I am going to build a Raman spectroscope over the next few months. This thread is to document my work; I'm starting a new thread so as not to clutter
existing ones.
First Steps:
I've bought a Toshiba TCD1304AP CCD detector, a Peltier cooler, a Nucleo F401RE microcontroller, and a power supply. I'm waiting for my laser (25 mW,
532 nM) to arrive. Currently, I'm working on the optics and electronics design (using Eagle to lay out the PCBs). I should be done with the designing
by the end of the week.aga - 6-4-2015 at 13:10
Do you have a Plan of how you will construct this ?
What are you going to use to filter out the excitation laser's wavelength ?Zombie - 6-4-2015 at 20:46
Anxious to see what you develop.Metacelsus - 7-4-2015 at 04:37
Yes, I have a plan. This is a school project (the school said I could do a project of my choice), but I will be doing it at home. I'll most likely use
an edge filter for filtering out the laser wavelength. I also might try blocking it with a thin mirror after it comes out of the diffraction grating.
I am going to use OSHpark for the PCBs, as I can't etch double-layered PCBs very well at home.
The laser arrived yesterday, by the way.blogfast25 - 7-4-2015 at 06:11
A diagram would be useful here.aga - 7-4-2015 at 11:22
use an edge filter for filtering out the laser wavelength. I also might try blocking it with a thin mirror after it comes out of the diffraction
grating.
It will be very interesting to see what results you get with each of those methods, either singly or together.
Please post details of the filter (and where to get one) .
I am going to use OSHpark for the PCBs, as I can't etch double-layered PCBs very well at home.
I use a strip of pcb and selloptape. Line the two transparencies up with the strip of pcb off to one edge and tape them to the strip.
When you put your board in there, use a couple more tabs of sellotape to hold the thing in place.
If that mob do them cheap enough, may as well save yourself the bother tho.
Work has been going slowly, but steadily (my time is split between this and the TEMPO project). After a bit of trouble with the 3D printer, I managed
to print all the parts for the spectroscope portion (I'll build this section before the Raman scattering one).
On another note, I learned not to trust sketchy $4.95 ATX power supplies from the surplus store. Of two I bought, one did nothing, and one made a
small explosion (a bad capacitor, I think).
Steam - 26-4-2015 at 11:46
I am very interested, how do you plan to collect/ analyze your data?
I am attempting to build my own raman spectrometer. Metacelsus - 26-4-2015 at 17:32
I'll use a CCD hooked up to a STM Nucleo board, with a USB connection to a computer.byko3y - 26-4-2015 at 18:17
Will any window, lense and diffraction grating make noise in the spectr? I would be glad to make my own IR or Raman spectrometer, but the restriction
on materials holds me back. Otherwise Raman spectrometer could be really cheap.Metacelsus - 27-4-2015 at 04:59
I'm not sure how much noise will be present. Theoretically, my spectroscope (a Littrow configuration) should be able to get down to 1 nm resolution.
However, this requires a very small slit width. If the signal-to-noise ratio is too small, I may need to increase the slit width to let more light in,
thus sacrificing resolution.smaerd - 28-4-2015 at 04:02
Cheddite did you hook up a ghost load on the ATX supply which did nothing? Often times they won't do anything unless one of their power lines atleast
has a small resistive load IE a series resistor & LED or something.Metacelsus - 28-4-2015 at 05:04
Yes, I did.IrC - 15-5-2015 at 00:37
I have been working on several variations of Lock-In Amplifiers since it is possible to detect signals buried in noise which can be orders of
magnitude greater. Doing some searching as it relates to this area I found a very good paper that I thought I should post here. Have you given any
thought to this approach? This thesis appears to be very applicable to your project.
A lock in amplifier solves a different problem. For a Raman signal on a CCD the thing to do is integrate and background subtract. Frames as long as
possible as often as possible and just add the signal together. SNR will improve with the square root of the number of frames.Metacelsus - 17-5-2015 at 11:44
I got my CCD working with the Nucleo, finally! Spectra (and more details) are soon to come.
The oscilloscope picture is the CCD partially blocked with a finger.
(Pictures of oscilloscopes, or at least 1950-s era ones, apparently don't show up very well: an interesting effect of the speed of the eye being
slower than the speed of the camera.)
[Edited on 17-5-2015 by Cheddite Cheese]smaerd - 17-5-2015 at 12:46
*watches eagerly*
This looks like a really low cost build. I'm really excited to see how you build your optics path. I like your approach so far. I may follow in your
foot steps after I wrap up two of these projects and you have yourself some success.
edit - I really wish there was a subforum for building scientific instruments/tools. There's so much ingenuity on here.
[Edited on 17-5-2015 by smaerd]aga - 17-5-2015 at 14:03
The hell with a subforum - keep it going !
I also eagerly await an outcome.Metacelsus - 18-5-2015 at 15:28
I've set up a program to save the CCD data on my computer; here is a preliminary example of the CCD being physically blocked.
smaerd - 18-5-2015 at 16:05
Looking good! How are you talking to your microcontroller?Metacelsus - 18-5-2015 at 16:30
I'm emulating a serial port over USB. (virtual port com3)smaerd - 18-5-2015 at 17:10
That's an interesting approach. I always used the RXTX libraries with JAVA which is pretty limited, but its cross-platform, and pretty easy once you
get the hang of it. The nice part of it is it lets you manipulate and display the data live (RAM permitting), gives you access to a UI if you want to
perform tasks like background subtract or whatever. If you choose to go that route let me know and I can set you up with a lot of code that will save
you tons of time.
Looks great though I check this thread probably five times a day.
First Spectrum (not Raman, though)
Metacelsus - 21-5-2015 at 12:31
This is of a red-orange laser diode:
Average of 8 scans, integration time 10 ms each. (I've found that more scans are better than longer integration times.)
I still need to calibrate the spectroscope to convert pixel number into wavelength, not to mention building the Raman scattering portion.
[Edited on 21-5-2015 by Cheddite Cheese]aga - 21-5-2015 at 12:35
The spectrometer is a huge part of the story, and both smaerd and yourself are well on your way.
I feel a Tricorder bashing it's way into existence.
[Edited on 21-5-2015 by aga]m1tanker78 - 22-5-2015 at 06:01
Nice job so far Cheddite. What type/resolution ADC are you using?Metacelsus - 22-5-2015 at 11:28
I'm just using the ADC that's built into the Nucleo F401RE. It has 12-bit resolution, and is fast enough for my purposes.aga - 22-5-2015 at 12:47
The ADC is fine.
Just make sure the signals are free of external/internal noise by using earthed screening tin (soup can), filtering capacitors and maybe even ferrite
beads in the signal path for hi-freq noise.
Is it ready yet ? C'mon !Metacelsus - 22-5-2015 at 15:27
This part holds the laser, the 40x objective lens (cannibalized from a microscope), and the beamsplitter (which hasn't arrived yet).
