Need creative ideas for measuring LED intensity

[Kinnza]

I think the way suggested by David Campen is by far the easiest setup, and should work fine when you just want to perform comparative measurements. Calibrating it would be difficult, you could know from it very well the rank of outputs compared, but saying what is the actual difference between each rank would be more complex. As it is easy to build, it may worth the try.

A copper piece for the end of copper tubing might do the job well. Put the LED on the open end with several layers of kapton tape for insulating thermally LED from cooper piece and it may work fine enough for your purpose (you could build a reflective piece for the open side with just the aperture for LED dome).

For measuring the way I suggested, you need a photodiode well matched with photopic response, usually found on good light meters (cheap ones usually are not well matched on the short blue range). But you could use any as far as you know accurately its response, so you can calculate its measured LER, being able to find radiometric output from the lm output. The key of all is having the SPD.

I have processed royal blue spectrums using digital camera and it works fine. When you use RAW or TIFF formats, data includes the actual response of the CCD array to each wavelength, so still when response is lower on the short blue range, it is detectable and mensurable. Just Signal to Noise ratio becomes lower, but not a problem for your purpose.

On the other hand, some years ago I was searching for photodiodes with flat response. I wanted to build a good light detector using 3 photodiodes (RGB) with flat response. I found two but not the third, so I give up the project. I did not remember if it was the Blue or the Green the one I couldnt find with a flat response, but I believe it was the G. Anyway, maybe now there are more photodiodes out there. Unfortunately, I dont have the datasheets or reference of the models I found, but you can search for it. Finding a photodiode with flat response in the B would simplify your task a lot and make it very straightforward.

 

[Kinnza]

Kinnza, I just found the lowest price spectrometer I have ever seen. Generally, I have found pricing between $1499 and $2499. This one is just $749 and it originates from Canada - of all places. If you get a minute can you have a look and see what you think.
http://www.aseq-instruments.com/index.html

Stephen Lebans

WOW! Really cheap for its performance.

I think it should work perfect for lighting measurements. We are not performing Raman spectroscopy. The spectrometer itself looks good, compared with similar ones cost 1/3.

But I miss more detailed info on the page. Clearly they are new, and want to enter the market by offering low prices. Sofware included seems little developed. Probably they will improve it with time, but for now, it has some shortcomings. Specifically, it dont have an automatic function for calibrating. Calibrating is done by introducing manually three coefficients for a 2nd order polynomial fitting. I doubt anybody does it. So actually, recalibrating it would be near impossible, still having a NIST certified calibrated lamp.

Anyway, it is a powerful spectrometer for very cheap. The initial calibration should be enough for lighting measurements, where we dont need extra accurate results. Still not being for the moment as tuned as other brand spectrometers, usually we dont need such fine tuning, and I would expect they improve the software with time. Spectral range for the standard configuration is impressive, using a very good CCD array, for 200-1100nm. Probably S/N ratio on the UV range wont be good, but at least it measures it. Configuration A, with range 300-1000nm and resolution below 3nm, S/N 300:1 looks very good for a general purpose spectrometer.

 

[Kinnza]

If I am interpreting what you are trying to do, then all you are looking at measuring is radiometric output of the LED?

1) You need some way of diffusing the light source to even out the distribution. You can make a poor mans integrating sphere with a box or sphere and some really good pure white flat paint. Even better, paint for the inside of signs ... >95% reflective and highly diffuse.

2) You need a somewhat linear light sensor. A wide bandwidth photo-diode. Do not operate it forward biased as a current source, but back-biased as a light based current modulator. It will be more linear in this mode. You may even be able to do it with a monocrystalline solar cell (just need a piece of one).

Semiman

Could a royal blue LED be used as photodiode that way?

 

[slebans]

WOW! Really cheap for its performance.

I think it should work perfect for lighting measurements. We are not performing Raman spectroscopy. The spectrometer itself looks good, compared with similar ones cost 1/3.

But I miss more detailed info on the page. Clearly they are new, and want to enter the market by offering low prices. Sofware included seems little developed. Probably they will improve it with time, but for now, it has some shortcomings. Specifically, it dont have an automatic function for calibrating. Calibrating is done by introducing manually three coefficients for a 2nd order polynomial fitting. I doubt anybody does it. So actually, recalibrating it would be near impossible, still having a NIST certified calibrated lamp.

Thanks for looking at this, Kinnza.
The software and lack of automatic calibration were the two issues that bothered me as well. Still, the price is incredible compared to what is currently on the market. I think we need to buy one and test it out.

Stephen Lebans

 

[SemiMan]

Could a royal blue LED be used as photodiode that way?

My guess, and it's only a guess is that the lack of optimization of the LED to act as a broadband photodiode would negate its use. Solar cells, especially back-contact monocrystalline solar cells have reasonably flat response across the visible range even into the blue.

Check out this paper http://tayloredge.com/reference/Electronics/Photonics/HighEfficiencySolarCells.pdf

There is a response on the right side of page 2 for the Sunpower back contact cell. The external quantum efficiency looks to be within about 5% across the visible spectrum. The bigger issue will be the lack of linearity though I believe for Sunpower cells it is not bad. Reverse biasing it will improve the response linearity.


On another note, I am thinking of asking that company for a loan of their spectrometer. Serious consideration of buying it if it works and I would be willing to write them a review. I do consulting on LED fixture design and have been putting off building a goniophotometer ... can't justify the cost of buying one. This would be a good additional piece. I only need something good enough for first pass testing .. before spending on lab testing. In theory optical simulations should be enough, but alas .... the real world like to interfere with that.


Semiman

 

[MikeAusC]

You can get a 3 colour channel + white sensor with documented response for $5 - why use undocumented sensors ?

The "Clear" channel of the TCS3200 Colour Light to Frequency Converter has a response at 350nm that's 28% of its peak at 720nm. The sensitivity rises roughly linearly from 350 to 720nm.

You also have R, G and B channels available so you can make an assessment of the spectral distribution. The colour channel to be measured is set by applying DC to two input pins. The sensor area is 1mm x 1mm,

It has internal scalers so you can set the maximum to to 600, 120 or 12kHz. The Dark output is <2Hz. Many Multimeters have a frequency measurement capability.

It draws 2mA at 5 volts.

And all of this costs . . . . . . $6 !!!!!!!

 

[SemiMan]

You can get a 3 colour channel + white sensor with documented response for $5 - why use undocumented sensors ?

-- The original poster was looking for radiometric output, not photopic output, hence a visual response sensor will not cut it.

-- On second thought though, a photodiode and/or solar cell will not work either. While the QE is very linear, the power response is not and that is what is needed.


..... back to the optically black tube ... though it is tough to make such measurements.

Semiman

 

[bshanahan14rulz]

So why not use a TEC? they even make some tecs with special coatings for absorbing energy. Saabluster mentioned it up in post 3...

 

[MikeAusC]

. . . . Problem with this is I need to take readings from both royals and various whites. Depending on composition solar cells start to get terribly non-linear around 460nm. . . . .

Why not just paint a Temperature Sensor (Glass Thermometer - Thermocouple - Semiconductor Sensor) with a non reflective coating on the side facing the LEDs and cover the other side with an insulator. Measure the rate of temperature rise.

If you want to maximise the radiant energy absorption from one direction, put the sensor at the focal point of a parabolic reflector. Metallic reflectors should have a flat spectral response.

The response time will be much faster than heating a pipe or TEC and you will have a narrow zone of absorption.

 

[MikeAusC]

Do you really want to measure Infrared radiation > 700nm - I can't imagine this helps with Reeftank lighting.

An XP-G Warm White has radiant energy at 700nm that's 28% of its radiant peak at 610nm.

One low-cost way to reduce IR sensitivity is to use dichroic filters that are used in halogen downlights. From the front they look like they're solid silvered, but if you look at the back of the reflector when the light is on, you can see redddish light.

Halogen downlights must use dichroic reflectors, otherwise a 50 watt downlight would focus 45 watts of heat on the subject !

If you remove the halogen globe and put a temperature sensor at the focal point of a 10 degree, 50 or 20mm halogen downlight, you will have a very directional radiant energy sensor with reduced IR sensitivity. The IR component will pass through the dichroic reflector.

Hopefully someone with a spectrometer can measure the response of these readily available dichroic reflectors.

 

[saabluster]

So why not use a TEC? they even make some tecs with special coatings for absorbing energy. Saabluster mentioned it up in post 3...