea.

Using an off-the-shelf, residential, AC led lightbulb of a given lumen output? Maybe even a few color temps to check.

I have a 1500lm sylvania bulb (warm white) I tried for s&g. Using the 0.20 multiplier (I was estimating to be correct) yielded exactly 7500 lux… or 1500 lumens. Matching the label’s claim on the bulb. It stayed steady too, rather than declining like flashlights tend to. I measured for at least 2 mins and the number didn’t move.

I might pick up a few more new led bulbs this weekend to test further.

If they are relatively accurate to spec, residential LED bulbs might be a good reference standard for calibration. I’m not sure what type of degradation we’d see over time.

I’ll report back after additional tinkering.

nkresho · 2018-05-13T01:51:30.000Z

I picked up a couple of led bulbs to try out today.

Here’s what I got:

GE Nighthawk 3000k rated at 3000 lumens
spotlight type, so I was able to rest it on my base and measure through the Mt09r sized opening
15600 lux at start w/ 0.20 multi=3120lumens
15400 lux at 30s w/ 0.20 multi=3080lumens
15300 lux at 3mins w/ 0.20 multi=3060lumens
14900 lux at 10mins w/ 0.20 multi=2980lumens

pretty close to spot on with the 0.20 multiplier
right on at 30s if I change the multi to 0.1948

GE refresh 5000k rated at 1600 lumens
regular bulb type, so I put it through a Q8 opening and pulled it upwards, allowing all the reflector to sit inside the sphere. This may have skewed the reading as there might have been a direct path past my pingpong ball baffles
10320 lux at start w/ 0.20 multi=2064lumens
10280 lux at 30s w/ 0.20 multi=2056lumens
10180 lux at 3mins w/ 0.20 multi=2036lumens
9700 lux at 10mins w/ 0.20 multi=1940lumens

a bit higher than I anticipated, based on rating…

using a bulb I had lying around (has hours on it, used from time to time):
slightly smaller bulb type, I put it inside the q8 opening and inserted it until the reflector was totally inside the sphere, but no further.
Sylvania 2700k rated at 1500 lumens
8350 lux at start w/ 0.20 multi=1670 lumens
8250 lux at 30s w/ 0.20 multi=1650 lumens
8160 lux at 3mins w/ 0.20 multi=1632 lumens
7950 lux at 10mins w/ 0.20 multi=1590 lumens

30s reading using the 0.1948 multi gives 1607lumens still - a bit brighter than I expected too

So my conclusion is that my sphere is reading high with the regular lightbulb types, as they likely send direct light to the sensor. With the nighthawk, it’s a spotlight type, with an aluminum heatsink to rest on the ring. It reads as a flashlight would.

If I take those readings as correct, my new multi is 0.1948 and the 30s reading from the nighthawk AC spotlight bulb is spot on.

Reference light testing for integrating sphere

The nighhawk looks a lot like a single emitter COB through the reflector.

chops728 · 2018-05-13T02:26:37.000Z

I’m curious if the Bulb output changes when the Line Voltage changes——any idea what it was—-seems the only way to get consistent readings with AC powered bulbs would be a controlled power source

nkresho · 2018-05-13T02:40:37.000Z

I’ll check again after I find my kill a watt. Got it in a box somewhere.

ImA4Wheelr · 2018-05-13T02:45:17.000Z

chops728:

I’m curious if the Bulb output changes when the Line Voltage changes----any idea what it was—seems the only way to get consistent readings with AC powered bulbs would be a controlled power source

Just a guess, I would expect the normal voltage level range to not impact output. The few A/C led lights I have disassembled had buck drivers pulling down voltage to 90+ VDC. That gives a good amount of headroom.

nkresho:

I picked up a couple of led bulbs to try out today. Here’s what I got: GE Nighthawk 3000k rated at 3000 lumens spotlight type, so I was able to rest it on my base and measure through the Mt09r sized opening 15600 lux at start w/ 0.20 multi=3120lumens 15400 lux at 30s w/ 0.20 multi=3080lumens 15300 lux at 3mins w/ 0.20 multi=3060lumens 14900 lux at 10mins w/ 0.20 multi=2980lumens pretty close to spot on with the 0.20 multiplier right on at 30s if I change the multi to 0.1948 . . . .

Genius! Why didn't I think of using some rated flood lights for calibrating my integrating tube. Thank you for this post nkresho.

nkresho · 2018-05-13T11:13:48.000Z

Thank you. This night hawk is the way to go imo. 3000lm and really pretty steady as far as output. It does decline over time, bit not that much. Not super cheap at about $35, but readily available at lowes.

This will likely be my new reference standard if I don’t come up with anything better. It has a big heatsink and the flat front is as close as I can see to a flashlight. I can actually sit it on the sphere opening and the heatsink supports it on the sides. The regular lightbulb shaped ones have varying depths they sit inside the sphere and that changes the reading on the meter.

I might grab a mid grade meter at some point too. Something that will connect to a computer and log output over time. It’s kinda a juggling act with the current set-up and a stop watch.

Still haven’t found the killowatt, but I’ll post my measurements when I do.

nkresho · 2018-05-13T11:22:30.000Z

I updated the thread title for easier searching and more accuracy.

djozz · 2018-05-13T12:20:00.000Z

My references are 2 led-flashlights that I found have very steady output. So ‘my’ lumen is constant over time but not guaranteed precisely correct. I do not look for an absolute lumen reference because either the specs are just as reliable as those of a flashlight (not very), or it is really expensive (a documented calibration test source) and even then the accuracy is spoilt by the luxmeter’s spectral error, coating or varying reflectivity of the sphere caused by the light source.

The last cause of error btw can be decreased greatly by positioning a permanent auxiliary constant light source at the third 90-degrees position in the sphere (and add extra baffles to block the light path to the two other ports) . Use the luxmeter reading of that light with the to-be-measured lightsource in measuring position while switched off to correct the multiplier for the altered reflectivity (via the test link in my signature-line there are some examples how I made lights like that).

nkresho · 2018-05-13T12:48:44.000Z

Thanks. None of my lights are all that steady for output. Maybe the olights… They are very repeatable. I’m just getting the feeling that the ac flood light is a bit closer to the rated specs. Plus the power source is fixed, unlike batteries. I don’t have a higher end bench supply yet.

I’ve toyed with the idea of a third port, for threat very purpose. Still not sure if I want to go that route.

Thank you for your input. You’ve got way more experience in the spheres than I do. I have lots of reading to do. Thanks for documenting your tests so well.

I would love to send the floodlight to a place with a calibrated sphere for measurements. I’m sure some college students would not mind a little testing and comparison with my cheapo sphere and their real one.

Once my mind is at ease that I’m as close as possible, I’ll retest all my lights and have a solid spreadsheet to share. Then I can recalibrate my plumbing trap spheres to the constant.

This all started with my trap versions reading pretty low compared to others’. If I had been close I would have left it alone.