Howdy,

Recently one of the members of the forum mentioned that not all LEDs in a given "lot" have the same luminosity/brightness.
I have noted that fact in some of the bunches that I have bought - most bright, but some less bright.

Also based on the comments of another member, I thought I would build a little "light meter" to measure the relative luminosity of my LEDs so as to classify them into similar levels of brightness.

So I set up something like this:
5V
|
photoresistor (varies with brightness between 150Ω and 18kΩ)
|
connection to Arduino (analog pin)
|
10K resistor
|
Ground

On the Arduino I programmed a few simple lines taking the voltage input, converting it to a range between 0-100 and then displaying it on a 3 digit 7 segment LED display.
It seems to work just fine.

I'll probably upgrade to a Light to Voltage converter, but I'll buy one next time I order some other parts.

My questions for the group:
1. Do you know of any "standard" LEDs that I could use to calibrate my rudimentary "meter"?
(I understand that I don't really measure lumens, but would like to have some "standards" to use between measurement runs)

2. Are there other ways of calibrating/linearizing my rudimentary device?

I thank you in advance!

 

My questions for the group:
1. Do you know of any "standard" LEDs that I could use to calibrate my rudimentary "meter"?
(I understand that I don't really measure lumens, but would like to have some "standards" to use between measurement runs)

2. Are there other ways of calibrating/linearizing my rudimentary device?

You are using a non-linear device, so probably not. If you were to get a LDR that matched the spectral response of the human eye, you might be able to make your device more capable of measuring lumens.
See here.

Provided that you measured LEDs with the same spectral components, it would be possible to standardize them with a well-defined basis(such a thing isn't plausible), keeping all things constant and knowing the resultant output of the device would allow you to have a reference.

However, considering the spectral components of each type(indeed, each individual light) are unique, it seems nigh impossible to do this. If you were to filter all but a small portion of the sun's spectrum, you could relate light intensity, at that frequency to the sun's light, but that is a relative measurement...Basically, the best this device will do is give you relative light intensity...

 

LED binning.. google it

 

If you only want to bin them to equal brightness (relative to each other, rather than a lumen number), make a small foil tube that couples an LED to the photoresistor. You may want to use a constant current source, or connect as you plan to use them in a circuit.

Test each one, then throw them in different trays/bags based on the result number give or take a digit (or several digits, depending on quality of LED lot).

You could go further and after getting the number, point it at a clean, bright white piece of paper or card stock. Use known good warm/daylight/cool LEDs at low current to illuminate one edge of the paper. Darken the room. Then separate out "warm" (yellow tinge) "daylight" (white) and "cool" (blue tinge), for better sorting.

In the end you would have
Warm - dim ok bright
White - dim ok bright
Cool - dim ok bright

Or however many bins you want to make. If the lot is from DigiKey or Mouser, then they'll mostly go into the same one. If the lot is from China, from eBay, you might not have enough bins.

 

Are you taking into consideration the led lens? For example a 8° led will shine a brighter spot than a 120°. If you are comparing all leds of one type none of this applies.

 

Excellent comments as usual!

So I'll use my overal setup and change out the sensor to see what works best.
I probably can readjust my upper and lower limits of sensitivity to be able to spread out the "brightness" to at least classify them into 4-5 groups.

"LED binning" is very interesting and something I'll study a bit more on.

Consideration of the lens is a good point. Clearly an SMD device needs to be evaluated differently than a T-1.75 device.

I've just bought a variety of LED smd devices, so I'll be able to evaluate those too.

I thank you all very much for your input!

 

I wanted to update this with some work I've been doing that might help other newbies.
Jameco had a sale on SMD LEDs a few months ago and so I bough a few different kinds for a few bucks.
Based on the responses to this thread, and a few others I've read, I thought I'd experiment a bit and subjected 8 different SMD LEDs to variations in V and I and got the following:

Voltage/Current relationships
0805 1 - Unicolor Red 660 nm
On___1.4V .02mA
Nice__1.5V .67mA
Bright 1.6V 1.0 mA

0805 2 - White superthin
On___2.4V 0.01 mA
Nice__2.7V 0.37 mA
Bright 2.8V 1.3 mA

0805 3 - Amber 610 nm
On___1.6V 0.12 mA
Nice__2.1V 14 mA
Bright 2.8V 40 mA

0805 4 - Red 660 nm
On___1.4V 0.02 mA
Nice__1.8V 14 mA
Bright 1.9V 15 mA

0805 5 - Yellow clear
On___1.6V 0.05 mA
Nice__2v 8 mA
Bright 3V 50 mA

0805 6 - Amber yellow 595 nm
On___1.5V 0.05 mA
Nice__1.9V 7 mA
Bright 2.2V 20 mA

0805 7 - Ultrabright white 600mcd
On___2.4V 0.01 mA
Nice__2.6V 0.5 mA
Bright 3V 20 mA

1206 1 - Green clear
On___2.1V 0.01mA
Nice__2.4V 0.14 mA
Bright 2.6V 3 mA

"On" is when you can barely see the light.
"Nice" is that scientific term that means "I can see it easily, and it doesn't hurt"
"Bright" is another scientific term that means "bright".

1. What is compelling is the amazing variation in voltage/current requirements between LEDs.
2. Newbies - As I've learned form the experts here, remember that
"EVERY COMPONENT IN EVERY CIRCUIT NEEDS TO BE ANALYZED IN THE CONTEXT OF THAT CIRCUIT"
- even the lowly LED.

I thank all the experts who have made this possible.

 

I applaud your efforts!

Things like this (and newbies reading things like this) go a long way toward developing a feel for what is, and what is not reasonable, to expect in circuits.

I wish your data had another sig fig on the voltage measurements. It would be interesting to estimate the "n" values for these LEDs and see how much variability there is in that parameter.

 

Thanks very much!
This forum has been very enlightening because of all of you experts.

1. I have a cheap power supply bought on Amazon that I did this with - Sometimes when you are adjusting the voltage you can't even get to to sit on the first decimal point!
If I were to upgrade to a "good" power supply would the readings be accurate enough, or should I use external volt and ammeters?

2. "n" values - intriguing - what are they? - couldn't find 'em with Google.

3. I'll play some more and probably post some additions to this thread. But my next LED project is figuring out the "constant current" drivers, and how I might use them (Most of my LEDs are for flashing in a circuit so I can see what's going on).


Thanks again!

 

Thanks very much!
This forum has been very enlightening because of all of you experts.

1. I have a cheap power supply bought on Amazon that I did this with - Sometimes when you are adjusting the voltage you can't even get to to sit on the first decimal point!
If I were to upgrade to a "good" power supply would the readings be accurate enough, or should I use external volt and ammeters?

2. "n" values - intriguing - what are they? - couldn't find 'em with Google.

3. I'll play some more and probably post some additions to this thread. But my next LED project is figuring out the "constant current" drivers, and how I might use them (Most of my LEDs are for flashing in a circuit so I can see what's going on).


Thanks again!

Whenever possible, you want to use appropriate meters to make measurements.

The equation that describes the current in a diode or BJT as a function of the control voltage (diode voltage for a diode and base-emitter junction for a transistor) is exponential. The denominator of the exponent is proportional to the thermal voltage. The proportionality constant, 'n', is often called the "ideality factor" and it is generally between 1 and 2. One rule of thumb use to claim that it was generally safe to assume it was 1 for a transistor and 2 for a diode, but I think that most modern diodes are generally pretty close to 1 as well. But this is small signal diodes, not power diodes, laser diodes, or light emitting diodes -- I have no feel at all for what the n-value for those devices typically are.

At room temperature, an n-value of 1 means that the current will increase by a factor of 10 for every ~60mV increase in control voltage.

 

Ok.
So I took an ultrabright white LED, SMD, 2.8V, 20mA.
Connected to the powersupply and started turning up the voltage in increments of 0.1V - 1 minute for each increment of 0.1V (to take away any change in current as a function of heating the LED).

From 2.4V (when I barely could see it light up) to 15V - the relationship is perfectly linear - (There's a small little ditzel around 20mA, but that probably is my problem).

At 2.4V the current was 0.01mA and at 15V it was 35.7mA.
Slope = (15/(35-2.4)) = 0.46.
Which doesn't sound anything like the "n".
What am I doing wrong?

 

From 2.4V (when I barely could see it light up) to 15V - the relationship is perfectly linear - (There's a small little ditzel around 20mA, but that probably is my problem).

At 2.4V the current was 0.01mA and at 15V it was 35.7mA.
Slope = (15/(35-2.4)) = 0.46.
Which doesn't sound anything like the "n".
What am I doing wrong?

Was there a 380Ω resistor in the loop? Was current limiting enabled on the power supply? If it was a standard 20mA bright SMD Diode, it should have turned to magic smoke by 5V, and appeared close to a dead short at 15V.

Were you using good DMMs to measure both voltage across LED and current through LED at the same time?

Repeat that same experiment with a 10Ω Resistor and post how your power supply responds, using the resistor EXACTLY as you used the LED. If it is different, is there some sort of small resistance integrated in the LED to reduce overcurrent issues?

 

1. No resistor, I assumed you'd not want current limitation to see how the LED responded to "pure" voltage and current.
(If I toast the LED, no big deal - I learn something, and I replace the LED).

2. Had the current limitation knob cranked up all the way, so no limitation - not a great power supply, but OK

3. Hmmmm - was working great at 15V and 35mA - hurt my eyes.

4. I have lame measuring devices - I just used the Volts off of the DC power supply, and I used my DVM810 for current. But I'm pretty sure they're not horribly far off.

5. HERE's the LED - I don't think there is anything hidden.

WAIT! WAIT! I'll redo it with the current limitation TOTALLY OFF!!!
For some reason I had it 1/2 way...stay tuned...story continues in the morning!

 

The saga continues and we have - Success!!!!
Nice curve and I didn't toast the LED! Hooray!

1. I'm not sure about the little "divot" in the curve - whether it is part of the nature of the semiconductors,
or error of my measurement - probably the latter.
Confirmed - my error - based on THIS. I had to change "range" on my meter.

2. The "+"s are the "brightness" - arbitrary by my eye. It actually started to give off light prior to the 0.01 mA - so obviously my ammeter isn't sensitive enough. The "++++" means "It hurts my eyes to look at it" - and it did go to +++++.

3. I noted something interesting where the arrow is. I always waited for the current to stabilize before I recorded the number. Around the arrow I noted that the current started "high", then slowly went DOWN. I wonder if that is temperature effect.

4. Slopes -
0-3V - 9.1/0.7 = 13 mA/V
3-3.3V - 29.9/0.3 = 99.7 mA/V
3.4-3.7 - 50.4/0.3 = 168 mA/V

5. I played around with the "ideal diode equation" and couldn't get anything reasonable as an output - but at least I got a curve.

Thanks for all your help!

 

If I were to upgrade to a "good" power supply would the readings be accurate enough, or should I use external volt and ammeters?

If you're looking for any kind of accuracy its always best to use external meters for voltage and current rather than depending on panel meters on power supplies. Usually these meters are for reference only and are not calibrated.

 

The saga continues and we have - Success!!!!
Nice curve and I didn't toast the LED! Hooray!

1. I'm not sure about the little "divot" in the curve - whether it is part of the nature of the semiconductors,
or error of my measurement - probably the latter.
Confirmed - my error - based on THIS. I had to change "range" on my meter.

2. The "+"s are the "brightness" - arbitrary by my eye. It actually started to give off light prior to the 0.01 mA - so obviously my ammeter isn't sensitive enough. The "++++" means "It hurts my eyes to look at it" - and it did go to +++++.

3. I noted something interesting where the arrow is. I always waited for the current to stabilize before I recorded the number. Around the arrow I noted that the current started "high", then slowly went DOWN. I wonder if that is temperature effect.

4. Slopes -
0-3V - 9.1/0.7 = 13 mA/V
3-3.3V - 29.9/0.3 = 99.7 mA/V
3.4-3.7 - 50.4/0.3 = 168 mA/V

5. I played around with the "ideal diode equation" and couldn't get anything reasonable as an output - but at least I got a curve.

Thanks for all your help!

From your graph and perceived brightness, I'd say don't run them above 2.8V/~15mA, if the plus signs are from your light meter, then all that is happening past 15mA is more UV is hitting the already excited phosphors (white and some blue LEDs), not getting brighter. If this were a Green LED, you'd see it differently.

The only thing >15mA will accomplish would be eating away at the liftespan of phosphors faster and adding a good deal more heat stress to the LED die itself.

 

From your graph and perceived brightness, I'd say don't run them above 2.8V/~15mA, if the plus signs are from your light meter, then all that is happening past 15mA is more UV is hitting the already excited phosphors (white and some blue LEDs), not getting brighter. If this were a Green LED, you'd see it differently.

The only thing >15mA will accomplish would be eating away at the liftespan of phosphors faster and adding a good deal more heat stress to the LED die itself.

Very good. Thank you.
So I'll probably do something similar when I use LEDs in all my projects.