`DIY` spectrometer first light!

[Gasman]

Hi guys

I`ve been trying something today which might be of interest to others thinking of having a go at spectroscopy!

In December I bought Jeff Hopkin`s excellent new book from Springer `Using Commercial Amateur Astronomical Spectrographs` mainly as it had a chapter devoted to the Alpy 600 which I had on order and also another another chapter which is practically a handbook on Rspec.

Chapter 4 of the book describes a `DIY` spectrometer based on a reconditioned medical spectrometer which includes the optical bench including 1800l/mm grating, 5v power supply (USA) fibre optic cable , 9pin to 9 pin serial lead,ccd detector and a really good software package called `Spectrum studio`. Its really not correct to call it  DIY  as it is a self contained unit only needing a computer and a means of getting the image onto the end of the fibre.  Jeff suggests that these units are available on ebay and lo and behold that very evening they were on sale from the States for not much more than the price of a Star analyser, £122 but the £70 (p+p,import vat) adds a bit more to the price so I ordered one and it arrived today!.

Out of the box

I had to get a few bits, UK power supply & usb to serial converter but the software installed easily and the spectrometer detected and the fibre optic cable just pointed at the window

Fisrt turn on

ponted to a desk lamp

and also an energy saving lamp which seems to id some lines ??

No calibration has been carried out yet so I`ve no idea how accurate these units are although they say a resolution of 2nm is possible with the correct calibration. These units are sold as NOT calibrated indeed they say that aligning these optics is NOT trivial and they point to a pdf on their web site explaining the process here -

http://www.science-surplus.com/products/spectrometers/spectrometer-resources

The unit comes with an 1800 l/mm grating which apparantly gives a 200 nm window although the whole visible spectrum can be covered by moving the 200nm window which is done by `tweaking the various optics inside the unit. A more expensive unit with a 900l/mm  claims to cover the whole visible  spectrum without this tweaking and an even more expensive unit is available with the optics already aligned, my unit seems to be pretty well aligned but I need a 0.05" allen key to get to the optics.

I intend to attach it to my ED80 scope with a flip mirror, the supplied SMA type fibre optic cable is only 19" long so a bit short given the sweep of the scope on the mount but  longer ones seem available.

I`ve not had the unit working long so will report back as I find out more about it but considering what you get for the money it seems an excellent addition for the amateur. My thanks go to Jeff for introducing me to it

Cheers

Steve

[nytecam]

Thanks Steve - interesting write-up - keep us posted with your results via your 80mm OG - it will be interesting to see how faint it will go

[Merlin66]

Steve,

there are a few members over on the astronomical spectroscopy group who have been "playing" with these Kits for the past few years.

I'm sure they can give you a heads up on issues (and limitations)

[Cath]

Looking at the spectrum captures you have it looks very much like it's all shifted in wavelength by quite a lot. It all seems to start around 700nm, where as I'd have thought your samples should have started much shorter in wavelength, probably around 300 to 400nm.

I don't know if the hardware needs aligning or the software needs setting up. Most probably the hardware needs adjusting/aligning.

[Gasman]

Hi folks

Cath yes I gather the calibration reqs 4 coefficients worked out using least squares from a minimum of 4 known wavelengths!. I'm just working out how to do it in Excel using a command called 'Linest' something I've not tried before! All part of the curve ;-)

Steve

[tcpip]

Hi Steve,

How are things going with the spectrometer? I've ordered ine today...

[Gasman]

Hi Irving

I mounted my fibre optic cable on the back of a flip mirror using an old crappy eyepiece but using an illuminated micrometric eyepiece for positioning. I`m on hols at present but can post a pic when I get back !.

cheers

Steve

[athornett]

Hi Steve

I have recently purchased one of the DIY Spectrometers from Science Surplus and am now trying to get it to work with my telescope. Came across your thread here. Noted you said that you "mounted my fibre optic cable on the back of a flip mirror using an old crappy eyepiece but using an illuminated micrometric eyepiece for positioning" on your last post 12/9/2014 - but no further posts - how did you get on? Any advice you can give me?

I have posted my own experiences so far of using this device on my astronomy club website @ https://roslistonastronomy.uk/category/spectroscopy/diy-spectrometer

Andy

[Saibot]

To revive this thread, I also got one unit and first captured spectra of Vega.
Next step was to build a beam splitter to have some light available for positioning the telescope. Inspired by this presentation by Jerry Hubbell

https://www.raclub.org/Documents/Programs/Engineering-a-fiber-fed-spectrometer-for-astronomical-use.pdf

I built my own cold mirror setup with slightly different components. You can see my setup and parts list here:

https://quantumdynamics.wordpress.com/2019/01/11/let-the-starshine-in-the-fiber/

Also two spectra I acquired are shown (Vega and Mirach). With my Skywatcher Heritage Newton 130/650 and from the polluted sky around here I estimate magnitude 2.5 or similar is the limit for stellar spectroscopy (60 s exposure).

Some caveats:

-the repurposed flip mirror requires a back focus of about 7-8 cm. I can do this since I have a Newton (130mm) with adjustable tube length
-I wrote my own software to collect the spectral data (serial port protocol, lots of timings/buffer sizes to get right), and at the same time to show the CCD image from the CCD superimposed
-many dry runs required to get everything working and aligned, around xmas distant xmas LEDs are a good target to identify LED colors and check the focusing of spectrometer and at the same time the CCD image

I just changed the alignment of the spectrometer. As shipped, the window started  in the UV range and ended in orange colors, I red-shifted it. This is a very delicate business and I am not entirely satisfied with my current setup, since it lost intensity compared to the original alignment. However I want to do lunar spectroscopy (in particular of the upcoming eclipse) and for this it should work.
The spectrometer instructions focus on how to get the line shapes and spectral window adjusted, but less about how to maximize intensity (which is important for starlight of course). Anybody with more comments about the internal alignment of the Science Surplus Spectrometer besides the instructions?

The attached image shows Vega, the red graph is the pickles reference. This was with the original alignment and without the cold-mirror.

Best, Tobias 

[Gasman]

Nice one Tobias

You have got much further than I did . 

Andy pm'd me back in August hence no posts here. I was struggling with getting the star on the slit if I remember correctly but it was the same time as the Alpy 600 was launched and I got side tracked with that with the intention of going back to the surplus spectroscope.

Your work might inspire me to give it another go ?.

Regards

Steve

[robin_astro]

4 hours ago, Saibot said:

With my Skywatcher Heritage Newton 130/650 and from the polluted sky around here I estimate magnitude 2.5 or similar is the limit for stellar spectroscopy (60 s exposure).

Hi Tobias,

Congratulations on your spectra !

You are definitely losing a lot of light if your limiting magnitude is 2.5.  For comparison a slit spectrograph would produce a similar resolution spectrum in a  second or two and would potentially reach around mag 12 with your equipment.  (Light pollution is generally not a  limiting factor for slit/fibre spectroscopy, except for very faint objects)

Designing and building an efficient fibre optic feed is difficult which is why most amateur spectroscopists use telescope mounted slit spectrographs.  I suspect you are losing most light transferring light from the telescope to the fibre and then matching the light from the fibre to the spectrograph.  The CAOS website has some useful information

https://spectroscopy.wordpress.com/fibres/

Cheers

Robin

[robin_astro]

1 hour ago, robin_astro said:

(Light pollution is generally not a  limiting factor for slit/fibre spectroscopy, except for very faint objects)

Actually, thinking about it, depending on the diameter of your fibre, you could be introducing a lot of unwanted sky background into the fibre.  With a fibre optic feed, it is important to match the size of the fibre to the size of the star image otherwise you potentially lose resolution and include contamination from sky background into the measurement which cannot easily be removed.This is different from a slit spectrograph where you can select the width of the slit to match the diameter of the star image and subtract the sky background based on measurements in the slit above and below the star.  I dont know what diameter  fibre you are using but the narrowest practical fibres I have seen are 50um.  The star image from  your telescope though would only be perhaps around 15 um diameter in typical seeing so even with a 50um fibre the area of sky collected would be ~10 larger than the star.  This is one reason why fibre fed spectrographs tend  to be used with larger telescopes.

Robin

[robin_astro]

14 minutes ago, robin_astro said:

I dont know what diameter  fibre you are using but the narrowest practical fibres I have seen are 50um.

I see from Hubbell's presentation he is suggesting using a 200um diameter fibre.  While it may help getting the star on the end of the fibre, this is a very poor match to an amateur size telescope.  You will be collecting light from an area about 200 times larger than the star which in high light pollution areas will be a disaster. 

These surplus spectrographs are lab instruments and perform very poorly as astronomical spectrographs. If you want to learn about amateur spectroscopy, there are much better sources of information out there. For example I can recommend Christian Buil's website as a good starting point.

http://www.astrosurf.com/buil/index.html

He is arguably the worlds leading amateur spectroscopist and observes from a very light polluted location

Robin

[Saibot]

Thanks Robin for the comments and suggestions! Indeed the background "noise" is an issue with additional sky light entering the fiber. The supplied fiber with the Science Surplus Spectrometer is not a bad choice, another fiber I bough induces considerably more losses. I checked the focusing by putting tracing paper on the fiber location for verifying a focused image.

I compromised on the equipment cost since I can only put my small telescope on a tiny balcony in downtown and see wanted to keep the budget below USD 500 (so far about USD 250 for the spectrometer and another 250 USD for the remaining parts: flip+cold mirror, fiber adapter). This is at a fraction of the costs of the more advanced setups, but it already allows me to study the brighter light sources in the sky, including the moon - and to become familiar with the spectroscopy setup and calibration. For point sources the star analyzer might provide a much cheaper and simpler solution.
Still have to see how much further I can get with the current setup... 
-Tobias

[robin_astro]

Hi Tobias,

Unfortunately even with the modular ALPY, the next step up from the Star Analyser is still a big one both in cost and complexity.   If you are looking for a low cost next step beyond the Star Analyser, you have access to a 3D printer and don't mind some DIY then you might like to consider the LOWSPEC design by Paul Gerlach.

https://www.thingiverse.com/thing:2455390

This is an excellent design and would be a much better spectrograph for your setup. It could be build for around a third the cost of a similar commercial model.  It uses off the shelf optics so  performance is slightly compromised compared with commercial equivalents but still has very good performance, includes a proper mirror slit guider which is easier to use than a beam splitter, can use different gratings to give a wide range of resolutions from R~500 to ~10000  and  will work far better than what you are trying currently.  Several people have built one and Paul can be found  giving updates on his design on the astronomical spectroscopy Yahoo group, for example here

https://groups.yahoo.com/neo/groups/astronomical_spectroscopy/conversations/messages/14661

Cheers

Robin

[robin_astro]

6 hours ago, Saibot said:

I compromised on the equipment cost since I can only put my small telescope on a tiny balcony in downtown

Don't  be discouraged by your location. I am lucky to live in a dark rural location but spectroscopy can be done in very light polluted locations compared with other astronomy.  This is where Christian Buil does most of his spectroscopy.

http://www.astrosurf.com/buil/pollution/measure.htm

and Valerie Desnoux, the author of the famous spectroscopy data reduction program, Visual Spec  observes from the centre of Paris less than a km from the Eifel Tower

http://valerie.desnoux.free.fr/paris saint-charles/Paris_obs.html

Cheers

Robin

[robin_astro]

You can also read the full story of the development of the LOWSPEC and the latest updates on the VdS spectroscopy forum here

https://forum.vdsastro.de/viewtopic.php?f=28&t=4432

Robin

[Merlin66]

Recent developments with Paul's LOWSPEC and the latest guide module are discussed on the astronomical spectroscopy forum.

https://groups.yahoo.com/neo/groups/astronomical_spectroscopy/conversations/messages/14693

[Saibot]

I took another image yesterday of M42 (Orion nebula) and also of Mirach (Andromeda) to test the fiber spectrometer in combination with the cold mirror  (and across the moon, but this requires more processing). In principle the equipment is working fine, I needed to adapt my programming for the  readout of the spectrometer after noticing that it takes 1-2 attempts to change the integration time (just to issue a "record spectrum once with 60s" command is not enough. Also as you can see I adjusted the grating to cover blue - red colors. I believe one advantage of the fiber digital spectrometer is that you see the spectrum live on the screen. The post processing is in principle easier, I just subtract a "dark" frame, no need to rotate images etc etc. For the moon there is enough light to cut down the integration time to 100 ms  and this should enable me to study the lunar reflectance quite nicely. Overall I can say it works for me, but it requires to code your own data acquisition and processing pipeline, besides the "output" coupler. 

[Saibot]

Today I took a sequence of spectra across the moon, so far I only extracted the ratio of blueish/reddish color (I averaged across about 40nm around each color), but will look into more detailed analysis to cross correlate for instance with the Clementine mission data. For me it shows that the setup has some spatial resolution on the moon and is working ok! Note the blueish crater on the left side: fresh material). For each spectrum I take also an image with the cold mirror CCD camera (PointGrey Chameleon Mono, USB2), which I then used to determine the location wrt to the spectra taken (I switched off the guiding during the image scan). The Orion nebula M42 image shown in the previous post shows a resolution (full width at half maximum) of around 1nm for the setup.

[robin_astro]

Hi Tobias,

Are you planning to use use  a solar analogue to divide the lunar spectrum by to calculate the lunar reflectance spectrum and how are you correcting for atmospheric extinction?

Cheers

Robin

[robin_astro]

The M42 spectrum is impressively clean considering your light pollution. Did you subtract a sky background spectrum ?

Cheers

Robin

[Saibot]

Hi Robin, thanks for your input. For M42 or stars I subtract a spectrum of a nearby sky region taken with the same integration time (60 s). For the moon normalization/response is more tricky. In the literature I find that people sometimes normalized to specific areas on the moon (i.e. Sea of Serenity) in "Lunar Spectral Types" by McCord et al Journal of Geophysical Research, vol 77, p. 1349 (1972). Another option is in principle to construct a sky background model around the moon, or to just obtain the spectral slope with respect to some "average". I am still checking out these different methods. Comparisons can be done to some extent with the Clementine mission data (just 2 points in the visible), or the Moon Mineralogy Mapper (M3) spectral cubes (I still have to see how to process this data better). Cheers, Tobias