Match LED Brightness

[Kasper]

To add, op said they are sourcing from JLCPCB smt library so I am guessing the 2 leds are not on opposite sides of the efficiency spectrum.  If op chooses the most expensive of one color and the cheapest of the other color that probably won't help.

[Kasper]

I edited my earlier post to add: efficiency and viewing angle may be so different that these quick solutions do not help.

[tooki]

As Siwastaja said, they still make the old-style low efficiency LEDs, which was my point (I dug out an actual old one simply because it was a sure way to ensure it wasn’t an efficient chemistry).

But no, your first method isn’t any help in practice. As Siwastaja also said, LED efficiency spans multiple orders of magnitude, even among modern LEDs. Choosing a similar drive current tells you nothing about whether two different LEDs will have similar light output.

[Zero999]

@StillTrying
I'm trying to spec out blue and yellow LEDs, I have some THT ones but SMD is a mystery.

It wouldn't surprise me if the yellow needs 50%+ more current than the blue to look similar brightness.

Yes, that's certainly true of old yellow LEDs. A much better alternative is to use phosphor converted LEDs which are just deep blue/violet/UV LEDs with a yellow phosphor. They have a higher forward drop, but are much more efficient than the crappy old LEDs and have a much broader spectrum, so are often specified in chromaticity, rather than dominant wavelength.
https://static.rapidonline.com/pdf/55-9182.pdf

[Kasper]

As Siwastaja said, they still make the old-style low efficiency LEDs, which was my point (I dug out an actual old one simply because it was a sure way to ensure it wasn’t an efficient chemistry).

But no, your first method isn’t any help in practice. As Siwastaja also said, LED efficiency spans multiple orders of magnitude, even among modern LEDs. Choosing a similar drive current tells you nothing about whether two different LEDs will have similar light output.

Best thing to do is look at the mcd rating and only compare leds that have the same viewing angle.
That will match them up pretty well.

I looked up 7 different smt LEDs on JLCPCB parts list.  Just used ones rated at 20mA.

Did not even find 1 order of magnitude difference in efficiency.  And found they all had similar viewing angles.

MPN:
   Yellow:
      19-213/Y2C- CQ2R2L/3T(CY), 17-21SUYC/TR8, FC-DA1608HYK-588J, E6C0805UYAC1UDA
   Blue:
      X L - 0 8 0 5 Q B C, E6C1206QBAC1UDA, MHT170UBCT
      
Yellow:
   Intensity: Min, Typ, Max (MCD): 70, 125, 220
   View angle: Min, Max: 120°, 130°
   Typ Vf: 2V
      
Blue:
   Intensity: Min, Typ, Max (MCD): 50, 111, 225
   View angle: Min, Max: 120°, 140°
   Typ Vf: 3V
   
Comparing yellows to blues, there was much less difference in intensities and viewing angles than there was in Vf.

Resistor voltage with 3.3V supply:
      Vr_yellow = 3.3V - 2V = 1.3V
      Vr_blue = 3.3V - 3V = 0.3V

Example currents with 3.3V supply and 65 ohm resistor:
   If_yellow = 20mA
   If_blue = 5mA
   
Typical luminous intensity at 20mA:
   Yellow: 125 MCD
   
Typical luminous intensity at 5mA (based on linear approximation of current vs luminous intensity):
   Blue: 28 MCD
   
Results:
   Difference in typical MCD based on using same value resistor for both LEDs: 125 MCD / 28 MCD
   4.5 times difference

   Typical difference in MCD between yellow and blue: 125 MCD / 111 MCD
   1.1 times difference

   Worst case difference in MCD (max yellow / min blue): 220 MCD vs 50 MCD
   4.4 times different

Conclusion:
   In this case difference in Vf is as impactful as difference in efficiency. 
   If you go on digi-key and look up the extremes then you can find some big differences in efficiency but that comes with different packages and different costs.  Odds are the purchaser is going to choose identical packages and similar costs and it seems like that will yield similar efficiencies. At least similar enough that there is room for current matching to cause an improvement in intensity matching.
   

[tooki]

Well, I just grabbed the electronics catalog that happened to be closest at hand (in this case, Reichelt) and on the first page of LEDs alone, just within the 3mm (T1) LEDs, the light output of the LEDs listed spans 3 mcd to 14,000 mcd at the same current in the same package. Yes, differences in viewing angle account for some of that, but my point remains that "Choosing a similar drive current tells you nothing about whether two different LEDs will have similar light output."

If you're starting by selecting two LEDs with similar rated output at the same current, as you did, then yeah, current tells you something. But if you're looking at LEDs as a whole, it tells you nothing.

[Psi]

I remember when i designed some ID badges for a gaming event around 7 years ago.
I did some datasheet research looking at 100's of leds to find the most efficient LED i could.
We wanted the ID badge lighting to last for 3 days straight on two CR2032 but be as bright as possible.

I bought around 30 LEDs from digikey that looked promising and ended one with a InGaN/GaN orange LED that was amazing. 
At 1mA it was about the same brightness as most others are at 20mA.

[Kasper]

If you're starting by selecting two LEDs with similar rated output at the same current, as you did, then yeah, current tells you something.

Before accusing me of cherry picking my data, you should look at their site.  Relying on such a big assumption just shows you don't have a good argument.

I did not select LEDs with similar rated outputs on purpose.  I got LEDs with similar rated outputs because that's what is on their site. 

I didn't even limit my selection for package size, viewing angle or cost.  I selected the first handful of blues and yellows.

There were a couple with lower intensity but they also had lower current so I did not include them and I was clear about that filtering for 20mA.  I almost included them as extra evidence of my argument that lower current generally yields lower output but my post was already too long.

[tooki]

If you're starting by selecting two LEDs with similar rated output at the same current, as you did, then yeah, current tells you something.

Before accusing me of cherry picking my data, you should look at their site.  Relying on such a big assumption just shows you don't have a good argument.

Bro, you lost the disagreement long ago. You made a claim that is readily disproven, and multiple people have done just that. You’re just making yourself look silly by digging in your heels on something that is so easily disproven.

Anyhow, I didn’t say you cherry picked your data. You selected LEDs with broadly similar light output. How that came to be is irrelevant. You chose randomly and so did I when I literally grabbed the closest catalog at hand. But the upshot is, your claim that LEDs of similar current will have similar brightness is plain and simply not true.

[Zero999]

I remember when i designed some ID badges for a gaming event around 7 years ago.
I did some datasheet research looking at 100's of leds to find the most efficient LED i could.
We wanted the ID badge lighting to last for 3 days straight on two CR2032 but be as bright as possible.

I bought around 30 LEDs from digikey that looked promising and ended one with a InGaN/GaN orange LED that was amazing. 
At 1mA it was about the same brightness as most others are at 20mA.

It sounds like a phosphor converted LED, similar to the one in the datasheet, I linked previously. Do you have the part number or data sheet?

I found the same yolk yellow LED in TME, for those in Europe who'll find it easier to order it from there, than Rapid. No doubt a similar type of LED can be found in the US.
https://www.tme.eu/gb/details/oscm4l5111a/tht-leds-5mm/optosupply/

[Kasper]

If you're starting by selecting two LEDs with similar rated output at the same current, as you did, then yeah, current tells you something.

Before accusing me of cherry picking my data, you should look at their site.  Relying on such a big assumption just shows you don't have a good argument.

Bro, you lost the disagreement long ago. You made a claim that is readily disproven, and multiple people have done just that. You’re just making yourself look silly by digging in your heels on something that is so easily disproven.

Anyhow, I didn’t say you cherry picked your data. You selected LEDs with broadly similar light output. How that came to be is irrelevant. You chose randomly and so did I when I literally grabbed the closest catalog at hand. But the upshot is, your claim that LEDs of similar current will have similar brightness is plain and simply not true.

Once again you prove you have no good argument by choosing fallacies over details.

[Nusa]

Keep in mind that fine-tuning by eye assumes that everyone that's going to use the device has the same eyes. Color perception varies greatly among the population.

[codingwithethanol]

WOW!

Lot of replies, not sure how to respond but i'm at least glad I started an interesting line of discussion. For the time being i'm going to design the circuit with same color LEDs, but in the future i'd actually like to order a variety of SMD LEDs and settle this via experimental data once and for all. I'll bump this thread when the badges come in so keep an eye out.

[Psi]

It sounds like a phosphor converted LED, similar to the one in the datasheet, I linked previously. Do you have the part number or data sheet?

My notes from back then are a bit of a mess.
But i think it was this one
https://www.digikey.com/product-detail/en/kingbright/WP7104SYC/754-1257-ND/1747656
I don't think it was a 'special' led.

I ordered around 30 different leds based on their datasheet specs and price.
Out of the 30 that one was the brightest for the least current.

[Kasper]

WOW!

Lot of replies, not sure how to respond but i'm at least glad I started an interesting line of discussion. For the time being i'm going to design the circuit with same color LEDs, but in the future i'd actually like to order a variety of SMD LEDs and settle this via experimental data once and for all. I'll bump this thread when the badges come in so keep an eye out.

I applaud your enthusiasm but I think a digikey search would be a more thorough way to settle it though even that is left open to arguments over how tightly it should be filtered for package size, viewing angle, current, cost, etc.

On a side note, I am interested what the part numbers were for the 2 LEDs that you were talking about at the start of this.  That is too small of a sample to really prove anything, but I am curious.

[tooki]

If you're starting by selecting two LEDs with similar rated output at the same current, as you did, then yeah, current tells you something.

Before accusing me of cherry picking my data, you should look at their site.  Relying on such a big assumption just shows you don't have a good argument.

Bro, you lost the disagreement long ago. You made a claim that is readily disproven, and multiple people have done just that. You’re just making yourself look silly by digging in your heels on something that is so easily disproven.

Anyhow, I didn’t say you cherry picked your data. You selected LEDs with broadly similar light output. How that came to be is irrelevant. You chose randomly and so did I when I literally grabbed the closest catalog at hand. But the upshot is, your claim that LEDs of similar current will have similar brightness is plain and simply not true.

Once again you prove you have no good argument by choosing fallacies over details.

Your claim has been completely disproven, end of story. No fallacies, no cherry picking. Your stubbornness is astounding, given that it’s literally trivial to disprove your claim. One 20mA LED with 3mcd, another 20mA LED with 14,000mcd. Not even in the same ballpark of brightness.

[james_s]

Buy some samples of the LEDs you wish to use, then experiment with variable resistors to get the brightness you want. Measure the resistance of the pots then and select fixed resistors of similar value.

This is waaay easier than trying to calculate from the datasheets and will give better results anyway. You're always going to need a different resistor value for each color, the forward voltage for example of a red LED is usually around half that of a blue LED.

[kripton2035]

a device like this one may help too.
http://www.neufeld.newton.ks.us/electronics/?p=1145

[Kasper]

If you're starting by selecting two LEDs with similar rated output at the same current, as you did, then yeah, current tells you something.

Before accusing me of cherry picking my data, you should look at their site.  Relying on such a big assumption just shows you don't have a good argument.

Bro, you lost the disagreement long ago. You made a claim that is readily disproven, and multiple people have done just that. You’re just making yourself look silly by digging in your heels on something that is so easily disproven.

Anyhow, I didn’t say you cherry picked your data. You selected LEDs with broadly similar light output. How that came to be is irrelevant. You chose randomly and so did I when I literally grabbed the closest catalog at hand. But the upshot is, your claim that LEDs of similar current will have similar brightness is plain and simply not true.

Once again you prove you have no good argument by choosing fallacies over details.

Your claim has been completely disproven, end of story. No fallacies, no cherry picking. Your stubbornness is astounding, given that it’s literally trivial to disprove your claim. One 20mA LED with 3mcd, another 20mA LED with 14,000mcd. Not even in the same ballpark of brightness.

I counted 6 fallacies in your last post, 7 if you count the values without links or even part numbers.  I use evidence and you use fallacies.  It does not look like I am wrong.  If I am, please give explanation instead of spewing fallacies.

I gave you a chance and dug through JLC library to learn more and I learned you were wrong.  I shared my findings, you replied with nothing but fallacies.  You finally tried to share some evidence though without part numbers it really is just another fallacy.  I gave you another chance and looked on digikey and once again I find that you are exaggerating. 

Digikey.ca search: LED - discrete, smt, in stock, 20mA, blue and yellow.  Min intensity: 3.2 mcd.  Max intensity: 5700 mcd

That is huge range, still smaller than your numbers because you exaggerate too much.  Odds of someone ordering 1 on either side of that range are slim.  Odds are they will pick identical packages and similar costs.

If the search is narrowed by package:
0603: 6 mcd  - 1600 mcd
0805: 4 mcd - 1990 mcd

Still a large range but odds of purchaser choosing one on either end of that spectrum are slim.  If they choose cheapest LEDs or most in stock, the range is significantly reduced.

Cheapest 10 in package 0603, cut tape:
   intensity: 63 mcd - 170 mcd
   viewing angles: 120° - 140° (2/10 not listed)

Highest stock 10 in package 0603, cut tape:
   intensity: 104 mcd - 800 mcd (8/10 are between 104 mcd and 180 mcd)
   viewing angles: 60° - 130° (8/10 are between 120° and 130°)
   
This helps explain why I found such a limited range of intensities and viewing angles when I looked on the library in question, JLC's library.

If you must continue, please give evidence instead of fallacies.

[Zero999]

Of course simply selecting the same forward current doesn't give equal brightness for two different colour LEDs, even if differences in beam angle are taken into account. Implying that it's a good rule of thumb, is just not true. As mentioned above, the radiometric efficiency depends on the chemistry and the luminous efficacy, depends on the sensitivity of the human eye to the light being produced. A classic example is a royal blue vs modern, high efficiency emerald green LED. The royal blue LED will have a better radiometric efficiency i.e. optical power out, in the visible spectrum vs electrical power in, but the emerald green LED will look brighter and have a higher luminous efficacy because the human eye is much more sensitive to that colour. It will also depend on whether the observer's eyes are dark adapted or not. In low light conditions, the peak sensitivity of the human eye shifts to shorter wavelengths, which will suppress reds and accentuate blues. In bright daylight, the reverse happens.

Going back to the original post, it might not even be desirable to have all colours equal brightness, because it will probably look unnatural. The blue and red LEDs will look much brighter than they should be and the green and yellow LEDs much dimmer. It might make more sense to make each LED emit the same optical power, or scale the optical power output, to the expected black body spectrum of an object at a certain temperature.
https://en.wikipedia.org/wiki/Black-body_radiation#Spectrum
https://upload.wikimedia.org/wikipedia/commons/1/19/Black_body.svg

[tooki]

If you must continue, please give evidence instead of fallacies.

Wow man, what a joke.

I don’t even think you know what a fallacy means. Your reply only serves to confirm what I have been saying. 

I didn’t limit myself to what jlcpcb carries, nor to only blue and orange, because your original claim didn’t, either. You made a broad general statement that simply isn’t true, and not only have I and others disproven it, now you have done the same. 

I don’t need to provide part numbers and links for my statement to be true. I gave you the source (the Reichelt catalog), and if you wanted to, you could go find it. (It is online.) I certainly am not going to go out of my way for someone who is so utterly arrogant in their position that it’s clear that NOTHING will make you realize you’re spouting bullshit.

[Kasper]

Edit to add: efficiency and viewing angle may be so different that these quick solutions do not help.
[...]
If they have similar current they should have somewhat similar brightness but you could go further...

Longer version:
Look up the amount of light output per current input and pick a current for each LED that will yield similar outputs.
[...]

Of course simply selecting the same forward current doesn't give equal brightness for two different colour LEDs, even if differences in beam angle are taken into account. Implying that it's a good rule of thumb, is just not true.
[...]

I gave a quick solution that probably would have been an improvement.  Not hard to improve on using the same resistors for different colors.  Using the words 'should' and 'somewhat' I explained it is not a great solution and went on to describe a better solution.  I even went back to clarify there may be cases where it doesn't help.  Even though it most likely would have helped with LEDs from where OP is sourcing (JLC) or even with LEDs from digikey as I have clearly shown.

[Kasper]

If you must continue, please give evidence instead of fallacies.

Wow man, what a joke.

I don’t even think you know what a fallacy means. Your reply only serves to confirm what I have been saying. 

I didn’t limit myself to what jlcpcb carries, nor to only blue and orange, because your original claim didn’t, either. You made a broad general statement that simply isn’t true, and not only have I and others disproven it, now you have done the same. 

I don’t need to provide part numbers and links for my statement to be true. I gave you the source (the Reichelt catalog), and if you wanted to, you could go find it. (It is online.) I certainly am not going to go out of my way for someone who is so utterly arrogant in their position that it’s clear that NOTHING will make you realize you’re spouting bullshit.

I politely request evidience instead of fallacies, I even said please but once again you reply with more fallacies and personal attacks 'utterly arrogant' and 'spouting bullshit' and a even little yelling 'NOTHING'.

I certainly am not going to go out of my way for someone who is so utterly arrogant in their position that it’s clear that NOTHING will make you realize you’re spouting bullshit.

You won't go out of your way to share knowledge but you are going out of your way to spew fallacies and personal attacks.  I guess we know what you are here for.

Fallacies and attacks are clearly your top priority but if you are going to call someone wrong, you should expect them to ask why.

blue and orange

blue and yellow.

[tooki]

We’ve told you why you’re wrong. You already provided evidence of this yourself. But since you won’t accept your own evidence as evidence, it’s clearly utterly pointless to go to the much greater effort to dig up datasheets and stuff.

P.S. They’re not personal attacks, just simple statements of fact.