Question: "Full Spectrum" grow LED Testing

soheilkh is asking a question about general
Follow this topic

by soheilkh | December 11, 2016 01:12 | #13766

What I want to do or know

I'm testing a "Full Spectrum" grow LED with way more red LEDs than blue but I get a very strong blue which is stronger than red. I'm trying to find out why am I having this problem.

Background story

I'm using a customized spectrometer I made from scratch which is very similar to the "Desktop Spectrometry Kit" as for the housing and angles and grating (DVD). The camera I'm using is Microsoft LifeCam Cinema 720p Web Camera as my camera. I taped the DVD grating to the front of the camera. To clean the DVD I used dishwasher soap to clean the residue of the DVD blue hue. I calibrated my spectrometer with CFL light which seems to be accurate.

I used a white A4 paper that I put on top of the slit to overcome the overcompensation of the strong LED even without it the blue seems stronger. Any one know why or how can I

The LED has 240 epileds in total with the following colors: Red 168pcs, Blue 48pcs White 16pcs UV4pcs IR4pcs



Do you have some reason to think that the spectral output of the lamp should be different? Maybe the result you got is a good representation of what the lamp produces.

The webcam you are using probably uses automatic white balance, so the relative intensity of the red and blue channels is adjusted based on an algorithm designed to make natural looking photos in different light environments. So it is not easy to know how much your spectral result is skewed from reality. But automatic white balance probably does not shift the color very much so the result should be similar to reality.

Is it possible that the paper you are shining the light through changes the color balance?


Is this a question? Click here to post it to the Questions page.

The reason I was doubted the result is that I have way more red LEDs than blue. Both colors seems (visibly to eyes) to have comparable brightness, so I don't know how can the camera picks up more blue than red.

As for the white balance I couldn't figure out a way yo change it since none of the browsers (Chrome, Firefox) use allow me to change the white balance setting for the input webcam. I know of the softwares such as Amcap that I can play with settings such as white balance. I might be able to take a picture in there and then upload and compare the result with the one that browser got from the cam directly.

There is no way I can change the white balance to see the difference in the color. The paper is just a plain A4 white paper so I doubt that is the cause unless it is better at filtering red than blue.

Thank you Chris,

The raw spectra these DIY devices produce are not absolute calibrated per-color, so they may be more sensitive to blue than red. You could either independently confirm the ratio of red/blue using an absolute method, or use a known light source to absolute-calibrate your camera. There's some about this process here: #gain-correction -- but it's not so easy to follow, and it'd be great to get a clearer guide published sometime.

Thanks Warren, (1) is there any specific filter that I can use to fix overcompensation of the light in your experience? maybe that will help in increasing color accuracy (like a filter that is known to subdues all colors equally? so I can double or triple fold it depending on the overexposure) (2) how about removing IR filer? what effect could that have on my color issue (other than being able to see more IR and/or UV light)

Is this a question? Click here to post it to the Questions page.

Well, the compensation is based on the sensitivity of the semiconductor, mostly (i believe). I think you could try to block the wavelengths that are over-represented, but it might be hard to find a filter that was the exact inverse to the sensitivity of the camera. It's a pretty interesting idea, in theory, though.

But do you mean overexposure in general? That may be a separate question -- just an excess of brightness overall, not inconsistent response on a per-color basis.

Is this a question? Click here to post it to the Questions page.

Reply to this comment...

Log in to comment