Raman Probe design WIP

[andy]

I have been working on a raman probe re-work of OpenRAMAN.

I had these goals in mind:

• Reduce price by allowing the lenses to be glued in place after alignment.
• Use an off the shelf spectrometer which saves money, allows use of commercial software, and allows flexibility in the laser excitation wavelength
• Use a fiber laser
• Shrink the size for cost and flexibility (for example placing in a vice and pointing down)
• Keep compatibility with liquid cuvette holder
• Enable microscope use

The total price not including the cuvette holder or the tools needed for building (which the end user doesn't need), I estimate to be $1000. The full system with spec and laser is $2,600. I aim to have comparable performance to the OpenRaman performance edition which I price at about $4,200.

Another comparable probe is this one which retails at $2,250 for the probe by itself (full system is $4,000+). Granted, what I have learned is that people interested in these devices are not really cost sensitive. My hope is maybe seeing how low cost this can get might open new markets, but I think that will take adapting the system to something specific.

I completed my design and built a prototype today, which I can confirm that the parts fit and the laser aligns. I need to wait on another part to see how well the raman effect is captured. I am hopeful, but previous trials on a breadboard didn't go so well so we'll see. I attached some photos.

Code: Select all

+--------------------+---------------------+
| AC127-019-A        |  $              59  |
+--------------------+---------------------+
| AC127-019-A        |  $              59  |
+--------------------+---------------------+
| FLH05532-10        |  $             129  |
+--------------------+---------------------+
| SM05L03            |  $              33  |
+--------------------+---------------------+
| SM05L03            |  $              33  |
+--------------------+---------------------+
| SM05V05            |  $              32  |
+--------------------+---------------------+
| SM05V05            |  $              32  |
+--------------------+---------------------+
| SM05SMA            |  $              32  |
+--------------------+---------------------+
| SM05SMA            |  $              32  |
+--------------------+---------------------+
| 1/2" Dichroic      |  $             132  |
+--------------------+---------------------+
| Mirror #46-618     |  $              25  |
+--------------------+---------------------+
| Bandpass #84-772   |  $             160  |
+--------------------+---------------------+
| Fiber              |  $             150  |
+--------------------+---------------------+
| ASEQ Spec          |  $             849  |
+--------------------+---------------------+
| Laser power TO     |  $             199  |
+--------------------+---------------------+
| Laser TO           |  $             500  |
+--------------------+---------------------+
| Case (If made 25+) |  $              38  |
+--------------------+---------------------+
| Plastic Cover      |  $              90  |
+--------------------+---------------------+

[Luc]

I confirm that typical price for a Raman probe is usually above 2k$. Big part of the reason are already the higher quality filters they use (steeper rejection) which, alone, already cost $1600. Optics can become quickly very expensive!

https://www.chroma.com/products/sets/49 ... ngpass-set

[andy]

I ran into some issues:

• First, I tried to use a bandpass and not a longpass filter at the output side at first, which produced garbage in the output. Fixed that but unfortunately the lens I got back was too thin for the holder, so for now I put it in the output lens tube, which means most of the signal is cut off (starts at like 560nm).
• Not sure how much it matters but at first I also had the dichroic potentially facing the wrong way.
• The locking rings for the input and output focusing lenses ended up being a bit too large for the holder, I'm going to try and lathe out that part of it to make room for it to get a proper focus both in the input (collimating the laser) and the output (collecting the signal)
• I tested with fixed collimator and the spot size ended up rather large, I think some of the laser light was being occluded, reducing the input power. I'm going to try again with just the achromat lens and a smaller na fiber which produces a very small spot, and when focused just right, seems to collimate ok.
• The spectrometer I'm using has a much lower resolution then I had originally thought... Using the formula from this post, the math is ((1.5)*(990nm -500 nm)*(50 um)) / (8um * 3648) = 1.26nm. I might try a 25um fiber like Luc had suggested a while back to see if this simulates a smaller slit size. This would bring it to 0.6nm according to that formula.
• I'm not sure about the output fiber; the NA or the size what it should be. Previously in my breadboard I had tried to use a single mode fiber (P2-460B-PCSMA-1) but to this day have never seen anything come through that fiber. I don't know if it is meant for like high power optics and not low-light or what the issue is. For this test I used the generous Ø600 µm, 0.50 NA which is so large it is very tolerant to being misfocused. My spectrometer (ASEQ LR1) claims an NA of 0.22. Once I fix the focus ring I'm going to try and try a few other fibers I have.

I attached what I got back anyway on Isopropyl with 10second exposure at 50mw.

[Luc]

The issue with single mode fiber is that it's very difficult to couple the signal into it. And the big disadvantage of the larger core fiber is that you loose a lot of light.

I remember that my boss tried to make an OpenRAMAN reproduction at the office except he decided to change the design to use a 400 µm core fiber with a large NA (that and other things because why use something that works if you can build something that doesn't?). He then complained that the signal was very weak and that the spectrometer was shitty until I showed him with basic math that his fiber was throwing away 99.98% of the collected light. It's all about étendue of the system (pi*Surface*NA²). You can consider this as a form of energy/luminance conservation rule.

With a 600 µm fiber, 0.5 NA and a 50 µm slit spectrometer with 0.22 NA (that's already a lot btw -- to be checked because it's what Wasatch claims to obtain with their patented design), you'll get at best ((4/pi)*(600/50))) * (0.22 / 0.5)² coupling which is a 2% efficiency. If the spectrometer is f/4, it drops below 0.5%.

The best is to match the fiber to the spec of the spectrometer and to take great care in alignment. It's also part of the reason fiber manufacturers invented configurations like https://www.thorlabs.de/newgrouppage9.c ... up_id=7719

[andy]

Thanks for the note! That is funny the story about your boss using the wrong fiber. I wonder if other systems re-focus the light or something?

I went ahead and bought that fiber you linked and I'll try it. My spectrometer has a 50um slit so it will keep only 25% but then with 7 cores instead of 1 maybe there will be a 175% improvement.

I ended up diagnosing more issues with my design:

• Critically, the output side lens tube was misaligned. I figured this out because when I manually held the output fiber there was an angle which had much more response. I confirmed by running the laser through the output side and seeing it was not straight. I tapped the part manually so I must have tapped it at a slight angle. Edit: I was able to fix this today by using a drill bit slightly larger then the thread basically removing the tap. I drilled carefully this time, sliding several 1/2" locking washers onto the bolt of the drill bit to "guide" it straight. I put the lens tube through and screwed tight a locking ring to hold it. This worked and it is now aligned on my prototype. Wont be an issue in production because a CNC machine will tape the thread.
• The longpass lens was too thin for the holder. I used a washer but you can see how the irising is. I bought a brass spacer and will try that. With my design the lenses will be glued on helping some of the irising but I should lookup the theoretical spot size is given the 45 degree angle of the 1/2" beamsplitter.
• I discovered my laser says it needs 100um core at minimum so I was losing a lot of power using smaller fibers

SNR still appears to be around 1:50 at this stage. More testing to come!

[andy]

My biggest struggle continues to be with the getting light collimated out of the fiber, and coupling back in (basically the only fundamental differences between my design and OpenRaman...).

Today I used the thorlabs multimode collimator for the input laser (I discovered that it screws on to the 1/2" lens tube so that is nice), so that kind of solved the collimation out. Only issue the spot size is rather large so I think some of the output signal may be shaded out, I have to check this.

I tried using a fixed focus collimator on the output side with the single mode fiber. I put a mirror up where the frosted glass would go and was able to get things lined up that way and even see a lot of signal come through, but when I tried to move to the raman test, I didn't see much other then laser raleigh scattering if I removed the filter. I'm not sure why even the florescence wasn't really showing up, but I think again it's just alignment. I don't know if the tube is perfectly centered (though from the laser test I know it's pretty close), and I don't know if the fixed focus collimator is centered on the tube (I glued a thread ring as a improved adapter). No matter what I did adjusting the beamsplitter I didn't see anything. I don't know if it's just such a small adjustment that I'm missing it; of if the collimator is pitched (pointed off axis) and that's causing an issue. Another thought I had (not sure if this is right) is slight amount of mis-collimation could adjust the focus point of the fixed focus collimator, and at the distance here (over 150mm), maybe is outside the spec of the multimode collimator.