LED Spec sheets and graphs

[idleprocess]

I have long been reading spec sheets on LED's and have noticed that while the manufacturers go to great lengths to produce nice graphs showing performance (Vf vs Current and Relative Flux vs Current are common), they almost never include that data in an immensely-more-useful table format. I have often stared at these graphs and guessed what the values are, but short of taking the truly anal step of printing out the graphs on a full-size sheet of paper and measuring X/Y offset, I can't really extract values from them.

A table of typical values could be used to derive equations to approximate performance as well as have some discrete values to work with rather than trying to guess what the value is by looking at a graph where the performance curve almost never intersects the X & Y grid.

Is there a source of tabular data? I can't expect to get samples as a hobbyist for testing and lack numerous pieces of test equipment (such as an integrating sphere or even a light meter calibrated for LED's). I've sufficient skill in Excel to come up with decent back-of-the-napkin performance metrics should I come across a source of this data. I understand that there are manufacturing variations that will make these calculations of limited value, but I just want to be able to perform apples-to-apples comparisons at this point.

 

[RoGuE_StreaK]

Main one I look at currently (no pun intended) is Relative Flux vs Current, and for that I could see reasons why they don't have tables; it's given (well, for the Cree xm-l and xp-g) as % vs current, and this percent is relative to the rated flux of each bin. So for example the XM-L is rated at 50% flux at around 350mA, but 50% of what? There are eight bins listed, you have to choose your bin, check it's rating for 700mA (100% baseline) and extrapolate from there.
Just a quick thought on possible reasoning.

PS. There's never any harm in attempting to get samples, you'd be suprised as to how often one will arrive! :sssh:

 

[idleprocess]

Discrete (X,Y) coordinates of relative flux at sample currents would allow rapid extrapolation of where to expect a given bin to perform - which is why flux is expressed as a percentage rather than lumens. Forward voltage could be expressed the same way for manufacturers that offer Vf bins.

About 4-6 points is all it would take to determine a workable curve.

 

[idleprocess]

duplicate post

 

[pretmetled]

I have long been reading spec sheets on LED's and have noticed that while the manufacturers go to great lengths to produce nice graphs showing performance (Vf vs Current and Relative Flux vs Current are common), they almost never include that data in an immensely-more-useful table format. I have often stared at these graphs and guessed what the values are, but short of taking the truly anal step of printing out the graphs on a full-size sheet of paper and measuring X/Y offset, I can't really extract values from them.

I recently went through pretty much the same thing. I wanted the relative intensity vs wavelength graph for several white leds, and ofcourse no tabular data anywhere to be found. So I did the following:

- open datasheet in pdf reader
- zoom in on graph for maximum fit
- take screenshot
- cropping screenshot optional, a bit depending.
- open screenshot in GetData
- do a get data points / digitize area + set the axis scale
- export data to text file
- And you're done.

It looks to be more work than it actually is. Works pretty fast if you do it in batches.

Oh, and GetData can be downloaded here:

http://www.getdata-graph-digitizer.com/download.php

Hope that helps.

 

[LilKevin715]

Checkout out JTR's lengthy White LED lumen testing thread. It hasn't been updated in a while, but it does have a lot of good info.

 

[videoman]

If anyone needs a good tool to determine CREE led performance ( and this is only for Cree leds) it's called Cree Product Characterization Tool. It is really a free downloadable table that is extremely useful to compare led performance across the board. All you need to feed in is the led model and the bin, the chart will automatically let you see at a glance all the relevant values, no math involved here. http://pct.cree.com/

 

[RoGuE_StreaK]

Nice link videoman, should be highly useful for checking out if something's worth the extra cost if driving at a certain current, eg. a T5 XM-L vs an R4 XP-G

 

[pretmetled]

Checkout out JTR's lengthy White LED lumen testing thread. It hasn't been updated in a while, but it does have a lot of good info.

Yep, that's a very informative thread. In fact, it's that sort of useful thread that brought me to this forum in the first place. :)

If anyone needs a good tool to determine CREE led performance ( and this is only for Cree leds) it's called Cree Product Characterization Tool

Useful, thanks! On that subject, this one might also be of interest:

http://www.national.com/en/webench/led.html

 

[idleprocess]

I recently went through pretty much the same thing. I wanted the relative intensity vs wavelength graph for several white leds, and ofcourse no tabular data anywhere to be found. So I did the following:

- open datasheet in pdf reader
- zoom in on graph for maximum fit
- take screenshot
- cropping screenshot optional, a bit depending.
- open screenshot in GetData
- do a get data points / digitize area + set the axis scale
- export data to text file
- And you're done.

It looks to be more work than it actually is. Works pretty fast if you do it in batches.

Oh, and GetData can be downloaded here:

http://www.getdata-graph-digitizer.com/download.php

Hope that helps.

I'll have to look at that - sounds like a workable way to extract useful data from graphs without so much guesswork.

If anyone needs a good tool to determine CREE led performance ( and this is only for Cree leds) it's called Cree Product Characterization Tool. It is really a free downloadable table that is extremely useful to compare led performance across the board. All you need to feed in is the led model and the bin, the chart will automatically let you see at a glance all the relevant values, no math involved here. http://pct.cree.com/

Neat, exactly what I was looking for in this instance!