Proposal of electronic & light output criteria.

[shankus]

I have been considering criteria for light output. It seems that there is no industry standard for stating output of a battery powered lighting device. With ambiguous terms such as “useable light”, and “moon mode”, one is left asking questions about what these terms mean, and at best, these terms are wholly subjective. What one user considers usable light, might be unusable to another user.
Given that there seem to be no standards for light output, I figure who better to establish some criteria than the armchair experts and engineers (and degreed engineers) at CPF.

I would like to see members in the lighting industry, and members that are electrical/electronic engineers weigh in on this topic. If you know a member that is in these groups, please PM or email them to look at this thread.
Also, members that review lights input would be valuable and appreciated.


Here are my proposals:

Non-regulated lights
Perhaps a light would be considered non-full-brightness when it falls to 50% of it’s light output with a given cell type. For example, with fresh alkalines, “light A’s” output is X candelas (or lux or lumens etc.). After 8 hours, it has fallen to 0.5 X candelas. So, “light A” lasts 8 hours on alkaline cells.
And/or, “light A” lasts 6 hours on 2000 mAh NiMH cells.

I have also seen suggestions for ratings of light outputs of 75%, 50%, 25%, as well. It seems that stating the battery life of a light would be simpler with one rating, instead of three or four, though. And it is unreasonable to rate a non-regulated light at anything near 100% brightness, since it is dimming from the moment it is turned on.
Perhaps this rating could be called “BL (for Battery Life)”.



Regulated lights (designed for constant light output):
Perhaps a light would be considered non-full-brightness when it falls to 75% of it’s light output with a given cell type. For example, with fresh alkalines, “light B’s” output is X candelas (or lux or lumens etc.). After 8 hours, it has fallen to 0.75 X candelas. So, “light B” lasts 8 hours on alkaline cells.
Perhaps this rating could be called “BLOR (for Battery Life, Output Regulated)”.



Regulated lights (other):
It gets tricky here, I think. Some regulated lights are designed to mimic an unregulated light’s output (MadMax), or to just boost voltage to a working level or to get the most life from the cells (Arc AAA).
Perhaps one of these lights would be considered non-full-brightness when it falls to 50% of it’s light output with a given cell type.
Perhaps this rating could be called “BLR (for Battery Life Regulated)”.


I’m suggesting this because the ambiguity in light output is frustrating. It seems that no manufacturers have reasonable ratings for the battery life of their lights. Usually you have to cut their stated life by ½ or more, to estimate real usability.

 

[INRETECH]

I suggested a very nice and standarized format a couple of months ago, saying that the light should be:

INIT-C/MINS

Init-Candlea

Mins = Number of minutes before the light drops to 1/2 intensity

I know that it would make some lights "look good" and others "bad"; but it would be a start - I got SHOT down nig time from a lot of people, so I decided to drop it

 

[Jonathan]

I don't want to get into a 'shoot each other' type debate again; but we do need to recognize that this will be a complex issue, since lights are rather complex beasties. _Any_ attempt to simplify lights to a single number or pair of numbers will, in my opinion, oversimplify, and thus hide important details.

I do believe that in the last discussion, there was some feeling that you (Mike) were suggesting standards which made Inretech direct drive lights look comparatively better than regulated lights, and the response to this feeling was to 'shoot you down'. I want to put this on the table up front to get everyone (myself included) to own up to their own biases, to recognize that _all_ simplifications of the vast number of parameters which describe a light will _always_ be biased. I don't think that Mike is in any way wrong to suggest a particular measurement that will have a bias, because _all_ the measurements will have different biases.

I would suggest that an improved version of Mike's two number suggestion would be the _average_ _lumen_ output, and the time for the lumen output to fall from initial to half.

My reasoning is that a light with steady output and the same operating time should get a higher 'score' than a light where the output drops over the same time period. For example, a light which initially produces 100 lumen, and then falls in a linear fashion to 50 lumen at 100 minutes would get a score of 75/100 whereas a light which produces 100 lumen for 100 minutes and then suddenly falls off would get a score of 100/100...whereas by Mikes suggestion they would both get exactly the same 'score' (that is, if they both had the same light output distribution so that their initial candlepower were the same, not just their initial lumen output).

I would also suggest lumen output rather than candlepower to remove issues of just how tight the beam is.

Again, any single number, or even a pair of numbers, will _not_ do a light justice...but thinking about what numbers are useful _will_ help us pick out better lights.

-Jon

 

[FreeBSDboy]

If you have the equipment handy to be able to measure the intensity of the light, then you certainly should also have the ability to make time based plots of that intensity. I say just give a plot of intensity and voltage v time and let people draw their own conclusions.

It is not right to make value judgements inherent in your data reduction method (ie, lights stop being usefull at X% intensity) because then it certainly will not fairly represent all groups.

Then if people want to argue, they can argue about the merrits of what light charastics are better, in the comfort of the knowledge that the testing data was not biased in any particular manner.

The part that would need to be standardized is the testing method of collecting the information. (ie: ambient light conditions and distance to your measuring equipment, battery type, etc)

Just my 2 cents.

 

[Roy]

On some of my Run-Time Plots, I've been known to report a number that I call the "half brightness value". This is nothing more than how long it takes a flashlight/battery combination to reach half of its initial brightness. It must be remembered that the "half brightness value" should NOT be used in compairing the performance of different flashlights.

I selected 50% because I seem to remember that the first the human eye will notice a dimming of a light is at about 50% of the starting brightness. This is easy to measure and does not require expensive equipment, just a photo-cell and a multimeter. Distance between the light and photo-cell is not a critical factor either.

 

[shankus]

I didn't know this was a "sore" subject. I also didn't intend the criteria to be biased in any way. I suggested different criteria for lights in different catagories for just this reason. It isn't reasonable to compare a constant brightness regulated flashlight to a direct drive light with an "across-the-board" measurement. They are different pieces of equipment, with different design philosophies & purposes.
What I'm suggesting is an adaquate way to express the runtime of a light, no matter how bright it is. The only brightness of concern would be that of the light compared to itself, from initial battery condition, to some standard percentage of brightness. The light would only be "competing" with itself.


[ QUOTE ]
Jonathan said:
I would also suggest lumen output rather than candlepower to remove issues of just how tight the beam is.


[/ QUOTE ]
I was hoping someone would bring this up, due to my ignorance on the subject. I remembered reading that some measurements were different than others in the method that they were taken.

 

[Jonathan]

FreeBSDboy,

I do not believe that it is possible to have data reduction that is free of bias. The method that you suggest, providing a _graph_ of intensity versus runtime, is _much_ better than the simple 'pairs of numbers score' the rest of us have been discussing. With a graph you _know_ if the light output remained on full and then fell off sharply, or fell of to 75% at the very beginning, but then held on for a long time at that plateau, or something else.

But the graph is quite a bit more data than a single pair of numbers; in essence you are doing less data reduction, so your results are fairer. But even your suggested method has its built in bias. For example, you compare intensity, rather than total light output, which means that a system with a tight beam reflector will 'look better' on paper than a system that produces the same total light output, but in a broader beam. Additionally, what happens if as the light is used, the intensity falls off around the edges, but stays the same in the center? Total light output could go way down, but the intensity measurement would stay the same. (Hmm, maybe I should figure out a way to do just this...I could see it being quite useful /ubbthreads/images/graemlins/smile.gif

-Jon

 

[Quickbeam]

[ QUOTE ]
Non-regulated lights
Perhaps a light would be considered non-full-brightness when it falls to 50% of it?s light output with a given cell type.

[/ QUOTE ]

The reason this won't work is that the alkaline discharge curve is non-linear. 25% dropoff is attained very early on, as is 50% while 75% and 90% take a long time, especially with LED lights. By stating times for brightness levels of 75%, 50%, 25% and 10% you can figure out the actual curve.

Examples:

Non regulated light:

75% = 20 min
50% = 2 hours
25% = 7 hours
10% = 24 hours

Regulated light

75% = 4 hours
50% = 4 hours 10 minutes
25% = 4 hours 15 minutes
10% = 9 hours

Semi-Regulated light:

75% = 2 hours
50% = 3 hours
25% = 4 hours
10% = 24 hours

If I were just to state that 50% was at 2 hours the assumption in the mind of the consumer would be that 75% was at 1 hour and 25% would be at 3 hours since most people think in terms of linear curves unless told otherwise...

A graph of output over time is the only way for runtime to be completely accurately displayed. Barring that, a list of percentages at intervals will work, although not perfectly.

 

[INRETECH]

We need to make it a few numbers, say two - so that people can quickly judge the products, yes - it will make some products look GREAT and others look BAD; but it will allow a STANDARD to judge flashlights by

I am so tired of seeing meaningless claims on products, like a LED flashlight with 10x5mm LEDs that puts out 10,000 cp

A customer came up to me at our booth one time with a flashlight that used 2x123 batteries and drove a 5w Luxeon, it put out a **LOT** of light, but it could only run that LED for about 5mins !

I would like to see my standard "resumed" as:

LUMENS/MINS till 1/2 brightness

And used "across the board" for all products

 

[Blikbok]

Quickbeam's system seems very fair and very adaptable all types of lights, and easily printed on packaging or typed onto CPF. Not all of us have mastered posting pictures, let alone graphing output from an analog instrument. /ubbthreads/images/graemlins/smile.gif It's simple, quick, and conveys most of the information most people need, without graphs.

I feel the most important thing to combat is the "thousands of hours" claims by some marketeers.

 

[INRETECH]

I disagree, one standard for all - it would be fair and honest; the customer could easily compare products without having to worry about type of flashlight, when the customer pushes the button he/she wants light; how the internal operation of the flashlight works doesn't interest the customer

When I use a voltmeter to measure a battery, it doesn't matter what kind of battery it is - I still use the same voltmeter, some batteries measure higher - some lower

 

[Quickbeam]

[ QUOTE ]
yes - it will make some products look GREAT and others look BAD;

[/ QUOTE ]

...which is exactly why the industry won't adapt to that kind of a standard. If my products look bad, especially to an unfair degree, I wouldn't use that method. Graphs and/or listed percentages don't make any product look good or bad, they just state facts. Let the consumer then decide if the runtime matches their need. Plus it doesn't matter how the light output is measured - Lux, foot candles, lumens, etc. 50% output is 50% no matter what...

Now everything I've said has to do with RUNTIME versus a generic axis called "LIGHT OUTPUT". That's why it will work. Once you start adding in lumens, candlepower, foot-candles, etc. it will get too confusing - there is simply no way to give this information to the consumer without them being very well educated on the subject.

For example, if I use LUX or foot-candles, that's only telling me the throw of the beam due to differing optics and reflectors. Plus these will give different readings if measured at different distances. They don't tell anytihg about beam width and brightness of the corona. If I use Lumens, that only tells me total output, not how well the beam throws, the quality of the beam, and how wide of a beam is put out by the light...

The BEST way to show this information is to use a light output beam profile pictures like Craig uses in addition to the % output runtime graphs ...

This gives output at various parts of the beam, thereby displaying intensity and beam contour.

So I guess what we really need is a piece of equipment that produces graphs like the one above only at a standardized distance with a standardized unit of measurement allowing for no deviation by a manufacturer so they could "fudge the numbers" along with a runtime graph of % output...

I'm afraid there is no easy answer.

 

[Jonathan]

[ QUOTE ]
INRETECH said:
We need to make it a few numbers, say two - so that people can quickly judge the products, yes - it will make some products look GREAT and others look BAD; but it will allow a STANDARD to judge flashlights by
...
I would like to see my standard "resumed" as:

LUMENS/MINS till 1/2 brightness


[/ QUOTE ]

[ QUOTE ]
INRETECH said:
I disagree, one standard for all - it would be fair and honest; the customer could easily compare products without having to worry about type of flashlight, when the customer pushes the button he/she wants light; how the internal operation of the flashlight works doesn't interest the customer

[/ QUOTE ]

This is a good point. The innards of the flashlight are irrelevant as far as a measurement of the light output are concerned. In fact, you would want to use the same sort of measurement across types of flashlight so that you could make general comparisons between the different types.

However, under slightly different standards, the relative strengths of different lights will make different products look better or worse. So you can't simply pick a standard and apply it equally to every light; you _also_ need to pick a standard data reduction method that fairly represents the different lights.

If we were to pick as the simple pair of numbers standard ' initial light output, and minutes to 90% brightness', then direct drive LEDs would be made to look unfairly bad relative to regulated systems, even though the same measurement was 'fairly' applied to each system. In a regulated system, once you fall away from 90% brightness, you are pretty close to the end of usefulness of the light. But in a direct drive system, you have quite a bit of time spent between 90% and 50% brightness, likely very useful lighting.

Again, I would suggest a slight modification to your proposal:

(Average Lumens) / minutes to 1/2 brightness

This would essentially average out any changes in brightness during the useable light period, IMHO more fairly measuring both regulated and unregulated sources.

As an example, consider three lights:

1) Starts at 100 lumen. Falls to 50 lumen in a linear fashion over an hour.
2) Starts at 100 lumen. Holds 100 lumen for 15 minutes, and then fall off
3) Starts at 100 lumen. Holds 100 lumen for an hour, then falls off.

By the (suggested as a bad) standard of initial lumen/time to 90%, the ratings would be:
1) 100/12 2) 100/15 3) 100/60
But this is unfair, because I would claim that light 1 is 'better' than light 2.

By the (proposed by Mike) standard of initial lumen/time to 50%, the ratings would be:
1) 100/60 2) 100/15 3) 100/60
Which IMHO ranks 1 and 2 properly, but IMHO 1 and 3 should have different ratings, with 3 better.

By my proposed standard of average lumen/time to 50%, the ratings would be:
1)75/60 2) 100/15 3) 100/60
Which IMHO ranks all three lights in proper order. Of course that is my bias /ubbthreads/images/graemlins/smile.gif

-Jon

 

[INRETECH]

Again, it doesn't MATTER how the insides work to the customer, whether its:

) Direct drive
) Voltage-mode
) Current-mode
) Linear
) Buck
) Ripple
) R^2
) Cuk'
) Black Magic

The customer just wants to know the "bottom line", people email me or visit my booth, they always ask the two following questions:

1) How much ?
2) How many lumens (or cp) ?

They never ask "how does it work"

Like the sticker on the cars: Miles Per Gallon and Price

Same exact treatment for all, in fact - this would "entice" companies that get a "bad" score to re-think their product

The customer KNOWS that since the product is battery based, it is going to start out bright and slowly dim

The customer just wants to know HOW BRIGHT and HOW LONG; if they buy that little 5w flashlight that I saw that ran on 2x123 batteries, it would might be something like 120/5 (120 Lumens to start, and 1/2 brightness in 5 minutes)

Or in the case of the TriLight - the score might be something like 54/200 - meaning 54 Lumens to start with and stays about 1/2 brightness for 200 minutes, not as bright to start with - but lasts a long time

There are people out there that want "sports cars" - that is a LOT of light for a short period of time and there are people out there that want "SUVS" - a flashlight that gives a LOT of light over a long period of time; having ONE standard is the **ONLY** way to go, the person will be able to look on the side of the box - see the numbers and JUDGE what he or she is looking for in a flashlight

Before the FCC stepped into the music industry, there were so many "ways" that companies measured power on their stereos that it confused so many customers

) Peak power
) Peak music power
) Etc

And then the FCC just came in and said "Power RMS"

 

[Jonathan]

[ QUOTE ]
INRETECH said:
Again, it doesn't MATTER how the insides work to the customer, whether its:
Or in the case of the TriLight - the score might be something like 54/200 - meaning 54 Lumens to start with and stays about 1/2 brightness for 200 minutes, not as bright to start with - but lasts a long time

There are people out there that want "sports cars" - that is a LOT of light for a short period of time and there are people out there that want "SUVS" - a flashlight that gives a LOT of light over a long period of time; having ONE standard is the **ONLY** way to go, the person will be able to look on the side of the box - see the numbers and JUDGE what he or she is looking for in a flashlight


[/ QUOTE ]

So I come along, and as a competing manufacturer build a _regulated_ 3 luxeon lamp. The light output starts at 48 lumen, and _stays_ at 48 lumen rock solid for 200 minutes, and then the output falls off to zero.

The customer doesn't care what is inside the box. They just care about the light output.

Well, I claim that _my_ hypothetical lamp produces more total output than the Trilight. For some significant portion of the 200 minutes run time, probably the majority of the time, my hypothetical lamp (and remember, this is just a hypothetical; I don't know if I could really build it) is brighter than the Trilight. But by the scoring standard that you have proposed, my lamp gets a lower score (48/200).

I am _not_ arguing against a simple 'MPG' number. I am arguing against the particular simple number that you are suggesting. And I am not even arguing far away from the particular simple number that you are suggesting; I am arguing that you should replace the initial light output value with the _average_ light output value.

-Jon

 

[INRETECH]

How would you quanify "average light output" ?

 

[Jonathan]

I would suggest lumen output averaged over time.

The time period for the averaging would be from initial turn on with fresh batteries until the defined end of discharge, when the lumen output falls to 1/2 the initial value. So if the claimed run time were 100 minutes, then the average would be calculated over that entire 100 minutes.

A source which started at 100 lumen, and linearly dimmed to 50 lumen would have an average output of 75 lumen. One that started at 100 lumen, and held that 100 lumen until the bitter end, and then suddenly dropped off, would have an average output of slightly less than 100 lumen. One that started out at 100 lumen, fell very quickly to 60 lumen, held steady at 60 and then suddenly fell off would have an average output of slightly more than 60 lumen, etc.

-Jon

 

[INRETECH]

So, you are saying add up the Lumens for each minute that the light intensity is greater than 1/2 the initial light and then divide that total by the number of minutes ?

Seems fair enough

 

[Jonathan]

[ QUOTE ]
INRETECH said:
So, you are saying add up the Lumens for each minute that the light intensity is greater than 1/2 the initial light and then divide that total by the number of minutes ?


[/ QUOTE ]

Essentially yes, although I was hoping to be slightly more general than that.

1) Rather than making the cutoff of the averaging process fixed at 1/2 the initial light output, I suggest be the cutoff be whatever the standard is using for the 'end of time' cut-off. So if everyone in the discussion decided that the end of the run time should be determined by 1/4 initial light output, then the average should similarly go to 1/4 initial light output.

2) The averaging process is _continuous_, not once per minute. The once per minute approach would closely approximate the continuous averaging process, except for lights with _very_ short run times. The technique is to measure the light output at regularly spaced intervals, sum the light output, and then divide by the number of measurements, and the frequency of the measurements has to be high enough to closely approximate the continuous averaging function. Maybe you need to measure once per second, but then your sum gets divided by the number of second.

-Jon

 

[Doug Owen]

Lots of interesting and well thought out discussion going on, I like that is a forum......

IMO, this is an excellent proposal for a *single* rating, 'battery lifetime'. Minutes (in continuous run mode) to half light output. Very similar to the consumer to the MPG times gas tank capacity in the car analogy (how far will my new car go before 'running out of gas'?), for sure a valid issue. This measure of performance is easily within the ability of lots of folks (completely unlike light level measurements), so it's easy for us to keep each other honest.

The standard should be the same for all fashlights, regardless of circuit type. Run time is run time.

I suggest it's an excellent measure, a well understood standard very hard to slant or otherwise 'cook'. Other important characteristics (like output level and discharge profile) can be dealt with using other specs. The bottom line is a guy buys a flashlight to make light. A bigger, more powerful one if he wants more light. The how big and how much light are valid decisions. It's however, quite reasonable to ask 'how long do the batteries last?'. This plan offers an objective, honest answer. Let's not complicate it or dilute it by trying to make this simple measure do more?

Or so I see it.

Doug Owen