PAR / light spectrum sensor

Hi all,

I’m looking to buy or build a PAR (Photosynthetically active radiation) sensor, but my googling and other searching has come short.

This is basically a light spectrum analyser in the 400-700 nm range, measuring the amount of energy the incoming light has that is useful for photosynthesis if plants. Plants use the light on both ends of the spectrum, but not the middle half.

Sunlight is of course full spectrum and a simple lux meter does the job well there, but LED lights (especially dedicated growing lights) are not, and that’s where the PAR comes in play. The PAR of such lights can easily be 2-3 times that of sunlight with the same brightness when measured with a regular lux meter.

There exist commercial sensors, but those are mighty expensive.

I was thinking maybe a CCD array (as found in cameras - after all they are sensitive to colours) with some software smarts could do the job, but no references to this.

Any ideas on what kind of sensor could do this? What measurement principles can be used? How is it done in commercially available sensors?

Do you mean a colour sensor without using the green channel. Leo..

Excellent, that could just be it! I just had a quick look at the data sheet, it seems the different colours do need some adjustment for sensitivity but that's rather trivial to do. It won't be very accurate but even with a 10-20% error I'd be happy, considering the price.

Mmm... The problem is indeed really really hard. Based on this Adafruit sensor I did some more searching and ran into some possible solutions.

Measuring just RGB is apparently not enough to be able to calculate PAR for light sources such as LEDs, which is what I mostly use (in case of sunlight there should be a direct relationship to the overall brightness, as the spectrum is known).

Then there's this solution: using a VTB8441BH photo sensor, but again with a few colour filters so that one will have the same problem.

Finally I ran into whose spectruino sounds very interesting. It basically uses a diffraction grate to separate the wavelengths and a CCD to take the image of it. It may need a but more power than an Arduino or ESP8266 to process, though.

Anyone else tried this, either successfully or not?

The otherwise worthless sheet describing the Spectruino Three states that it will produce a spectrum of light intensity versus wavelength at 10 nm resolution, and the manufacturer claims it to be Arduino-based. But at $415, it would not be considered cheap by most. What more do you need?

There are a couple of DIY projects to create a spectrometer at very low cost from a webcam and a piece of replica diffraction grating, or a DVD, but calibration is an issue because webcams have built in autoexposure controls that are difficult or impossible to defeat.

Hi, wvmarle. I hope you are still watching this topic.

I would also like making a spectrometer to investigate LED's and other lights. My idea is using a slit, a lens, a diffraction grating, another lens, a linear sensor and Arduino (Atmega 328p) with a color graphic display. A friend of mine made me in a few minutes a sketch of optics with all calculations. I bought some gratings for a few Euros, the two lenses (70€), then I've been busy. Now, if I have some time, I'd like to proceed. The sensor could be a TSL1401CL.

Currently working with the AS7265x array. 18 channels; planning to add a UV A/B sensor for total 20 channels.
Working quite well; some oddities that have to be investigated though. I don’t fully understand the workings of the sensors yet.

Uhm… I think you can’t get a good resolution if you don’t use a diffraction grating.

It depends on the resolution you're after. With sufficient grating of course you can get 1 nm resolution or even better, but do you really need that? This 18-channel sensor offers some 20-30 nm resolution. Pretty good. Probably good enough, as I'm not trying to get excitation spectra of elements but emission spectra of lamps.