Measuring Candela (Please Help)

Flashlight Modding and DIY Parts LED Testing
1

I need to know if my methodology is bad or if my equipment is? I am getting readings that are about 35% higher than what is published on reputable brands.

When I measure Candela I setup my light meter, move 10 meters (sometimes 15 with a really powerful light) away from the sensor and take the readings.

I can’t see the light meter from there so I setup a camera to record the readings. I make 5 slow passes across the light meter, from left to right in order to find the brightest part of the beam. It’s rarely in the dead center.

Then I go back and look at the footage frame by frame, and write down the 3 highest numbers. This produces consistent repeatable results. But they seem high.

Example:
Acebeam W20 Medium mode official 82,656 cd

My readings across multiple tests tonight
106 kcd
107 kcd
111 kcd
110 kcd
107 kcd
109 kcd

And so on, you get the idea.

Example 2: Lumintop thor 1. I could find no official rating, but 1 lumen measured it at 148 kcd . It should be noted his measurements for high mode are very close to the factory rating.

My Tests
201 kcd
201 kcd
200 kcd
197 kcd

So my question is, am I “cheating” By seeking out the brightest spot in the beam?

Or is a LX1330B really just 35% off?

Note: The reason I don’t use the peak hold function on this meter is because it doesn’t really do either of those things well. It does catch peak sometimes, but if I pass the light over the sensor more than once it drops the old reading and starts new.

2

Hmmm interesting situation. Before I finished reading, my initial thought was to ask for data on the full sweep, not just the peak measurements. This would tell us how much the peak reading differs from the “typical” hotspot value, and thus how variable/anomalous it is. But since you get fairly consistent peak readings across the 5 passes, it seems that random fluctuation is not the problem.

After I read everything, I checked that indeed, your measurements are around 35% higher than others’ measurements across the two lights, so maybe there is a systemic issue with the measurement or with the device. But I’m hesitant to jump to conclusions here because Acebeam’s published specs are not necessarily consistent with 1lumen’s measurements, so the +35% across two lights could be a fluke. In other words, you don’t yet have enough data to conclude that you are indeed over-measuring.

I would suggest measuring some better-known lights (like Convoy buck-driven throwers), for which there are plenty of independent, mutually consistent measurements by the community. This will give you a better sense of how your measurements compare to widely-accepted values.

Also: LEPs might be particularly finicky due to chromatic aberration, and it’s possible that there’s a very small region inside the hotspot that is much brighter than a “typical” point of hotspot. Measuring a reflector light can rule out this issue.

Additionally: I don’t know how other reviewers measure peak intensity, but your method seems very thorough, which means that you’re more likely to pick up higher peak readings that others miss. In other words: the running maximum only goes up, never down, the more points you measure.

1 Thank
3

I don’t publish numbers yet, but I use the max function of my meter (and its Bluetooth function) and move the spot a little bit around to find the highest intensity. Not too long, because the output tends to change significantly in the first minute and getting its peak and ANSI intensity is difficult.

Do you have reflections or ambient light? I’ve added a black tube to my meter to block stray light and only accept light from the front.

1 Thank
4

Don’t listen to me: I’m just repeating what members with more skills and experience have written.
It is very hard, if not impossible to compare measurements of yourself with those of others.
Your candela is the result of multi variables with multi tolerances.
So when you come with 120 kcd and 1lumen has 119 or 121 kcd I would be surprised.
Besides: 1 Lumen is not one person, but half a dozen. All of them using their own equipment.
So you can expect half a dozen outcomes, with variations in outcome of at least 10%

I always read the outcomes of candela measurement as: ball park and to be cross referenced with others.

When you mod a light and you improve it by 10%, I do believe you.
When someone else mods the same light and improves it by 10%, I believe him/her too.
And I’m sure that all four values (you ↔ him/her, before ↔ after) are different.

So when I’m looking for a new light, I look for several reviewers, and check if they also reviewed a light I already have.
That way I get an impression of what to expect from the new light. Relative, not absolute,

1 Thank
5

As SammysHP said, it’s important to make sure you’re not getting reflections. I get higher readings when I measure in a hallway, for example, than I do when I measure outdoors on a lawn. Because the hallway reflects a lot of sideways light back into the middle.

Some lux meters also just aren’t calibrated well. It might be worth checking with a different model of lux meter. That could also help with the broken peak hold function.

1 Thank
6

Ansi says to measure between 30 seconds and 2 minutes. There can be a variation between 30 seconds and 45 seconds. Some lights might start ramping down a little at 30 seconds or even before that. Some batteries might just start warming up and producing the greatest amount of energy at 40 or 60 seconds. Are you starting with a battery at 4.2 volts or 4.15 volts. Are you allowing the battery to rest after charging and cool down to the ambient 22C plus or minus 3C standard.

I’ve seen some much higher amperage readings with a light that 1 Lumen has tested and published. But then I’ll try a pile of batteries and I can match their numbers.
In the real world I don’t typically charge batteries to full so I don’t like to test things that way.
I’m not doing candela testing so none of this matters.
I’m not saying it’s not important.
What I am saying is, , variables.
You may be inadvertently stacking the deck in some ways. And they may be inadvertently cutting the deck in some ways.

1 Thank
7

I actually had no idea.

I have A/B tested two methods.

A: Outside at night with the light meter on a tripod. Away from anything reflective, or that light could even bounce off of like trees etc.

B: A baffled light tube. Think of 4" diameter telescope body, with a set of baffles that decreases in diameter until the final one is only the size of the sensor. Inside that tube is a black felt (Muso black). My thought was to eliminate reflections and be able to measure inside the house at low humidity.

W20 Outside:
107k
111k
110k

W20 Baffled Tube:
106k
109k
107k

Here is a very rough sketch of what the tube is like.

baffle tube.PNG
baffle tube.PNG466×178 24.5 KB

Not waiting long enough could be a little part of it I suppose. I’ve been measuring on medium mode with the lights to obtain a tighter control on this factor. for both lights the output is almost perfectly stable between turn on and 30 seconds though. With the W20 I get 169 lumens at ten seconds and 168 lumens at 30 seconds.

Additional Testing Needed

I will check this. I could get some readings in the direct center and off center with the light sitting on a tripod. I am considering getting a telescope tripod that has micro adjustment controls on it. That way I could take a reading then move an inch and stop.

I guess maybe I do need to see if a different brand of light meter makes a significant difference.

3 Thanks
8

Very clever construction, and that’s a very informative test. It looks like there is no significant difference between the baffled tube and outdoor measurements–the <2% difference in average value can easily be explained by uncontrolled factors.

Before moving on to the next meter, I would suggest measuring some more familiar reflector-based throwers with the existing meter, as a “sanity check” to make sure you are indeed over-measuring. It’s hard to get reliable reference data for LEPs for a variety of reasons, including but not limited to:

  • Very few reviews/measurements at all
  • Not all reviewers use proper methodology, with a very common mistake being insufficient measurement distance
  • LEPs are by nature more finicky than LEDs. We don’t have reliable datasheets/tests of the underlying LEP module, and the extremely small LES is very sensitive to quality+alignment of secondary optics.
  • I’ve observed massive discrepancies between reviews/measurements of LEP lights (on 1lumen, youtube, etc.), and published specs. It is possible that performance control is loose.

Plus, 2 lights feels like not enough data to fully support the conclusion of over-measurement, and to justify spending much effort on addressing the conclusion.

1 Thank
9

It looks like 1 lumen measured the Thor 1 at start, 30 seconds and 10 minutes.
So I would try all three of those time periods. And they measured amps although you don’t know when they got the amp readings. Get some amp readings and that might help you narrow things down. If you get higher amps that could easily explain your higher lumens. And it states he was using a flat top keeppower 1200mah battery.
That should be a 10 amp battery.

1 Thank
10

This would be my advice: either have a calibrated light source with known intensity, or have your luxmeter calibrated.

I have had my pro luxmeter (Hagner E4X) calibrated at the beginning of this year. It read 5% off from a few years ago.. so they do slightly change over the years, but I’m not sure how much other luxmeters can change over the year(s). I’ve dropped or bumped it a few times, so that might be the reason???
I’ve bought a Testo 545 Luxmeter last year and shipped it to 1 of our 1Lumen reviewers, and its readings were 20% off. But I didn’t want to open it up to see if I can manually tune it.

But as long as you have a calibrated reference, you only need to do a quick calculation at the end.
Basically, you can’t rely on anything unless it’s calibrated, or you have a light source with a known beam intensity, where you will get your correction factor from.

I have had quite a few luxmeters, and some of them can be opened, and manually adjusted. (that’s probably what calibration looks like anyway). And you have some small dials that you can use to get the measurement finetuned.
I finetuned/calibrated a "LX101 Light Meter"and a Extech HD450 based on my pro meter, and its readings were very close to my professional meter. I havent tested it recently, so it might have gone off a bit again.

You could open up the LX1330B, and see if you can see some small screws that you can adjust the readings. Having a light source with known beam intensity is actually the best, either way if you want to finetune it yourself, or just use a calibration factor… that’s the way to go.

I send a flashlight with known lumen output to each new 1Lumen reviewer.
I measure it in my integrating sphere, and include the candelas measured with my pro meter at 5 meters distance at X seconds.
The reviewer can base their integrating sphere measurements and intensity measurements on this calibration light.

Regarding your ways of measuring. Doing an average is fine.
I do them manually, where I put my flashlight down, and manually move the sensor around to get the highest readings when I do 10 or 20+ meters measurements. My luxmeter doesnt have a Max reading setting.
My eyes are still doing okay to see the screen of the luxmeter at 5 meters disance, so that’s where I do it the other way around, and move my flashlight instead of the sensor.
Also, ideally a measurement should be done indoors at X degrees ambient temperatre.
I can’t do any 10 meters measurements indoors, so I have to do it outdoors.. with temperatures always changing.

Besides that, I would advise measuring a certain type of flashlight always at the same distance. Small EDC at 5 meters, long range at 10 meters or 20 meters..
But be consistent with the distance. That way, even if you’re readings are a bit off, people can still compare the measurements with its competitors you also tested. I wouldnt worry about 5% difference or even just a bit higher.. If it’s consistently higher than 10 or more, I would.

2 Thanks
11

Wow, what an illuminating read!

I’ve always assumed that dedicated lux meters can be used without calibration, so thanks for correcting that misconception. Performing calibration with a known/stable/reliable light is great advice.

I do have another question: it seems to me that constant-factor calibration only works if you assume that the response curve is linear: upon receiving L lux, the meter shows a reading of cL, with some constant c that can be estimated from the calibration light. For example: if your meter has response curve L^0.9, then no constant-factor correction can fix the readings.

In your experience, do the response curves of most meters appear truly linear, or deviate little enough from linear that the error introduced is negligible?

12

Today I gathered 4 light meters, and rotated them one by one to the same spot on my kitchen floor below the large overhead LED light.

254 lux -LX1330B Meter 1
258 lux- LX1330B Meter 2
240 lux-Tklamp Meter
247 lux- Smart Sensor meter

My Tklamp integrating sphere comes with a calibration light. I doubt using the calibration light in the sphere means the external puck sensor is “calibrated” too since its a separate sensor, but it tested the lowest so I plugged that into my baffle tube.

I also put the flashlights on a tripod, and moved the light beam across the sensor and STOPPED for 2-3 seconds(so the light was perfectly still) before moving on.

W20 Peak Reading 112k
Thor Peak Reading 204k
6P Tuned Peak Reading 20k

So swapping the meter didn’t seem to affect the outcome a lot. Neither did taking the reading while the light was holding perfectly still. As long as I search for the brightest spot in the beam I still get the repeatable readings.

I also put the W20 on the tripod, pointed the tripod away, then tried to visually pull a random spot near the center as best I could and got 110k. So in the case of the W20 dead center happens to be a pretty good spot to measure.

What do you suggest for obtaining a calibrated reference?

1 Thank
13

Matt. I test for umen, and yes Henk is right.

We use our own tools, but they’ve been calibrated against known baselines. It doesn’t really matter which instrument is used, there’s a calibration factor involved.
I use a Testo 545 luxmeter. Used to use a Uni-T 383S prior (for a long time) and that one was okay for an inexpensive meter, but the Testo is a professional, lab-grade instrument. Even so, it’s been calibrated (as mentioned) and is very accurate.
The DrMeter you use is also okay, and will give you a good ballpark figure. My rule is a good figure is going to be +/- 10% of factory.

14

@QReciprocity42: I don’t know too much about the details, I’ve just come to a point that I trust my gear for what it is, and don’t overthink it. Especially knowing my luxmeter was only 5% off since the last calibration a few years before :slight_smile:

@Lumencraft-Matt: yes, it’s usually in or around the center. For me it’s a bit tricky to measure LEPs, because I have to walk 20 meters and manually measure it. Whenever I notice that the measurements don’t make sense, I do another round, and often also another test at 5 meters indoors.. (so the temps are better) but measuring LEPs at 5 meters is just a bit too crazy difficult..= a slight shake with my hand results in thousands of candelas difference :smiley:

1 Thank
15

About the calibrated reference: I would suggest getting a good calibrated luxmeter. And if you know a company that can do the calibration, you can always send it in when you start doubting your measurements.

1 Thank
16

Another factor in measuring candelas…

A lot of lights don’t collimate perfectly. Instead of shining rays in parallel like an = shape, the light comes out in a < shape or an X shape.

If the beam has an X shaped collimation pattern, you’ll get artificially inflated readings if you measure at the convergence point in the middle of the X. An accurate reading needs to be measured well past that convergence point.

The official specs may also have not been measured at a good place in the beam. It could have been measured before the convergence point, for example, too close to the light, in the middle of a subtle “donut hole”.

To check if a particular light has this type of beam, you would need to make several measurements at various distances, and check if the candela calculation changes. Like, maybe it measures at 80 kcd at 3 m, 100 kcd at 15 m, and then 80 kcd again at 50 m. Or maybe the official spec measured at just 1 m, which is usually not accurate for throwers.