The Otter DIY Raman Spectrometer

Another DIY Raman Spectrometer.

Similar projects worth following
Raman spectroscopy offers vibrational and rotational information about molecules, so it's a valuable tool for characterization.On top of that it's "easy" to build a Raman spectrometer. Well, easier than an IR spectrometer, which offer some of the same information - though with complementary selection rules.

The information presented here is a summary of my blog about constructing a Raman spectrometer:

It's still a project very much under construction, so don't go off and buy parts for the spectrometer unless you're willing to experiment.

The Otter DIY Raman spectrometer inhabits two boxes:

  1. A box for the laser and the sample
  2. Another box for the spectrograph

Raman scattered radiation has an intensity of 10⁻⁷ compared to Raleigh scattered light and even less of the laser of course, so to avoid the CCD completely drowning in laser light, the two are kept separate.

1. The Raman excitation setup

The Raman excitation and collection is the usual back-scattering setup, and the schematic looks like this:

The laser is reflected off of the 540 nm dichroic long pass mirror. It's focused onto the sample by a microscope objective, which also serves to collect and collimate the scattered radiation. The long pass dichroic mirror and the 540 nm edge filter transmits the Stokes-shifted Raman radiation, while rejecting the Raleigh scattered light (and the anti-Stokes Raman signals). Finally the planoconvex lens focuses the Raman radiation onto an optic fiber going into the spectrograph.

This setup is more or less a copy of many commercial Raman probes, and almost identical to the one presented in Mohr's article (J. Chem. Educ., 2010, 87 (3), pp 326–330).

The diffracted light of interest is from 278-4000 cm⁻¹ in the Raman spectrum, which with an excitation wavelength of 532 nm translates to 540.0-675.6 nm, because:

2. The spectrograph

The spectrograph is a two off-axis parabolic mirror design proposed and analyzed by Gil and Simon (Appl. Optics, 1983, 22 (1), pp 152-158) and first realized by Schieffer et al. (Appl. Optics, 2007, 46 (16), pp 3095-3101). It came to my attention after reading T.J. Nelsons project page.

The spectrograph consists of two off-axis parabolic mirrors (OAP) positioned so their tangential planes are normal to the grating's dispersion plane, and aligned so their off-axis focal points coincide. It's hard to illustrate in 2D, so the diagram is broken into several pieces:

I apologize for the poor illustrations.

The OAPs and the diffraction grating. The angle 2α is equal to the angular dispersion of the central rays.

The light enters the spectrograph from an optic fiber and is collimated by the first off-axis parabolic mirror (OAP1) and sent to the diffraction grating.

The collimated light hits the diffraction grating, which disperses the light. Finally the second off-axis parabolic mirror (OAP2) focuses the diffracted light onto a linear CCD (TCD1304DG).

It's not super-clear from the diagram that the two OAPs are stacked over each other. Mirror 1 is employed for convenience, while the "folding" action by mirror 2 and 3 facilitates precise focusing by adjusting the length of the path of the diffracted light.

Here's a simple 3D-diagram, rendered with openscad and meshlab:

And the .stl file if you want to have a closer look.

As the both the diffraction and the CCD are all (ideally at least) positioned in the exact focal points of the off-axis parabolic mirrors, there's no spherical abberation, and because of stuff I don't understand coma, astigmatism and field curvature is minimized. My guess is the abberations introduced by OAP1 is cancelled by OAP2, which requires the two OAPs to have a common focal point, making this design different from the classical Czerny-Turner spectrograph.

In the 2007 paper by Stephanie L Schieffer an impressive resolving poweR of 2.5·10⁴ with a 10µm slit and 1500 l/mm grating is reported.

the stm32f103 peltier monitor firmware

Zip Archive - 202.35 kB - 07/09/2017 at 19:05


  • 1 × SLM laser JDSU µgreen 532nm 10mW
  • 1 × Nikon CFi VC 20x 0.75NA most 20x-40x fluor objectives will do.
  • 1 × raman edge filter 540AELP Ø=25.4mm
  • 1 × long pass dichroic mirror 540DRLP 18x26mm
  • 1 × SM1L20, SM1L10 and SM1L03 lens tubes plus a few extra SM1RR retaining rings

View all 19 components

  • PSU for the TEC cooled Linear CCD

    esben rossel06/02/2017 at 16:40 0 comments

    There are two (possibly more) good reasons for using a separate power supply for the TEC cooled linear CCD.

    1. The computer's USB will be unhappy to power the peltier-element.
    2. The opamp requires ±5V to operate

    The PSU consists of a small ±6V 10VA transformer for the CCD-circuitry, and a slightly bigger ±6V 24VA for the TEC alone. I've chosen to use two separate transformers to avoid an (inevitable?) asymmetric load if the TEC was to be powered by one of two outputs of a single larger transformer. (If this reason is invalid it's because I'm a whale biologist chemist and not an electrical engineer).

    The CCD PSU

    The CCD PCB has separate linear regulators (79L05 and 78L05)¹ for the analog and digital parts of the circuit that are all fed the same unregulated ±6V from the RC-filtered output from the small 10W transformer.

    The current draw is < 30 mA, so the voltage from the PSU will hopefully stay above 7V, or the regulators will drop out.

    The schematic is here:

    There's no other reason for the choice of components, than; it was on my shelf.

    The TEC PSU

    Is regulated with an LT1083 to make sure it the TEC's max-voltage (4.2V) is not exceeded. The schematic is:

    And here they are, all boxed up:

    [insert pretty pic of the psu]

    [1] It's a bad choice for more than one reason, I know. Expect to see a LT1964/LT1761 or similar if I ever decide to improve upon it.

  • The TEC-cooled linear CCD module

    esben rossel05/30/2017 at 21:54 0 comments

    The sensor in the Raman spectrometer is the linear CCD module. However, because the CCD drowns in dark current when the integration time is longer than 1s, the TCD1304DG is mounted on a cold tip that's thermo-electrically cooled using a standard TEC1-03506 running at 4.2V (but the photograph shows the CCD with a TEC1-12706).

    Here it is with the right peltier element. The temperature drop was ~25°C, which gives a reduction in dark current of almost an order of a magnitude. Of course it caused a lot of condensation on the CCD, but when complete, the spectrograph will be sealed up together with bags of dessicant.

    An STM32F103 "blue pill" connected to a 10kΩ thermistor mounted on the cold tip displays the temperature on an ILI9341 display.¹ The voltage drop across the TEC is also measured, just not at the time the photograph was taken. The firmware may be found in the files section of the project.

    Furthermore, the CCD is placed in a more elaborate circuit than the datasheet's typical drive circuit.

    The most notable new features are:

    • ADC input signal conditioning
    • Separate power supplies for the digital and analog sections

    The output from the CCD is upside-down and doesn't quite match the ADC of the STM32F401RE's input range of 0-3.3V. To fix these issues the output is fed to an AD8021 opamp² working as a differential amplifier:

    The gain is:

    G = - R₂ / R₁

    we have an input range of 1.9V (2.5-0.6V), but want 3.2V (actually 3.3V but I don't want to risk clipping the signal), so G should be -1.68.

    The level shift is:

    S = R₄/(R₃+R₄)·(1+R₂/R₁)·Vref

    The output voltage then becomes:

    Vout = G·Vi+S = -R₂/R₁·Vin + R₄/(R₃+R₄)·(1+R₂/R₁)·Vref

    There’s probably a point to the resistor values in the typical drive circuit (noise?), so R₁ is chosen to be 150Ω. The gain then dictates that R₂ should be 252Ω. Setting Vref to 5V (it is in the complete circuit) R₃ and R₄ can be found by solving:

    Vout = 3.2V    when Vin = 0.6V        and
    Vout = 0V      when Vin = 2.5V

    One solution is (almost) R₄ = 150Ω and R₃ = 330Ω

    The CCD-circuit in its entirety is shown here:

    If I were to redo it, I would substitute the 78L05 and 79L05 voltage regulators (U3-4 and U2) with LDOs LT1761 and with LT1964.

    [1] The LCD is driven using fagcinsk's stm-ILI9341-spi library.

    [2] I can't take credit for this, I stole parts of the circuit from David Allmon.

  • The spectrograph

    esben rossel05/30/2017 at 20:20 0 comments

    An overview of the spectrograph's geometry. On the left are the mirrormounts for the folding mirrors. In the top right corner is the mount for the off-axis parabolic mirrors. Center right is the thermoelectrically cooled CCD.

    Three of the 5 mirrors installed:

  • FAQ

    esben rossel02/17/2017 at 05:52 0 comments

    Q: What's the overall cost of this thing?

    A: A quick back-of-the-envelope estimate would be:

    • laser: 250€
    • microscope objective: 150€
    • diffraction grating: 125€
    • filter+DIC-mirror: 75€
    • OAP-mirrors: 350€
    • elliptical mirrors: 150€
    • lens tubes, fiber adapters etc: 100€
    • fiber optic: 100€
    • cases: 100€
    • linear CCD module: 25€
    • connectors: 25€
    • misc electronics: 100€

    So all in all around 1500€. It can certainly be made cheaper, but now you have an idea.

    Q: What's with the expensive laser?

    A: If you want an instrument capable of making reproducible measurements, you need a stable monochromatic light source. Cheap lasers are neither. However, with patience you can find used JDSU and Coherent lasers for somewhat cheap (100-300$).

    Q: What's with the not so cheap microscope objective?

    A: The intensity of the Raman signal is roughly proportional to the numerical aperture squared, so you want as high NA as possible. You can get cheap microscope objectives with NA > 1, but they are oil immersion with a working distance not much longer than standard cover glass thickness (0.17mm). The OEM version of the Nikon CFI VC 20x NA 0.75 was the cheapest high NA objective I could find with a long working distance (if you consider 1mm to be long) - and it's made for fluorescence applications so there won't be any autofluorescence.

    Q: What's with the expensive edge filter and dichroic mirror?

    A: You need efficient blocking of the Raleigh-scattered light, so you need proper filters. And they weren't that expensive - I bought production overruns from Omega.

    Q: Why don't you have a slit in your spectrograph?

    A: A slit is a waste of light if you cannot focus the entire length of the slit onto the linear CCD, and for that you need a cylindrical lens. Both the slit and the lens are expensive, so I chose to simply use the optic fiber aperture as the point source.

    Q: What's with the custom fiber optic patch cable?

    A: The raman signal is weak. I wanted a fiber optic that wouldn't pick up light pollution. It was 30€ extra - but of course around 80€ more than a lucky *bay find.

    Q: Why the super expensive mirrors?

    A: The off-axis parabolic are just not cheap, and once you get into a serious collection of mirrors, the tendency is to push it as far as possible.

  • The Raman excitation setup

    esben rossel02/17/2017 at 05:21 0 comments

    The schematic for the Raman setup is:

    In real life it looks like this:

    The laser is a 10 mW JDSU µgreen 532 nm laser. It operates in single longitudinal mode (SLM). The laser has no beam shaping optics, and the beam is somewhat divergent. This is "fixed" with a beam-expander (the long black tube). I'm not sure this is essential. Everything was carefully aligned one component at a time.

    I would recommend a kinematic mount for the dichroic mirror - I didn't have that at the time of writing, I had a hair-drier (to heat, soften and adjust the 3D-printed holder). It could look something like this:

    The "target" is an electroluminescent backlight that I use because it fluoresces in the 532nm laser light.

    Here the EL-backlight is placed in the focal point of the microscope objective. In the background you see a razor blade beam dump.

    Everything, literally, needs to be carefully aligned. Below you see the difference in light output before and after the last alignment of the dichroic mirror (oh and this time it's not the EL-backlight fluorescence we're looking at, but the fluorescence of a solution of tetraphenylporphyrin):

View all 5 project logs

  • 1
    Step 1

    Spectrograph geometry calculations

    Here's a quick walk-through to calculate the geometry of the spectrograph. I guarantee nothing concerning the correctness of what follows (but please do correct me if you find errors).

    At the heart of everything is the diffraction equation:

    The relevant wavelengths are given by the Raman shifts:

    If we're interested in wavenumbers from 150-4000cm⁻¹ and λ₀ = 532nm, the range for λ₁ becomes 536.3-675.8 nm. The edge filter's cut-on wavelength is 540 nm, so the range is in fact 540-675.8 nm - this also means that the spectrometer's lower limit is 278 cm⁻¹.

    d is given by the grating in the case of 1200 lp/mm it becomes

    Because the CCD is 29.1 mm wide and the focal length of the focusing off-axis parabolic mirror is 152.4 mm we can calculate what angles of the diffracted light we would like. For 540 nm to fall on the edge of the CCD, the angle should be:

    The angle for 675.8nm is of course identical except for the sign. It gives us an angular range of ±5.45° around the center angle. The angle of incidence can now be found by solving this set of equations:

    Of course not any solution will fit the spectrograph's geometry. The angle of dispersion (γ) for the central ray must match the angle between the off-axis parabolic mirrors as seen from the diffraction grating (2α):

    2α is determined by the position of the mirrors and is:

    Where lm is the distance between the mirrors. Their diameter is 25.4mm, so lm cannot be shorter than this.

    I'm don't think there's a (practical) solution to the equations, but setting lm to 27mm, 2α becomes 10.16° and then these values more or less cover the spectral range:

    The CCD then covers the spectrum from 538-680 nm.

    For better explanations go here:

View all instructions

Enjoy this project?



dgourlay wrote 05/18/2021 at 16:35 point

I apologize if this has been asked/answered but apart from the camera.... what was your parts cost?  Have you looked a graphene vs graphite using your instrument?  That would be a very nice spectrum pair to look at.  Thanks for sharing this.  Doug

  Are you sure? yes | no

esben rossel wrote 05/19/2021 at 07:07 point

It's not been completed. Life (very literally) got in the way. Graphene vs graphite sounds like fun, but I don't have any graphene lying around.

  Are you sure? yes | no

dgourlay wrote 05/19/2021 at 12:27 point

yes, life does go on!  I hope that what ever did get in your way was worthwhile and left you moving on to bigger and better projects, one way or the other :)

  Are you sure? yes | no

Kevin wrote 11/12/2019 at 13:55 point

Hi Esben. I'm also building a raman spectrometer and I found your project very fascinating. However, I am wondering why do you have two longpass filters. Would the longpass dichroic filter not make the longpass filter redundant?


  Are you sure? yes | no

esben rossel wrote 11/13/2019 at 14:35 point

The long pass dichroic mirror will reflect most of the rayleigh-scattered light, but without the long pass (steep) edge filter I'm not sure the laser rejection is strong enough. This is the first part of my reasoning. The second is, that the commercial raman probes I've seen also use both, and I figure the professionals know what they're doing. If you can make it work without, I would love to hear about it, those filters are expensive.

  Are you sure? yes | no

Kevin wrote 11/15/2019 at 11:22 point

Yea, the reason I asked is because of all the commercial Raman with that design. I wanted to know why the long pass edge filter is needed after the dichroic. After some more research I think that the dichroic mirror allows some Rayleigh scattering to pass through still. Thus, the use of the long pass filter afterwards. Thanks for the response.

  Are you sure? yes | no

A. M. Aitken wrote 12/01/2019 at 16:32 point

Kevin, a high optical density filter (at least 4 or 5 OD) is essential because the Raman signal may only be 1 part in a million of the incident light.  A 45 degree filter that can do that would be awesome, but dichroic versions do not exist.  In reality all the imperfect filter does in the design is turn the laser beam through 90 degrees down the barrel objective and allows Raman light to travel straight through on the way back.  In other designs this has been done with a microscope slide with a dot of silver mirror on it and still others use a 50% mirror (which wastes half the laser light and half the Raman signal).  The 45 degree dichroic filter is better than these solutions because it wastes less laser light and blocks less Raman signal coming back, that is all.  The edge filter then does the real Job of making the Raman signal detectable.  If spectrometers were perfect we also wouldn't need an edge filter, but scattering can be around 1% even in a commercial unit.

  Are you sure? yes | no

neotech wrote 11/10/2019 at 19:08 point

Hello Esben,nice  project.  I will probably try out  the toshiba CCD chip,  with your schematics... check my video on youtube,,

  Are you sure? yes | no

Butch O'Dowd wrote 01/10/2019 at 13:47 point

I want to build one of these and I would like to see some data as well. 

  Are you sure? yes | no

David H Haffner Sr wrote 05/14/2018 at 18:08 point

Esben, the suspense is killin' me U have some data that we can see?

  Are you sure? yes | no

esben rossel wrote 05/14/2018 at 19:49 point

this past year my dayjob has been killing me. however the school year is coming to an end in 6 weeks time, so hopefully I will be able to complete the spectrometer some time before the end of summer. then I will have a spectrometer party and after that post some spectre..

  Are you sure? yes | no

David H Haffner Sr wrote 05/14/2018 at 22:47 point

After all this time I think U and UR students deserve a vacation :)

  Are you sure? yes | no

ivan003003 wrote 05/14/2018 at 01:51 point

Woooow !!!! Awesome work !!!

Sorry for my poor English,but i wish to ask you something.

Can this spectrometer be used as Hg (Mercury) detector ?

I need an Open-source portable device to detect mercury in close and open environments.

The commercial solutions are way far from my wallet,so for this reason im seeking for this kind of Open-source device .

Thanks :)

  Are you sure? yes | no

esben rossel wrote 05/14/2018 at 04:42 point

The answer is: That depends..

When you say mercury, do you mean elemental mercury ie. Hg, or are you looking for Hg²⁺. I'm guessing that elemental mercury would be hard to detect (but I'm not sure). There are examples of Hg²⁺-detection using SERS in recent literature.

  Are you sure? yes | no

ivan003003 wrote 05/14/2018 at 05:20 point

Thanks you ;)

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates