I couldn't resist! I've built the smallest of the spectrometers, but using a picam1 and a Raspberry vers. 1 (!). I modified the new version of the software to make it compatible with the old picam and removed some bells and whistles to make it less sluggish. Now it is kind of usable, I've just to wait! I also modified the source to average the spectrum on an arbitrary number of rows and changed the 'holdpeak' behaviour to let the signal accumulate for very weak sources. So, thanks for all, Les, that was fun.
You've been a busy boy Les. I have built V1 so am very pleased to see that you have upgraded the software. The calibration is brilliant kitchen physics. Great for teachers. Kudos to you. Now to get hold of a Pi4.
Thanks! It will run on a 3B+ without issues, but the boot time for Bullseye is a little slow. It may work ok on others so long as Bullseye will run on it.
@@LesLaboratory Is it possible to use an ESP-Cam that streams the video over WiFi that pipes it through a UDP receiver in order to make a stand-alone, portable version? That would be epic, Sir. That's exactly what I'm trying to do. I want to stream it to a tablet. 😁
Thank you for all your videos! Just came up with interesting idea: What can be interesting is to put linear lens array in front of tube, and get linear array of specters. Then visualize it in 3D. Or even better to create 2D array of spectrometers and visualize in 4D.
I've been using your spectroscope this weekend and it helped me with barium salts identification. I had many sodium contamination and it's a pleasure to use the spectrometer to find elements like sodium calcium barium and strontium... Also helped me to find sodium contamination wich was supposed to be calcium, but it's clearly sodium.Thank you very much for your work ❤🎉
Excellent videos-this has been so helpful. Have you thought at all about adding some features for the plant-growing hobbyist? I am using this to figure out if my Orchid LED strips are generating the proper photosynthesis wavelengths (PAR/ePAR), so if there is a way to make the Y axis show PPFD (Photosynthesis Photon Flux Density) or any other plant-related metrics, that would be fantastic.
Thank you. Very interesting! Getting ideas on how to study dormancy in spruce needles through spectroscopy. Found an article (Vegetation and Dormancy States Identification in Coniferous Plants Based on Hyperspectral Imaging Data) that gives the most likely band candidates to investigate. Maybe I can record the data to train a model that can identify "sleeping" trees
@@GemFacets my dear friend, because I am also interested in the distribution of precious stones, did it work for you? Can you give me some details?Thank you.
I am curious as to what kind of readings one would obtain by looking at stars through a telescope with this device. I am also wondering if the camera can be adapted to a wireless esp-cam streaming via UDP. Your work is spectacular! Keep doing what you're doing.
Galactic redshifts are on the order or 10s of nm. With a suitable telescope you might be able to measure redshift with your setup. I think you would also need a chamber filled with hydrogen to get an adsorption spectrum? My gr 12 physics is failing me...
Something I would like to get into is stellar spectroscopy, I just need time...I have had a couple of ideas about how this might be accomplished and calibrated. Red-shift would be especially interesting...
@@LesLaboratory How cool would it be to map the spectra coming from a Sonoliuminesce experiment where water cavitation occurs in standing-wave nodes? If one were to identify helium in those produced emissions, I think that would be an epic breakthrough.
This looks great! A few cool things to add would be: 1. Average mode for reduced noise (continuous integration?) 2. HDR mode (combine multiple different exposures to avoid clipping the high intensities but also get low noise on low intensities), possibly with log Y option 3. Blackbody calibration (incandescent light bulb?) to get a more realistic absolute intensities 4. Relative spectra (compared to a saved reference spectra to be able to get absorption spectra)
1) Peak hold integrates the max values of all peaks over a period of time, but I guess you are thinking of grabbing say 30 frames worth, averaging then displaying? 2) Like image stacking in astophotography?. LogY sounds interesting. 3) This may be done at some point, but software might not be the best solution. Picams (like many cameras) have poor linearity across the spectrum. The current most sensible solutions are to debayer the camera sensor (not easy, see my other videos!) or acquire a B&W USB camera. Both have relatively linear responses when compared with color cams. 4) A cool idea!
@@LesLaboratory 3) Couldn't you calibrate the software by measuring the change of brightness while doubling the distance to the blackbody emitter and determine the change in measured intensity to the theoretical change in intensity?
@@stefanmayer444 for any given individual wavelength yes, but the color filter on the bayer layer results in a non linear spectral response. It should be a bell curve, from about 400nm to 900nm or so, but it is way off and this would need compensating.
@@LesLaboratory Okay, so Ideally a monochrome sensor, but we can still hope for a software compensated solution? Would be interesting to see the accuracy between a monochrome sensor and a software compensated RGB sensor.
This is an amazing creation! Coming from an audio perspective I love how the same tools of spectrogram and spectrograph are equally revealing and amazing, just at different frequencies. :) If the software ever developed the ability to determine CIE 1931 xy chromaticity coordinates, along with the ability to calculate the true luminosity in nits, this would let the it be used to match a relatively inexpensive tristimulus colorimeter to the absolute accuracy of a spectrophotometer for display colour calibration. This match is done by measuring the same “pure” RGBW colour patches on the same display with both meters, and then the software take those two sets of chromacity measurements and use them to match the values produced tristimulus meter which can measure quickly and at low light levels to the spectrometer. Currently the tools to do this sort of measurement to 10nm accuracy, i.e. good enough for a LCD/CRT/bulb projector are ~$1.5K. And to get to a narrow enough range to be able to accurately measure the primaries of a laser based source would require a $15K instrument. There is a strong community of amateur home theatre colour calibration enthusiasts (and professional I’m sure) that would be eternally grateful if this window opened for them via your DIY creation. Me included! Cheers, take care and thank you. -Jonathon
Wonderful! Two questions: 1) If using a pi camera that does NOT filter out IR, how far into the IR spectrum can the spectroscope & software support? 2) Have you or are you interested in exploring integration of something like ArgyllCMS? I'd love to be able to get the output of >>>spotread with this level of precision and resolution.
The physical spectrometer is designed for visible light, and the deepest you can get is about 800nm or so. The software however does not care too much about range. When I was testing, I was using test data from 300nm to 1500nm, so it should be fine. You could always roll your own spectrometer head and indeed I am thinking of looking into this myself, there are too many interesting things to be seen in the IR that I am missing! I have not used Argyll CMS or spotread, and am certainly not an expert in color spaces. What is your application?
@@LesLaboratory It started out as an interest in accurate CRI and TM-30 analysis on home and studio lighting sources - essentially looking for DIY alternative to the Sekonic C-800. That said my interest has evolved to include effects of deep red and IR in home lighting, wellness, etc. I'm interested in UV as well, but that brings with it many more challenges and safety considerations that haven't yet been overridden by curiosity. I'm subscribed and will be following along if you do venture into a wider spectrum!
Does it need a high speed pi like a pi4 (I see you have a fancy cooling rig!) or will it run with a low-end one such as a pi zero ? The zero seems better suited to a self-contained unit.
It ill run on at least a 3B+, but I have not tried it on the zero yet. If Bullseye will run on the zero, I would reckon this software would, although I expect it will be slower than on on the Pi4. Since the software now supports full-screen, it might be possible to run it in a minimal desktop environment.
Thanks for the video and the weblinks! 1) Which RPi cam would you suggest - the normal one or the NoIR? I mean, for instance some of the most prominent emission lines of oxygen lie in the near-IR, will I see them with a normal cam? On the other hand, if I had a NoIR cam, would I still be able to properly see all other lines in the visible range? 2) Would it make sense to use a 12MP cam instead of a 5MP cam? Thanks in advance!
Will start building now. Thanks! Would it be possible to extend this into the IR range with the appropriate sensor/grating? I have used my Pi IRCUT cam to test some dodgy green laser pointers. With the filter turned off it was possible to see two lines of IR with diffraction. Just beginning my deep dive into optics/lasers so thanks for creating this.
Thank you very much for this wonderful spectrometer! I was thinking about building a spectrometer for colorimetric analysis for cheese making and earth chemistry and as part of an AES spectrometer - your spectrometer and software are much better than anything that I could have built myself! There are two tricks from experimental physics that I thought I'd mention in case they were of any interest: One is to use an "optical chopper" (usually a "chopper wheel" or "chopper disk") and phase sensitive detection to improve the signal/noise when measuring weak light sources. I think modern cameras and a Pi might be fast enough for phase sensitive detection to be implemented in software now though I'm not sure. Another is to use a reflective diffraction grating instead of a transmissive diffraction grating - that can allow the wider range of wavelengths to be measured since the light doesn't pass through the grating.
For anyone who has built this, did you have to make any changes to the libcamera settings? When I point the spectrometer directly at the FL reference light, I get high-intensity white lines that start to disappear when I turn the meter away from the source. Thanks
Les, this is an excellent series and I'm happy I found it. I haven't read all the comments, but I thought I'd ask about turning the spectrum to a diagonal and taking the pixel samples from that diagonal. For a 5MP camera you should get: 3240 diagonal of 2592 x 1944 3240-2592 = 648 648/2592 = .25 so 25% more pixels I will be reviewing your code over the next several days , but I was wondering if you had thought about doing this to improve resolution and/or have any comments about the feasibility? Thanks!
So, you could do this, but... The resolution of the camera is already very high at 2592 horizontal pixels. Of these the software scales down to 800px, and then processes that data. The reason is, 1/3 the data gives a >3x speed up in processing, so we can get a live view. Secondly, but more importantly the spectral detail that can be resolved depends on the optics more then the camera (although sensor size plays an important part). It is not likely you could split and resolve the Sodium D lines for example no matter how many pixels you try and throw at it, so we would just be processing wasted data. This could all be fixed by designing a better physical package with and adjustable slit, precision ruled gratings, coated optics and a larger sensor. The problem then is that the cost skyrockets.
Well that all makes sense. I'm trying to implement this on a Raspberry Pi 5 and have found that the latest and greatest isn't so great. I can't get the camera to work on it and I'm seeing all kinds of Google stuff about problems with both the camera and underlying graphical system compatibility with older stuff. Very frustrating. Today I will retreat to a RPi 4 and if I can't get that to work I'm going back to an RPi 3. I really like the work you've done and I'll report back on what I get working. Thank you so much for this project. @@LesLaboratory
Just to help someone else who might have problems with the RPi 5 as I described earlier: I retreated to the RPi4, same camera, same tests AND IT JUST WORKS! While this does not preclude the inference that I've done something stupid, it does point a stronger light on the RPi5 as having camera problems.
Not sure if this is the right place for this question but I’ll give it a shot. On your nitrogen laser low-voltage Power supply to Drive the ZVS board to power the flyback transformer what are you using might I ask? I’m finding power supplies that are producing 0 to 30 V at 20 A is that enough or too little? Whatever information you’re willing to share would be appreciated i’d like to say that your nitrogen laser build is brilliant. thank you
Thanks! I run the ZVS driver at 12v and 1A, there is no need to go crazy, for this application, you only need 20kV at a few milliamps. I have a ZVS PSU I built that does 30V at 6 or 7 Amps, but it is way to powerful to drive a Nitrogen Laser, all it would do is start a fire! Less is more with stuff like this.
Very nicely done! Is the vertical scale linear? so that, assuming I find a source with known intensities, would I be able to calibrate the vertical scale as function of wavelength? if the camera gain if fixed, it should be possible.
So love this, and got the software part to run. I see you made a note about max 800 pixels, and it crashes if you try to higher. Why the restriction? is it due to the display limits or a driver problem? I am using a USB monochrome camera on a pi5, and out of the box results are great, just seeing if i can utilize the full res.
@@FernandoMoreiraR There comes a point with the spectroscopes where a higher resolution doesn't gain you much, and the program runs slower so 800px seemed like a good compromise between speed and resolution. It is all open source so you can hack it up to whatever you want.
Great work. Are you using the Pi camera that you stripped the bayer layer? Is there any calibration to the intensity of the light at different wavelength?
I was just using the ordinary version of the camera in this video. Intensity is arbitrary units from 0-255, however it can save data as CSV which you can import into whatever program you like to do post processing
Wonderful idea and very clear presentation, thank you. If I source a suitable camera would the spectroscope and your software be compatible as far into NIR as 1850nm? I see some of the Pi cameras with Sony IMX 219 are without IR filters but I haven’t yet been able to find exact specifications as to wavelengths filter version cuts in. Any subscribers in the know?
OK, so a little more research completed. For anyone looking outside of the visual spectrum a camera without an IR filter is what you need. However, most compatible cameras seem to run out by 1000nm. The ultra low light IMX462 without IR filter seems to be about the best option with a little more width to 1100nm (from what I’ve read so don’t quote me) It’s only 2MP and uses a bigger M16 fitting too, anyone who’s built one, would these be a problem if matched to a suitable lens? I don’t have any suitability info on the spectroscope. I did email the suggested retailer but had no reply - any scholarly subscribers with a pearl or two of wisdom?
1.Can the spectrophotometer accommodate a cuvette or similar sample holder? Specifically, I am interested in using it to analyze absorbance and transmittance of solutions within a cuvette. If so, do I need to make any modifications to the software and hardware from what is shown in the video, or can I use the setup as it is? 2.Lastly, this might sound basic, but could you clarify what the Y-axis represents on the graph? I understand that the X-axis indicates wavelength, but I am uncertain about what the Y-axis measures-whether it is transmittance, intensity or any other value.
@@praveenragavesh1807 you would have to make a suitable cuvette holder, and build a suitable light source to measure transmittance. The values on the Y axis are intensity which are arbitrary values from 0 - 255. This is good enough for most purposes, but if you want them tied to a specific value, you would need an additional sensor.
Not sure you have mentioned it but the camera may need to be on for 5-15 minutes before calibration to have all parts at operational temperature and also note the temperature of the room at time of calibration
Amazing work ! Keep playing with spectra and spectrometers - there are amazing things that can be done with them. Just look at all the spectrometers used on space missions and not to mention all the amazing applications they have here on the ground as well. I once saw a hand held spectrometer for geologists that could give you the major chemical makeup of a rock - that device costed several tens of thousands of dollars/euros. But i really think it costs that much mainly because of intelectual property rights and so on, the actual hardware shouldn't be that expressive to build a spectrometer with an equal performance - maybe I'm wrong but that's what I understood. What do you think ? My dream is to be able to built highly sensitive spectrometers like that geologic one that I've described or even better with regular off the shelf components and granted maybe a camera ccd a bit better than average but the rest would be software i guess. In your experience, or anyone else reading the comments for that matter, do you guys think it would be possible to build spectrometers capable of identifying specific chemicals ? In my humble opinion spectrometry is the future and at the moment sensitive spectrometers are only available to very very expressive research labs but just imagine how much innovation would be enabled by enabling people like you here to build your own very sensitive spectrometers... Imagine having a NASA style curiosity Rover quality spectrometer available to you.. The possibilities would be endless. I can't wait for these kinds of open source hardware information to improve and to become more widely available. Thank you for your work and for documenting and sharing what you're doing here on UA-cam. All the best to you !
Thanks! I think what you are describing there is a Raman spectrometer if it can identify samples. It is something I intend to look into if I get time. The software is where the complexity will be especially if you want to autodetect, though for Raman setups the optics are fairly critical as well.
Excellent stuff ! I think I shall have a play with this as I have a 10 quid ebay usb microscope that I would happily pull the sensor out of or maybe it might work 'as is ' what do you think ? You have done an amazing job with this and I hope schools and colleges know about it as back when I was a student I would loved to make something like this ! and you get a very capable scientific instrument at the end of it. This has the lot, Lasers (everyone's favorite!) optics, software, electronics so it's all there...cheers.
Thanks! Sure, a USB camera should work with the USB version of the software, and will probably work more or less as is, though you will have to tweak the focus! There have been a few colleges e-mailed me since the first version last year, but yes, this kind of kit should be in every school and even Universities for that matter! It is such a pity that the "Amateur Scientist" and other great columns are no longer a thing!
Hi, this is really amazing! Do you think it could be used to determine elements and organic matter in the liquid (water)? Or will more sophisticated equipment be needed? What should be used as a light source, laser, UV or IR? Or more sources asynchronously, e.g. IR for organic, UV for chemical elements...
I think it would take more, work and probably a more sensitive camera. Really for determining composition of materials, you would be looking at Raman spectroscopy or hyperspectral imaging.
Is there any way to customize waterfall sample period or to make composite image automatically? Now waterfall shows only about 30 seconds. Would be awesome to be able to make long composite or set e.g. for 2 hour waterfall image. Thanks
It could be done, but I have questions: What is the application? What would be the required update frequency? (Important, since we have limited RAM on the pi!)
@@LesLaboratory To see light spectrum changes during the day for example. There are studies that prove light spectrum is changing throughout the day (of course) and it's highly beneficial to get some of the "specific light into our eyes" in the morning and also in the evening to trigger hormone production that helps us sleep, relax, aware and so on. I would like to set up a little monitor to collect data during the day and compare sunny day to cloudy day, etc. It would be great to have the option to snapshot CSV data on particular times. I would suggest update frequency to be editable. Not sure how it impacts RAM or if it can be stored on the disk to lessen memory consumption.
Would it make sense to have a way to calibrate the sensitivity as well, or there is no reliable reference that will always have the same ratios of intensities between frequencies across the whole spectrum?
Probably. It can probably be done with a tungsten Lamp and some algorithm to normalize the spectrum. Absolute intensity is another matter, and is probably not reasonable for all circumstances.
You specify a zoom lens, but are you actually adjusting its focal length, or just using one setting? I have played around with more than a few RPi cameras. It has always been my experience the M12 zoom lenses are far worse optically than fixed-focus lenses. Its not even close.
Great stuff, I have a spectrometer made from a 3d-printed black enclosure and a usb Webcam with it's IR filter removed. This looks way better! Are there differences in performance between the two?
The two versions shown in the video have comparable performance to each other. The entrance slits in each are the same size, and so the miniature one is just as crisp and bright.
Donated! Wow Les it's incredible to see the v2!! So many nice features. How about CIE1931 display with the blackbody graph and the Rec.709, P3, Rec.2020 color spaces in the future versions?
Thanks so much for your support! :-) Now that would be interesting, though it ill take a little while to wrap my head around the math for that. What would be the intended purpose? Measuring the color temp of light sources?
@@LesLaboratory In our Video Recording Lab at the University we are teaching students different light sources properties. The CIE1931 in 2D or even better in 3D is our basic graph for understanding the light sources/cameras/HDR monitors wide color gamuts. It would be great to have it as an option on your awesome software - if possible.
@@aviationodyssey1892 A couple of people have asked about this, and I am looking into it. It looks like it might take a bit of figuring out, but at this stage I have the general gist of it. It looks like luminosity data is relative rather then absolute, which solves a headache. When displaying the data, I assume a simple marker on the Chromacity diagram is what is required? Would it also be useful to print out the x & y coordinates? Edit: Although I have the gist of it, by the looks of the academic paper I looked at, it will likely take a while to implement it sensibly, but I am sure it ill be an adventure!
Hi-- art conservator here! I'd also be interested in this! I am building this spectrometer to identify gas fills in neon/gas-discharge tubes and would use chromaticity measurements (CIE1931) to characterize the color of the neons/gas-discharge tubes and assess for color matches and color shifting! Thanks so much for this incredible work.
@@taylorhealy7889 You are welcome! It will make it into the next iteration as either part of the existing interface or an add-on. From the requests I get for this software, I might be as well building a modular plug-in system, but we shall see.
super great video Les, great respect. I do have a questin. Do you know if the resolution is enough to detect mercury an cadmiun in food. Hav ea great day, Les. Cheers
Thanks! In order to detect heavy metals with a spectroscope, you would need to heat them to incandescence, and there would have to be sufficient quantity to be able to observe the characteristic peaks. For ppm or ppb values that you would likely encounter in food, you would have to consider X-Ray-Fluorescence (XRF). For qualitative detection is is possible to use Phenyl-,2-phenylhydrazide.
Love your work. Unfortunately I needed something that handled the UVA and UVB range (primarily 311nm and 365nm so this build wasn't suitable for me. Great stuff though. Awesome.
Hi Les, I am part of a team working on creating a novel toroidal plasma accelerator. We are interested in using this setup to examine the black body radiation of the plasma for our study. However, the spectroscope you are using is out of our budget. Do you have any cheaper recommendations for us? I was looking at either using a "pocket jewelers spectroscope" or a 600 line Diffraction Grating sheet which could be made into a homemade scope with a 3D printed housing. I look forward to any ideas you may have!
If the scope you are using gives a usable image it should work ok, but I have never used one of the cheap ones. I used the patton hawksley scope, because I had one on hand. There is no reason why you can't build your own. With a self build you can get into the IR. If you can afford (or scrounge) the optics, the UV is not out of the question either.
Awesome build! I need a spectrometer for some time now but always was holded up by cost. I need it to go down to 360nm, is your software going this low? Thanks for sharing!
The software will, however the spectroscope shown will not. UV requires silica optics for a transmission spectroscope, so you would need a custom build.
There are NoIR cameras available from the Pi Resellers, For cameras with an M12 mount and No IR filter, just look for the Picam night vision knockoffs.
Hi, I am using your software for an IR spectro (with the NoIR pi camera), what should I change in the code to extend the wavelength around 1000 nm ? Thank you very much, your job is awesome :)
The software will handle the new range, but the optics might not. The spectroscope I used was designed for visible light, and has a range that extends to about 800nm or so. I suppose if you got the camera to look into it at a weird angle, you maight get close to 1000nm, maybe.
What can be interesting is to put linear lens array in front of tube, and get linear array of specters. Then visualize it in 3D. Or even better to create 2D array of spectrometers and visualize in 4D.
A question, how does is it possible to calibrating the amount of power measured by this spectrometer? like for example, uW/cm2/nm. I'm looking for solutions! Thanks in advance!
Folks, I built Les's excellent spectrometer, but the small zoom lens seems to come in a number of variants. The one I have won't let me focus the spectrum with the PiCam. The lens I have looks the same superficially, but has the following spec: F1.6, 3.6 - 10mm (6MP). Can you suggest one that works please?
My suspicion is that the focal plain at the back of the lens is missing the sensor. (M12 mounts arent created equally, some are taller than others) If you unscrew the lens from the m12 mount a little does the focus get worse? If so, remove the m12 mount, and carefully file down the front threaded end of the mount by 1mm, this will allow the rear lens to get closer to the ccd re-assemble and try again. Don't remove too much material in one go, you don't want the lens to be able to be screwed up against the sensor!
@@LesLaboratory THANKS - You are probably right, but feels savage to take a file to a lens or lens mount though. Has my lens got the same focus range as yours?
Build a nitrogen pulsed 2 joule per second and made a rough spectrum uv machine that was hooked up to a photo diode and oscilloscope 100 Mhz. Tested many chemicals including cobalt chloride, copper sulfate, hcl, ethanol, ethylene glycol, water, methanol, etc.
Hello. I have watched your video about the Raspberry Pi spectrometer. I would like to know if what is the unit of intensity that your software is displaying. Thank you.
Woohoo! I'm excited about this video, and thought you might've put the pispectrometer project aside for the time being. Would you be willing to accept pull-requests on this new repository?
No. It is possible though, but you need to be aware the spectral response of PiCameras is very nonlinear. This can be fixed by de-bayering the camera (see my other videos) and it is potentially fixable in software.
Hello, I was wondering if you had any idea what might be the cause of this issue. When I try to calibrate, I get an error message that reads, "loading of calibration data failed(missing caldata.txt or corrupted data)". The problem is that no matter how many times I try, the error keeps coming in. If you have any idea what I should do, that would be really helpful (I've already tried creating the caldata.txtx file by myself and it's still not working).
A question, what OS are you using? For some reason my raspberry does not detect the camera, I tried with another raspberry or changing the camera, it is not a hardware problem, I think it is a software problem.
Thanks! I have not tried this, and all I can say is, its a maybe, but unlikely. The problem is that regular glass will absorb significant quantities of UV light, and many glass types are quite opaque at these wavelengths. You would need to construct a spectrometer head with either Silica optics, or go for a reflection grating setup instead.
The software is capable, but the spectroscope itself is designed for visible light only. With a custom hardware build, but UV an IR imaging are possible.
@@LesLaboratory I looked into sensors capable of measuring CRI LUX, while they are cheap, the PAR (which is by far one of the most important) is quite expensive. The sensors start at 150$.
I found the Arducam problematic in terms of drivers etc, that said, that was back on Buster, I have not tried it on Bullseye. If i get time over the holidays, I will spin up a now install and see ow it measures up!
For the moon, it would probably work, but for stars, well they are already a point source for all intents and purposes, so a different physical setup would be required, that would dispense with the slit.
Do you anticipate there being a relatively easy way to modify the python script for the USB version, to not crash when using a resolution other than 800x600? I've been looking for a means to interface with a Thunder Optics SMA-E spectrometer on a raspberry pi (or another SBC, but RPi is most likely to find existing open source support/software) which I can also use with GPIO to stop a Titanium anodization process, inexpensive configurable optical endpointing... I was able to view an image from this SMA-E (USB) spectrometer via guvcview but I cannot configure the resolution to 800x600 (It is 640x480), nor can I get pyspectrometer2(USB) to run without crashing - I don't have a picam to verify there are no other issues... but guvcview works with it and python3-opencv IS installed. Thanks for any help you can provide
Probably with modification, though it has caveats I guess. LED monitors would give a different spectrum than CCFL monitors, regardless of the test image being used for calibration.
Hello, I am pretty dumb. I downloaded python to PC. I downloaded your software. I can not get it to turn on. I can get the python screen to come up [ the one that appears to allow commands to be entered] and the html page for python. Somehow I am missing something. Maybe put your program folder in python folder? I am wanting to use pc for a while before I get the raspberry.
![A Small, Cheap Micro-Spectrometer - Review [Pt 1]](http://i.ytimg.com/vi/MOXryggwr_Q/mqdefault.jpg)
I couldn't resist! I've built the smallest of the spectrometers, but using a picam1 and a Raspberry vers. 1 (!). I modified the new version of the software to make it compatible with the old picam and removed some bells and whistles to make it less sluggish. Now it is kind of usable, I've just to wait! I also modified the source to average the spectrum on an arbitrary number of rows and changed the 'holdpeak' behaviour to let the signal accumulate for very weak sources. So, thanks for all, Les, that was fun.
You've been a busy boy Les. I have built V1 so am very pleased to see that you have upgraded the software. The calibration is brilliant kitchen physics. Great for teachers. Kudos to you. Now to get hold of a Pi4.
Thanks! It will run on a 3B+ without issues, but the boot time for Bullseye is a little slow. It may work ok on others so long as Bullseye will run on it.
@@LesLaboratory
Is it possible to use an ESP-Cam that streams the video over WiFi that pipes it through a UDP receiver in order to make a stand-alone, portable version?
That would be epic, Sir.
That's exactly what I'm trying to do.
I want to stream it to a tablet. 😁
Thank you for all your videos! Just came up with interesting idea: What can be interesting is to put linear lens array in front of tube, and get linear array of specters. Then visualize it in 3D. Or even better to create 2D array of spectrometers and visualize in 4D.
I've been using your spectroscope this weekend and it helped me with barium salts identification. I had many sodium contamination and it's a pleasure to use the spectrometer to find elements like sodium calcium barium and strontium... Also helped me to find sodium contamination wich was supposed to be calcium, but it's clearly sodium.Thank you very much for your work ❤🎉
Fantastic! I am glad this stuff is useful!
Excellent videos-this has been so helpful. Have you thought at all about adding some features for the plant-growing hobbyist? I am using this to figure out if my Orchid LED strips are generating the proper photosynthesis wavelengths (PAR/ePAR), so if there is a way to make the Y axis show PPFD (Photosynthesis Photon Flux Density) or any other plant-related metrics, that would be fantastic.
Thank you for this Les! I have always wanted a spectrometer but they were so expensive! I have recreated this for myself. Thanks again!
Awesome, glad you found it useful! :-)
Thank you. Very interesting! Getting ideas on how to study dormancy in spruce needles through spectroscopy. Found an article (Vegetation and Dormancy States Identification in Coniferous Plants Based on Hyperspectral Imaging Data) that gives the most likely band candidates to investigate. Maybe I can record the data to train a model that can identify "sleeping" trees
Gemmologists would love this. The spectra of gemstones can be very useful for identification.
For sure!
Student gemmologist, and cutter - just loaded up my shopping cart...
@@GemFacets my dear friend, because I am also interested in the distribution of precious stones, did it work for you? Can you give me some details?Thank you.
Man, I LOVE your vids. They blow up anybody's mind. Keep on going!
Thanks!
I am curious as to what kind of readings one would obtain by looking at stars through a telescope with this device.
I am also wondering if the camera can be adapted to a wireless esp-cam streaming via UDP.
Your work is spectacular!
Keep doing what you're doing.
Galactic redshifts are on the order or 10s of nm. With a suitable telescope you might be able to measure redshift with your setup. I think you would also need a chamber filled with hydrogen to get an adsorption spectrum? My gr 12 physics is failing me...
Something I would like to get into is stellar spectroscopy, I just need time...I have had a couple of ideas about how this might be accomplished and calibrated. Red-shift would be especially interesting...
@@LesLaboratory
How cool would it be to map the spectra coming from a Sonoliuminesce experiment where water cavitation occurs in standing-wave nodes?
If one were to identify helium in those produced emissions, I think that would be an epic breakthrough.
Thank you so much and congratulations on this beautiful project!
It's really a must-build for me now.
You are welcome!
This looks great! A few cool things to add would be:
1. Average mode for reduced noise (continuous integration?)
2. HDR mode (combine multiple different exposures to avoid clipping the high intensities but also get low noise on low intensities), possibly with log Y option
3. Blackbody calibration (incandescent light bulb?) to get a more realistic absolute intensities
4. Relative spectra (compared to a saved reference spectra to be able to get absorption spectra)
1) Peak hold integrates the max values of all peaks over a period of time, but I guess you are thinking of grabbing say 30 frames worth, averaging then displaying?
2) Like image stacking in astophotography?. LogY sounds interesting.
3) This may be done at some point, but software might not be the best solution. Picams (like many cameras) have poor linearity across the spectrum. The current most sensible solutions are to debayer the camera sensor (not easy, see my other videos!) or acquire a B&W USB camera. Both have relatively linear responses when compared with color cams.
4) A cool idea!
Very cool Ideas indeed! Intensity calibration with a black body would be awesome! That way you could determine stuff like the CRI of LEDs.
@@LesLaboratory 3) Couldn't you calibrate the software by measuring the change of brightness while doubling the distance to the blackbody emitter and determine the change in measured intensity to the theoretical change in intensity?
@@stefanmayer444 for any given individual wavelength yes, but the color filter on the bayer layer results in a non linear spectral response. It should be a bell curve, from about 400nm to 900nm or so, but it is way off and this would need compensating.
@@LesLaboratory Okay, so Ideally a monochrome sensor, but we can still hope for a software compensated solution? Would be interesting to see the accuracy between a monochrome sensor and a software compensated RGB sensor.
This is an amazing creation!
Coming from an audio perspective I love how the same tools of spectrogram and spectrograph are equally revealing and amazing, just at different frequencies. :)
If the software ever developed the ability to determine CIE 1931 xy chromaticity coordinates, along with the ability to calculate the true luminosity in nits, this would let the it be used to match a relatively inexpensive tristimulus colorimeter to the absolute accuracy of a spectrophotometer for display colour calibration.
This match is done by measuring the same “pure” RGBW colour patches on the same display with both meters, and then the software take those two sets of chromacity measurements and use them to match the values produced tristimulus meter which can measure quickly and at low light levels to the spectrometer.
Currently the tools to do this sort of measurement to 10nm accuracy, i.e. good enough for a LCD/CRT/bulb projector are ~$1.5K. And to get to a narrow enough range to be able to accurately measure the primaries of a laser based source would require a $15K instrument.
There is a strong community of amateur home theatre colour calibration enthusiasts (and professional I’m sure) that would be eternally grateful if this window opened for them via your DIY creation. Me included!
Cheers, take care and thank you.
-Jonathon
Wonderful! Two questions: 1) If using a pi camera that does NOT filter out IR, how far into the IR spectrum can the spectroscope & software support? 2) Have you or are you interested in exploring integration of something like ArgyllCMS? I'd love to be able to get the output of >>>spotread with this level of precision and resolution.
The physical spectrometer is designed for visible light, and the deepest you can get is about 800nm or so. The software however does not care too much about range. When I was testing, I was using test data from 300nm to 1500nm, so it should be fine.
You could always roll your own spectrometer head and indeed I am thinking of looking into this myself, there are too many interesting things to be seen in the IR that I am missing!
I have not used Argyll CMS or spotread, and am certainly not an expert in color spaces. What is your application?
@@LesLaboratory It started out as an interest in accurate CRI and TM-30 analysis on home and studio lighting sources - essentially looking for DIY alternative to the Sekonic C-800. That said my interest has evolved to include effects of deep red and IR in home lighting, wellness, etc. I'm interested in UV as well, but that brings with it many more challenges and safety considerations that haven't yet been overridden by curiosity. I'm subscribed and will be following along if you do venture into a wider spectrum!
Feed the UA-cam algorithm with a comment. Great video. Thanks!
Thanks!
holy moly Les your a wizzard have to look at allllll your projects and build them :D thx a ton!
Thanks! You are welcome!
Beautiful! I love all the improvements! Thank you!
Thanks! :-)
Great job! Please keep up the good and interesting work.
Thanks!
Definitely would be worthwhile implementing CSV output for data analyses. Would be very interesting to analyse water quality 🤔
When you hit save, it saves data as a graph image, and as a CSV file :-)
Does it need a high speed pi like a pi4 (I see you have a fancy cooling rig!) or will it run with a low-end one such as a pi zero ? The zero seems better suited to a self-contained unit.
It ill run on at least a 3B+, but I have not tried it on the zero yet. If Bullseye will run on the zero, I would reckon this software would, although I expect it will be slower than on on the Pi4. Since the software now supports full-screen, it might be possible to run it in a minimal desktop environment.
Thanks for the video and the weblinks!
1) Which RPi cam would you suggest - the normal one or the NoIR? I mean, for instance some of the most prominent emission lines of oxygen lie in the near-IR, will I see them with a normal cam? On the other hand, if I had a NoIR cam, would I still be able to properly see all other lines in the visible range?
2) Would it make sense to use a 12MP cam instead of a 5MP cam?
Thanks in advance!
Will start building now. Thanks!
Would it be possible to extend this into the IR range with the appropriate sensor/grating? I have used my Pi IRCUT cam to test some dodgy green laser pointers. With the filter turned off it was possible to see two lines of IR with diffraction. Just beginning my deep dive into optics/lasers so thanks for creating this.
did you see ir
Thank you very much for this wonderful spectrometer!
I was thinking about building a spectrometer for colorimetric analysis for cheese making and earth chemistry and as part of an AES spectrometer - your spectrometer and software are much better than anything that I could have built myself!
There are two tricks from experimental physics that I thought I'd mention in case they were of any interest:
One is to use an "optical chopper" (usually a "chopper wheel" or "chopper disk") and phase sensitive detection to improve the signal/noise when measuring weak light sources. I think modern cameras and a Pi might be fast enough for phase sensitive detection to be implemented in software now though I'm not sure.
Another is to use a reflective diffraction grating instead of a transmissive diffraction grating - that can allow the wider range of wavelengths to be measured since the light doesn't pass through the grating.
Fantastic as always Les!
Thanks!
Excellent job😉 much more refined software.
Thanks!
this is awesome!!! thanks for sharing,man.
Thanks!
For anyone who has built this, did you have to make any changes to the libcamera settings? When I point the spectrometer directly at the FL reference light, I get high-intensity white lines that start to disappear when I turn the meter away from the source. Thanks
Les, this is an excellent series and I'm happy I found it. I haven't read all the comments, but I thought I'd ask about turning the spectrum to a diagonal and taking the pixel samples from that diagonal.
For a 5MP camera you should get:
3240 diagonal of 2592 x 1944
3240-2592 = 648
648/2592 = .25 so 25% more pixels
I will be reviewing your code over the next several days , but I was wondering if you had thought about doing this to improve resolution and/or have any comments about the feasibility? Thanks!
So, you could do this, but... The resolution of the camera is already very high at 2592 horizontal pixels. Of these the software scales down to 800px, and then processes that data. The reason is, 1/3 the data gives a >3x speed up in processing, so we can get a live view. Secondly, but more importantly the spectral detail that can be resolved depends on the optics more then the camera (although sensor size plays an important part). It is not likely you could split and resolve the Sodium D lines for example no matter how many pixels you try and throw at it, so we would just be processing wasted data.
This could all be fixed by designing a better physical package with and adjustable slit, precision ruled gratings, coated optics and a larger sensor. The problem then is that the cost skyrockets.
Well that all makes sense. I'm trying to implement this on a Raspberry Pi 5 and have found that the latest and greatest isn't so great. I can't get the camera to work on it and I'm seeing all kinds of Google stuff about problems with both the camera and underlying graphical system compatibility with older stuff. Very frustrating. Today I will retreat to a RPi 4 and if I can't get that to work I'm going back to an RPi 3. I really like the work you've done and I'll report back on what I get working. Thank you so much for this project.
@@LesLaboratory
Just to help someone else who might have problems with the RPi 5 as I described earlier: I retreated to the RPi4, same camera, same tests AND IT JUST WORKS! While this does not preclude the inference that I've done something stupid, it does point a stronger light on the RPi5 as having camera problems.
Not sure if this is the right place for this question but I’ll give it a shot. On your nitrogen laser low-voltage Power supply to Drive the ZVS board to power the flyback transformer what are you using might I ask? I’m finding power supplies that are producing 0 to 30 V at 20 A is that enough or too little? Whatever information you’re willing to share would be appreciated
i’d like to say that your nitrogen laser build is brilliant. thank you
Thanks! I run the ZVS driver at 12v and 1A, there is no need to go crazy, for this application, you only need 20kV at a few milliamps. I have a ZVS PSU I built that does 30V at 6 or 7 Amps, but it is way to powerful to drive a Nitrogen Laser, all it would do is start a fire! Less is more with stuff like this.
Potentially quite useful
Very nicely done!
Is the vertical scale linear? so that, assuming I find a source with known intensities, would I be able to calibrate the vertical scale as function of wavelength? if the camera gain if fixed, it should be possible.
So love this, and got the software part to run. I see you made a note about max 800 pixels, and it crashes if you try to higher. Why the restriction? is it due to the display limits or a driver problem? I am using a USB monochrome camera on a pi5, and out of the box results are great, just seeing if i can utilize the full res.
@@FernandoMoreiraR There comes a point with the spectroscopes where a higher resolution doesn't gain you much, and the program runs slower so 800px seemed like a good compromise between speed and resolution. It is all open source so you can hack it up to whatever you want.
Great work. Are you using the Pi camera that you stripped the bayer layer? Is there any calibration to the intensity of the light at different wavelength?
I was just using the ordinary version of the camera in this video. Intensity is arbitrary units from 0-255, however it can save data as CSV which you can import into whatever program you like to do post processing
Its pretty good software, but can you also add a wavelength specific intensity measurement function please?
Eventually. I need to characterise the sensors first, then subtract the response curve from the data.
Wonderful idea and very clear presentation, thank you. If I source a suitable camera would the spectroscope and your software be compatible as far into NIR as 1850nm? I see some of the Pi cameras with Sony IMX 219 are without IR filters but I haven’t yet been able to find exact specifications as to wavelengths filter version cuts in. Any subscribers in the know?
OK, so a little more research completed. For anyone looking outside of the visual spectrum a camera without an IR filter is what you need. However, most compatible cameras seem to run out by 1000nm. The ultra low light IMX462 without IR filter seems to be about the best option with a little more width to 1100nm (from what I’ve read so don’t quote me) It’s only 2MP and uses a bigger M16 fitting too, anyone who’s built one, would these be a problem if matched to a suitable lens?
I don’t have any suitability info on the spectroscope. I did email the suggested retailer but had no reply - any scholarly subscribers with a pearl or two of wisdom?
1.Can the spectrophotometer accommodate a cuvette or similar sample holder? Specifically, I am interested in using it to analyze absorbance and transmittance of solutions within a cuvette. If so, do I need to make any modifications to the software and hardware from what is shown in the video, or can I use the setup as it is?
2.Lastly, this might sound basic, but could you clarify what the Y-axis represents on the graph? I understand that the X-axis indicates wavelength, but I am uncertain about what the Y-axis measures-whether it is transmittance, intensity or any other value.
@@praveenragavesh1807 you would have to make a suitable cuvette holder, and build a suitable light source to measure transmittance. The values on the Y axis are intensity which are arbitrary values from 0 - 255. This is good enough for most purposes, but if you want them tied to a specific value, you would need an additional sensor.
Not sure you have mentioned it but the camera may need to be on for 5-15 minutes before calibration to have all parts at operational temperature and also note the temperature of the room at time of calibration
The optical path is fairly short, so temperature should have no detectable effect on wavelength.Sensor gain is another matter...
Amazing work ! Keep playing with spectra and spectrometers - there are amazing things that can be done with them. Just look at all the spectrometers used on space missions and not to mention all the amazing applications they have here on the ground as well. I once saw a hand held spectrometer for geologists that could give you the major chemical makeup of a rock - that device costed several tens of thousands of dollars/euros. But i really think it costs that much mainly because of intelectual property rights and so on, the actual hardware shouldn't be that expressive to build a spectrometer with an equal performance - maybe I'm wrong but that's what I understood. What do you think ? My dream is to be able to built highly sensitive spectrometers like that geologic one that I've described or even better with regular off the shelf components and granted maybe a camera ccd a bit better than average but the rest would be software i guess.
In your experience, or anyone else reading the comments for that matter, do you guys think it would be possible to build spectrometers capable of identifying specific chemicals ? In my humble opinion spectrometry is the future and at the moment sensitive spectrometers are only available to very very expressive research labs but just imagine how much innovation would be enabled by enabling people like you here to build your own very sensitive spectrometers... Imagine having a NASA style curiosity Rover quality spectrometer available to you.. The possibilities would be endless. I can't wait for these kinds of open source hardware information to improve and to become more widely available.
Thank you for your work and for documenting and sharing what you're doing here on UA-cam. All the best to you !
Thanks! I think what you are describing there is a Raman spectrometer if it can identify samples. It is something I intend to look into if I get time. The software is where the complexity will be especially if you want to autodetect, though for Raman setups the optics are fairly critical as well.
Excellent stuff ! I think I shall have a play with this as I have a 10 quid ebay usb microscope that I would happily pull the sensor out of or maybe it might work 'as is ' what do you think ? You have done an amazing job with this and I hope schools and colleges know about it as back when I was a student I would loved to make something like this ! and you get a very capable scientific instrument at the end of it. This has the lot, Lasers (everyone's favorite!) optics, software, electronics so it's all there...cheers.
Thanks! Sure, a USB camera should work with the USB version of the software, and will probably work more or less as is, though you will have to tweak the focus!
There have been a few colleges e-mailed me since the first version last year, but yes, this kind of kit should be in every school and even Universities for that matter!
It is such a pity that the "Amateur Scientist" and other great columns are no longer a thing!
Hi, this is really amazing! Do you think it could be used to determine elements and organic matter in the liquid (water)? Or will more sophisticated equipment be needed? What should be used as a light source, laser, UV or IR? Or more sources asynchronously, e.g. IR for organic, UV for chemical elements...
I think it would take more, work and probably a more sensitive camera. Really for determining composition of materials, you would be looking at Raman spectroscopy or hyperspectral imaging.
Is there any way to customize waterfall sample period or to make composite image automatically? Now waterfall shows only about 30 seconds. Would be awesome to be able to make long composite or set e.g. for 2 hour waterfall image. Thanks
It could be done, but I have questions:
What is the application?
What would be the required update frequency? (Important, since we have limited RAM on the pi!)
@@LesLaboratory To see light spectrum changes during the day for example. There are studies that prove light spectrum is changing throughout the day (of course) and it's highly beneficial to get some of the "specific light into our eyes" in the morning and also in the evening to trigger hormone production that helps us sleep, relax, aware and so on. I would like to set up a little monitor to collect data during the day and compare sunny day to cloudy day, etc. It would be great to have the option to snapshot CSV data on particular times. I would suggest update frequency to be editable. Not sure how it impacts RAM or if it can be stored on the disk to lessen memory consumption.
Would it make sense to have a way to calibrate the sensitivity as well, or there is no reliable reference that will always have the same ratios of intensities between frequencies across the whole spectrum?
Probably. It can probably be done with a tungsten Lamp and some algorithm to normalize the spectrum. Absolute intensity is another matter, and is probably not reasonable for all circumstances.
You specify a zoom lens, but are you actually adjusting its focal length, or just using one setting? I have played around with more than a few RPi cameras. It has always been my experience the M12 zoom lenses are far worse optically than fixed-focus lenses. Its not even close.
Great stuff, I have a spectrometer made from a 3d-printed black enclosure and a usb Webcam with it's IR filter removed. This looks way better! Are there differences in performance between the two?
The two versions shown in the video have comparable performance to each other. The entrance slits in each are the same size, and so the miniature one is just as crisp and bright.
Donated! Wow Les it's incredible to see the v2!! So many nice features. How about CIE1931 display with the blackbody graph and the Rec.709, P3, Rec.2020 color spaces in the future versions?
Thanks so much for your support! :-)
Now that would be interesting, though it ill take a little while to wrap my head around the math for that. What would be the intended purpose? Measuring the color temp of light sources?
@@LesLaboratory In our Video Recording Lab at the University we are teaching students different light sources properties. The CIE1931 in 2D or even better in 3D is our basic graph for understanding the light sources/cameras/HDR monitors wide color gamuts. It would be great to have it as an option on your awesome software - if possible.
@@aviationodyssey1892 A couple of people have asked about this, and I am looking into it. It looks like it might take a bit of figuring out, but at this stage I have the general gist of it. It looks like luminosity data is relative rather then absolute, which solves a headache.
When displaying the data, I assume a simple marker on the Chromacity diagram is what is required? Would it also be useful to print out the x & y coordinates?
Edit: Although I have the gist of it, by the looks of the academic paper I looked at, it will likely take a while to implement it sensibly, but I am sure it ill be an adventure!
Hi-- art conservator here! I'd also be interested in this! I am building this spectrometer to identify gas fills in neon/gas-discharge tubes and would use chromaticity measurements (CIE1931) to characterize the color of the neons/gas-discharge tubes and assess for color matches and color shifting! Thanks so much for this incredible work.
@@taylorhealy7889 You are welcome! It will make it into the next iteration as either part of the existing interface or an add-on. From the requests I get for this software, I might be as well building a modular plug-in system, but we shall see.
super great video Les, great respect. I do have a questin. Do you know if the resolution is enough to detect mercury an cadmiun in food. Hav ea great day, Les. Cheers
Thanks! In order to detect heavy metals with a spectroscope, you would need to heat them to incandescence, and there would have to be sufficient quantity to be able to observe the characteristic peaks. For ppm or ppb values that you would likely encounter in food, you would have to consider X-Ray-Fluorescence (XRF). For qualitative detection is is possible to use Phenyl-,2-phenylhydrazide.
Love your work. Unfortunately I needed something that handled the UVA and UVB range (primarily 311nm and 365nm so this build wasn't suitable for me. Great stuff though. Awesome.
It is possible to do it if you de-bayer the camera and use quartz optics for the Spectroscope.
Hi Les, I am part of a team working on creating a novel toroidal plasma accelerator. We are interested in using this setup to examine the black body radiation of the plasma for our study. However, the spectroscope you are using is out of our budget. Do you have any cheaper recommendations for us? I was looking at either using a "pocket jewelers spectroscope" or a 600 line Diffraction Grating sheet which could be made into a homemade scope with a 3D printed housing. I look forward to any ideas you may have!
If the scope you are using gives a usable image it should work ok, but I have never used one of the cheap ones. I used the patton hawksley scope, because I had one on hand. There is no reason why you can't build your own. With a self build you can get into the IR. If you can afford (or scrounge) the optics, the UV is not out of the question either.
Very cool. 🌈
Awesome build! I need a spectrometer for some time now but always was holded up by cost. I need it to go down to 360nm, is your software going this low?
Thanks for sharing!
The software will, however the spectroscope shown will not. UV requires silica optics for a transmission spectroscope, so you would need a custom build.
Great video (Gemologys )
Have you found any IR filter free camera modules?
There are NoIR cameras available from the Pi Resellers, For cameras with an M12 mount and No IR filter, just look for the Picam night vision knockoffs.
Hi, I am using your software for an IR spectro (with the NoIR pi camera), what should I change in the code to extend the wavelength around 1000 nm ? Thank you very much, your job is awesome :)
The software will handle the new range, but the optics might not. The spectroscope I used was designed for visible light, and has a range that extends to about 800nm or so. I suppose if you got the camera to look into it at a weird angle, you maight get close to 1000nm, maybe.
Very good!
It would be fantastic if you would make your code for chip decap with excimer laser available
2nd guy on youtube who said the word excimer xD cheers to you! I also wanna make my students try to build something like this. And we got excimers!
Tell you what, I will write it up in the style of an academic paper, (might take a while, but will be worth it!) and fire it on GitHub, why not.
What can be interesting is to put linear lens array in front of tube, and get linear array of specters. Then visualize it in 3D. Or even better to create 2D array of spectrometers and visualize in 4D.
If buying a Pi4 for this project, I see there are options for 1,2,4 or 8GB of ram, which made me wonder which is suited best for this project?
A question, how does is it possible to calibrating the amount of power measured by this spectrometer? like for example, uW/cm2/nm. I'm looking for solutions! Thanks in advance!
Folks, I built Les's excellent spectrometer, but the small zoom lens seems to come in a number of variants. The one I have won't let me focus the spectrum with the PiCam. The lens I have looks the same superficially, but has the following spec: F1.6, 3.6 - 10mm (6MP). Can you suggest one that works please?
My suspicion is that the focal plain at the back of the lens is missing the sensor. (M12 mounts arent created equally, some are taller than others) If you unscrew the lens from the m12 mount a little does the focus get worse? If so, remove the m12 mount, and carefully file down the front threaded end of the mount by 1mm, this will allow the rear lens to get closer to the ccd re-assemble and try again. Don't remove too much material in one go, you don't want the lens to be able to be screwed up against the sensor!
@@LesLaboratory THANKS - You are probably right, but feels savage to take a file to a lens or lens mount though. Has my lens got the same focus range as yours?
Yes confirmed. Filing 1mm off has done the trick. Nice soft aluminium... Thanks again.
@@RobScott-l7l Fantastic!
Build a nitrogen pulsed 2 joule per second and made a rough spectrum uv machine that was hooked up to a photo diode and oscilloscope 100 Mhz. Tested many chemicals including cobalt chloride, copper sulfate, hcl, ethanol, ethylene glycol, water, methanol, etc.
Hello. I have watched your video about the Raspberry Pi spectrometer. I would like to know if what is the unit of intensity that your software is displaying. Thank you.
It is arbitrary between 0 and 255.
@@LesLaboratory which will be more proper? W/m² or lm/m²?
Woohoo! I'm excited about this video, and thought you might've put the pispectrometer project aside for the time being. Would you be willing to accept pull-requests on this new repository?
Thanks! I have just been pushed for time with life vs projects :-) What are you considering contributing?
What light sources do you reccomend?
@@CJ-gj7qc For calibration? Any Fluorescent lamp. They have clearly defined spectral lines. See the github repo for more details.
Is the pocket scope any less accurate than the larger one that’s 6$ more?
Nope they are about the same, but small is nice!
HI - looking for spectroscope for NIR 750-950nm for testing IR-led. can this instrument be made to cover that range?
Could you do a teardown of the spectroscope tube?
Have you tried radiometry with this guy?
No. It is possible though, but you need to be aware the spectral response of PiCameras is very nonlinear. This can be fixed by de-bayering the camera (see my other videos) and it is potentially fixable in software.
Hello, I was wondering if you had any idea what might be the cause of this issue. When I try to calibrate, I get an error message that reads, "loading of calibration data failed(missing caldata.txt or corrupted data)". The problem is that no matter how many times I try, the error keeps coming in. If you have any idea what I should do, that would be really helpful (I've already tried creating the caldata.txtx file by myself and it's still not working).
Could u build a uv vis type spectrometer that can measure concentrations? Cool.
A question, what OS are you using? For some reason my raspberry does not detect the camera, I tried with another raspberry or changing the camera, it is not a hardware problem, I think it is a software problem.
Very nicely done! I'm interested in spectra from 340nm and upwards. Is that possible with the current setup? Or does one need different hardware?
Thanks! I have not tried this, and all I can say is, its a maybe, but unlikely.
The problem is that regular glass will absorb significant quantities of UV light, and many glass types are quite opaque at these wavelengths.
You would need to construct a spectrometer head with either Silica optics, or go for a reflection grating setup instead.
Is it possible to get this to work in the ir range?
The software is capable, but the spectroscope itself is designed for visible light only. With a custom hardware build, but UV an IR imaging are possible.
Well ill be dam. Amazing work! Will try to recreate😁
@Les' Lab
What programming lanquage did you use to write spectrometer software?
It's all written in Python, and is available on my GitHub.
Is this capable of outputting CRI, LUX & correlated color temperature ?
Not yet! It is on my todo list, as there seems to be a fair amount of interest in this. I am thinking of implementing CIE-1931.
@@LesLaboratory I looked into sensors capable of measuring CRI LUX, while they are cheap, the PAR (which is by far one of the most important) is quite expensive. The sensors start at 150$.
What is the operating range of your spectrometer?
Can it see UVB light from 280 nm?
What is the lower limit and what is the upper limit?
How well did the OV9281 end up working?
I found the Arducam problematic in terms of drivers etc, that said, that was back on Buster, I have not tried it on Bullseye. If i get time over the holidays, I will spin up a now install and see ow it measures up!
Can we have infrared spectroscopy for fruit quality
possible to attach this to a telescope? take a look at some bright stars, maybe the moon?
For the moon, it would probably work, but for stars, well they are already a point source for all intents and purposes, so a different physical setup would be required, that would dispense with the slit.
That's really nice. I saw the first version. I wonder how difficult it would be to make an IR spectrometer w/a RaspberryPi ... :)
Is there any free software like that for a windows?
I built the software in Python, so if Python is installed, it will run. I have not tried this on Windows however.
Do you anticipate there being a relatively easy way to modify the python script for the USB version, to not crash when using a resolution other than 800x600?
I've been looking for a means to interface with a Thunder Optics SMA-E spectrometer on a raspberry pi (or another SBC, but RPi is most likely to find existing open source support/software) which I can also use with GPIO to stop a Titanium anodization process, inexpensive configurable optical endpointing...
I was able to view an image from this SMA-E (USB) spectrometer via guvcview but I cannot configure the resolution to 800x600 (It is 640x480), nor can I get pyspectrometer2(USB) to run without crashing - I don't have a picam to verify there are no other issues... but guvcview works with it and python3-opencv IS installed.
Thanks for any help you can provide
Can this be used to calibrate monitors?
Probably with modification, though it has caveats I guess. LED monitors would give a different spectrum than CCFL monitors, regardless of the test image being used for calibration.
Hello, if I don't have a raspberry pi, can I use my pc with a USB Webcam?
Yes. It was developed on a PC. Since it runs on Python, so long as dependencies are met, it should be ok.
Merci ;)
Do you sell this hardware yourself?
No, all the stuff is off the shelf hardware and so is already widely available.
H1 is less than H0
There is significant difference in the hypothesis
Hello, I am pretty dumb. I downloaded python to PC. I downloaded your software. I can not get it to turn on. I can get the python screen to come up [ the one that appears to allow commands to be entered] and the html page for python. Somehow I am missing something. Maybe put your program folder in python folder? I am wanting to use pc for a while before I get the raspberry.
Got 70% as a thesis.
Thesis in photonics 15 years ago.
It's a shame that all the important stuff (for my application) takes place between 150nm-350nm
Is it possible to run diy spectrometer on the Raspberry pi Desktop system with a usb camera?
Yes. This is documented in the GitHub repo. Most cameras work, so long as the native resolution is at least 800px wide.