The issues mentioned in this URL could certainly be an issue (i.e. for your spectrum example) if the grating is not correctly aligned with the pixels of the sensor.

To properly answer your question, you also need to account for optical effects (i.e. lens and distance to sensor).

What is confusing about this link is that they claim to use monochrome mode. A true monochrome camera has pixels "larger" sensitive to a broader range of the spectrum, rather than adjacent pixels sensitive to R,G and B. Generally, monochrome on RGB is simply a combination.

Hi there! Did anybody sorted out the "wobble" problem? I have finally finished my own spectrometer after a few months of hard work and this kind of problem is bothering me. I tought it is some kind of moire but I can't sort out yet. My spectrometer is using a linear CCD with 3680 pixels and a quality 300 lines/mm transmission grating. Here are some pics:Home built linear CCD spectrometer

No, haven't had a chance... still looking a bit moire-ed

https://spectralworkbench.org/spectra/show/900

I was going to try putting a diffuser in the beam path. If you make any progress i'd love to see!

Wow, you have quite an impressive setup! Would you be interested in open sourcing it by posting at http://publiclaboratory.org/note/add ?

Hi Warren, sorry for posting so late but I tought my comment will not be published...:)

Your wobble is acceptable compared to mine: some spectra

I have found some info on this amateur astronomers site: CCD fringing 1 and CCD fringing 2. I am afraid my problem resides in the CCD protective glass. That's it, the devil is in details, despite I am using the same ccd as some major spectrometer manufacturers.

Yes, I was thinking to make it open source despite my project is a commercial one. As soon I will sort out the annoying interference fringes and finish the firmware I will make it available. Anyhow it was way too much work for one person and improvements from third party is welcome. Saddly the other open source spectrometer projects are not available. I mean Spectruino and openspectrometer

Regarding total cost I am now at around 200 USD. Obviously electronics are the cheaper part, the CCD was 25USD, ADC and Chipkit board another 40USD I guess. The grating was expensive, 105USD from Edmund optics. First surface mirror and collimating/focusing lens was recovered from a scraped Minolta color copier. But I think one can get a good spectrometer and more important it gets access to the pure raw data from sensor unlike in webcams.

Hi Jeff.

I found that the "wobbling" problem is in fact a nasty interferrence pattern which arise in the protective glass which covers the ccd/cmos chip. That's it, it seems that proffesional spectrometer suppliers are changing that glass with AR coated glass on both surfaces. So the small interference pattern you noticed in your webcam it's ok. So now I am stucked with this problem and struggling to subtract this signal by some sort of algorithm.

huh, interesting. It's the glass reflection interfering with what? between the two surfaces? A friend suggested that its between the width of the slit and something about the sensor. I wonder if varying the slit width can remove it?

It seems that reflections between the air/glass/air/silicon boundaries adds constructively and destructively to make the interference fringes. Also is dependant of glass thickness. The fringe spacing is decreasing toward the red (NIR) edge. After all this chips are manufactured for imaging white light. Surfing the web I noticed that amateur and also professional astronomers have encountered this kind of problem when trying to build custom made spectrometers with comercial available ccd chips. Spectrometer manufacturers are changing that protective glass with some kind of wedged glass (not parallel surfaces) or AR coated glass. Obviously removing the protective glass is very difficult and risky operation. I also suspected that is some kind of slit diffraction projected onto the chip and I tried with several slit spacing from 50 microns to around 500 microns. The fringing was still there with exact spacing increasing toward red, only the amplitude was lower perhaps because of losing resolution with increasing slit width. Not to mention that I have changed the lenses, mirrors, diffractions gratings (glass 300 l/mm, plastic foil 500 l/mm, plastic foil 1000 l/mm) without any success. So that's it I am struggling now with math to find a way to supress the fringes without loosing signal information. By the way here is a simple method to calculate the slit width using only a laser pointer and a ruler slit width calculator

Oops! Sorry, the slit width calculator link is here Is in French but is easy to understand.

Looking back through the history to catch up while anxiously waiting to get my kit I see you have come across the Fabry-Perot interferometer effect caused by the reflections between the array and the cover glass. In the Ocean Optics spectrometers we had them remove the cover glass from the Sharp fax machine arrays they used and replace them with AR glass, but the black body curve still showed the waves I see in the sky plots some users have posted.

The only way to eliminate the effect is to show the spectrometer an approximate black body source with known spectral radiance characteristics and apply a correction to all subsequent images. The tungsten lamp calibration technique I suggested previously would help but I have no idea how much. If you show it a tungsten lamp and tell it what we know about the tungsten emissivity characteristics it will most likely be better than it is, and anyone can do it for the cost of a light bulb.

http://eprints.lancs.ac.uk/6736/1/inproc_326.pdf

Great idea -- if you can write out a sequence of steps for the tungsten correction method, maybe we can write a macro around it? Tom_H was also thinking of an intensity calibration method using a black body source, so perhaps the two could be combined into one process... see if you can find him on the mailing list!

pretty straightforward.

1)generate a file with the emissivity data of tungsten from the table in the attached file or the spectral radiance data of the light source.

2)show the spectrometer that light source

3)create a correction file that makes the display look like the file in step 1

4)apply that file thereafter to all readings

I was skeptical at first but the correction holds extremely well. I temperature regulated our optical bench by gluing a resistor on with epoxy and using pwm to regulate it, that might be necessary for optimum maintenance of the correction, we'll see.

When I first looked at a black body source with our first ocean optics unit I had no idea what was going on. It looked like a tsunami. Then I learned nobody was really using them to measure light. The learning curve began. I am thrilled to have found a path to share the knowledge gained.