Saturday, June 4, 2011

DIY Scanning Electron Microscope - Operation procedure

After getting back from Maker Faire (which is always a hugely enjoyable and inspiring event), I thought that my microscope might need some repairs. As it turned out, I only had to change the filament and tighten some screws that came loose during the trip back from the Faire. The microscope works just as well as it ever has -- I didn't even need to move my alignment magnets. I made this video to show everyone what using it is really like.

Also, if you haven't been able to attend Maker Faire yet, it really is as amazing and epic as you have heard. The intelligent and inspiring people who make it happen are a large part of the motivation that I had to build and display this microscope. In turn, I hope my project inspires others to create things and share their ideas with everyone. There's no better way to have fun and celebrate accomplishment at the same time!

35 comments:

  1. Definitely awesome seeing it actually work!! I can't wait to check out the documentation, projects like this are a blast to build and get running. If you need a set of eyes to review documentation let me know, more than happy to help.

    ReplyDelete
  2. well time to get started on fpga programming, isn't it;)
    Ben have you already made up your mind about what sorts of equipment you will need? let me repeat one thing. go for sram onboard and not dram. it will save you a lot of trouble. how is this thing going to be connected to your computer?

    ReplyDelete
  3. Simon, I've started on the digitizing project. I think I will eventually use an FPGA, but after initial research, I was trying to find an alternative, since programming FPGAs appears to have a very steep learning curve. I've never used one before, and there aren't nearly as many helpful resources on the net as there are for microprocessors.

    My first plan is to use the FPGA to create analog VGA or DVI and display the image directly on a monitor. That almost seems easier than dealing with a computer interface, but I am not sure. The real benefit of digital acquisition is that I can scan slowly and get a much better signal/noise ratio.

    ReplyDelete
  4. I have currently worked out how to generate VGA 1280x1024 video. The tricky part is to set up the phase locked loop circuit in the chip to generate an internal clock of 108 MHz which is the pixel rate and divide this up to generate LINE and FRAME (H/V) sync signals. (Using a set of counters) You then have 2 registers which contain the x and y coordinate of the pixel being output right now and you feed this as address input into a ram array and have a very basic framebuffer. I can give you the schematics, the verilog code , the Altera Quartus project file and some explanations if it helps. You would of course not use the 108 MHz for scanning but fill the framebuffer once and then display it repeatedly. It is a lot like programming C with the difference that everything you code will run at the very same time so you have to take care to have the right timing scheme. I'm not a guru myself but have mastered the VGA port anyway doing trial and error. The best thing is, once you have set up working code, you can keep it in place and just let it do its work and work on something else because there is so much space on todays fpgas and they can be easily splitted into functional blocks and thats quite fun.

    Just don't use complicated protocols, they will eat up all your time. If you want to do VGA out, use a simple parallel interfaced D/A chip or a simple resistor network whose output is quite acceptable. If you want memory use up the one in the FPGA itself first because its easy to use (just like an array in C) then switch to outside SRAM if necessary which requires a little bit of bus handling and timing stuff. Then if you want A/D conversion stick to a converter which has an parallel output and CLOCK IN or something because its much easier than the ones being accessed over some SERIAL 2 wire protocol, which requires (de)serializing signals and stop bits and stuff like that.

    Start with a scan generator consisting of a kind of fpga breakout board and 14 bit parallel D/A and OPAMP soldered onto a PCM with an .1 inch header which can be plugged directly to expansion port of the FPGA board.

    The most important things about FPGA boards (imho) are 1.and usb programmer onboard or shipped with the kit
    2.onboard sram
    3.a lot of i/o lines with 3.3 volt standart on a 20x2 .1 inch header (can use flat cables for interconnecting your modules (almost the same as IDE cables but with 40 PINS (IDE has one dead pin))
    4.good software for programming (altera has the free quartus which i like for some reason i think its esay)

    I have purchased the FT232H which could be good for interfacing with the PC but unfortunately i destroyed the onboard rom so i cannot say how good it really is. But the FT245 was quite good(only 1Mbyte/sec though)

    ethernet can also be done using very little
    fpga4fun.com

    well quite a long post i hope it helps

    ReplyDelete
  5. PCB is the word, not PCM // and "1.an usb programmer" // not 'esay' but easy

    ReplyDelete
  6. Simon, thanks very much. I anticipate a steep learning curve, but I really do want to get some FPGA experience, so it's time to dive in.

    ReplyDelete
  7. Hi Ben. I hope this is the right place for a post concerning SEM ... Just for info, I have started to build my own SEM. 1st step will probably be 'leak hunting' because I don't use a Bell Jar but a glass cylindre (H=600 mm / OD=250mm / Th=5mm )with Viton gaskets and a top lid in ... PVC. I hope not having degassing problems, but what can be interesting is that it's a good electrical insulator as all electric connections will be entering by this lid.

    ReplyDelete
  8. Ravaner, that's great! PVC might be OK if the temperature is kept low. I used acrylic in my vacuum chamber, and it only presented a problem when it was too close to the filament and got hot -- causing outgassing. Keep everyone informed of your progress.

    ReplyDelete
  9. Hello Ben. I don't want to monopolize your blog with my own project and so may be will be better to exchange through emails, mine is ravaner@free.fr. State of progress : I have install 12 (homemade) feedtheoughs on the PVC lid of my glass cylinder and, according to my capabilities of pumping, I reach 1E-5 torr which I consider as not too bad. As I've a quite long experience of tungsten filaments with my homemade TOF mass spectrometer, I want to test a Field Emission Gun. I know that this need an higher vaccum and very sharp tip, but at least for the fun I want to test that. Every solution seems to me better than tungsten filament (How many times I replaced mines ? ). Keep in touch for next progresses.

    ReplyDelete
  10. Ravaner, that sounds interesting. Feel free to keep posting comments. I am sure everyone would like to know if you are able to build a field emission gun, which would be quite an achievement!

    ReplyDelete
  11. Wow, incredibly impressed. We need to start a community around this project dedicated to the advancing of Open source community SEM's.

    I and the whole internet community is in awe of your work, and the incredible speed at which you got it functioning. There is much work left, but if we all work together this will become an indispensable tool of researchers worldwide who don't have access to a commercial SEM. In addition it will advance the teaching of the sciences to millions of students who would never have had the opportunity to use a SEM. Researchers, developing nations, institutions and schools worldwide are in real need of this.

    Let's start a community project page where we can consolidate idea's and information, I have a feeling we will see this expand on every front if a community with a common goal exists. There is much to work on, and if we all work together who knows how far it can go! Much of the work done can also carry over to other similar projects, the support hardware for the photomultiplier and high voltage supplies have millions of uses, and any digitizing circuits/software can also have other applications.

    CERN has just announced they are starting to try an opensource hardware model, there is some very useful hardware we can use there, such as regulated high voltage supplies. Open source/community hardware/software is about not reinventing the wheel, but having a global community work on a common goal for the advancement of science, which is what science is truly about.

    Are you opposed to making this a open source hardware/software project?

    ReplyDelete
  12. Alexander, sure I think this is a great idea. I am working with a book publisher right now to get many project details down on paper in a coherent form. I am a little too busy right now to also take on leadership of an open-source SEM project, but I would be happy to contribute. Let me know what you had in mind.

    ReplyDelete
  13. Hello, some news from the front ... Very, very basic test, I have generated an electron beam without any focusing system (Wehnelt not in use) and I've gotten a very very soft blue glow on my screen (ZnS + (Ag)). It is a beginning ... As I mentionned previously in my former rig (Mass Spectrometer) I used 40 µm tungsten wire as filament which is unable to keep an inverted "V" shape. As I don't want to produce 4A current with a 20 kV insulation, I'm using a completly crazy solution ... My filament is in fact the one included in a E10 light bulb of which I've broken the glass bulb. And it works ... I've created a special system which allows to perfectly center this filament according to the hole in the whenelt and the axis of anode and other electrodes. It's just a beginning, but it makes me optimistic for next steps ...

    ReplyDelete
  14. Hi Ben, amazing work. I'm a grad student that has been challenged by my professor to construct an SEM on about a $200 dollar budget not including the vacuum pumps. I am fairly certain that the only components I can't make are the photomultiplier and x-y scan generator. I was wondering how you produced your scanning voltages for your deflector plates. It would be a great help if you could steer me in the right direction.

    ReplyDelete
  15. Jon, I used the circuitry from an analog oscilloscope to generate the deflection voltages. By feeding the scope a 0-5V sawtooth wave, the scope will output -150 to +150 deflection voltages. Depending on the size of the scope's screen, the voltages might be higher or lower. You can find cheap analog 'scopes for under $100 on eBay, or even less at local electronic swap meets. I built my own 0-5V sawtooth generator circuits -- one for X and another for Y.

    ReplyDelete
  16. Thanks for the tip. Hopefully my university's surplus has some more analog o-scopes that I can pick up for little to no money.

    ReplyDelete
  17. Hi Ben, my SEM is in progress, my concern for the time beeing (and not the least) is that I get very small spot, well focused but ... accompanied by a stable track which looks like a spider web (not easy to explain). Did you meet this kind of default ? Do you know the reason ?

    ReplyDelete
  18. Ravaner, I am not sure what you are describing. Are you firing electrons at a phosphor screen? The spider web pattern appears on the screen? If your raster generator is working properly, you should have a square or rectangle of uniform brightness. You can defocus the spot to make the uniformity in scan pattern easier to see.

    ReplyDelete
  19. Sorry Ben, I was sure thay my explanation was not clear enough. For the time beeing I just check the quality of the spot (shape, brightness) on a screen. That means that deflection plates are at a fixed voltage. To describe what I get on the screen : a bright round spot at a corner of a (almost) triangle or diamond stable and less bright.

    ReplyDelete
  20. Ravaner, you might have some electrically charged dirt near your electron beam. Make sure the beam is shielded with a metal tube so that external fields do not affect it, and also make sure the tube is clean, and free of cracks or discontinuities. You may also have an aperture that is dirty or has a ragged (with burrs) hole. Your electron gun may have dirt or burrs that are affecting the roundness of the beam. If everything is clean an symmetric, you should get a round spot on your screen.

    ReplyDelete
  21. Hello Ben. Construction is in progress. Spot is now well swept. I've developped an ADC board connected to my PC through parallel connection. In a first time, i've tried to work with sample induced current, but in fact I get noises from X & Y deviation plates (saw teeth) and ripple from high voltage electrodes so I've to do as you did and develop an Everhart-Thornley detector. A question about that : how did you connect HV to crystal, metal coating ? annular connection ?

    ReplyDelete
  22. Ravan, I used an aluminum annular ring around the scintillation disc. The shape of the ring should accelerate electrons toward the scintillator disc to cause flashes of light. Good luck! Keep posting your project details.

    ReplyDelete
  23. Hi Ben. Did you cut scintillator you used by yourselves ? With common amateur tools ?

    ReplyDelete
  24. Ravaner, it is a P47 phosphor-coated disc that I probably bought from http://www.tedpella.com/scntilat_html/scint.htm

    ReplyDelete
  25. hello,

    i m on the way to complete the page mentionning your sem on my french website dedicated to homemade particle accelerators:

    http://www.accelerateurs.fr/semamateurs.htm

    thanks again to share your amazing project !!

    ReplyDelete
  26. Hello. As I got many problems in operating PMT, I leave the E.T detector for later. For the time beeing I have installed an electron multiplier as backscaterred electrons detector and it works. I get my first images on my PC. Quality is very poor and one need a part of imagination to recognize the crossing wires on the mesh I use as sample, but anyway it's a beginning ... I must precise that the best (less bad !) results are obtained with only 5 µA beam current. All that is really not an easy game.

    ReplyDelete
  27. have you looked at http://opencores.org/project,vga_lcd

    ReplyDelete
  28. Congrats on your writeup/article in Make Magazine!

    ReplyDelete
  29. very cool! I was thinking of doing a SEM project like this for quite some time now, that's how I stumbled upon your project. going to explore this further. Thanks

    ReplyDelete
  30. Do you think you could do a video on how the power supply for the electron gun works and powers the filament.

    Clark

    ReplyDelete
  31. Clark, the filament power supply is an unregulated DC power supply built from a variac, bridge rectifier, and smoothing capacitor.

    ReplyDelete
  32. Good to actually get to meet you at Maker Faire yesterday.

    Did you ever get around to running this with an FPGA? If not, I have a collection of FPGA boards around here that aren't currently in use; I suspect that you could get much better image quality with some simple image processing techniques (in particular, deblurring may go a long way, and dynamic contrast enhancement would probably work well, too).

    If you're interested, and have some free time, send me some mail -- I'm joshua at joshuawise dot com .

    joshua

    ReplyDelete
  33. Joshua, thanks for the offer. I haven't built my FPGA raster generator / image collector yet, but I think that I should really do it at some point -- there are so many projects and so little time!

    ReplyDelete
  34. Hi Ben. I got always troubles to find the right place for my posts ... Anyway my question is about the voltage you used for your Everhart-Thornley detector. Is it really 10 kV or less ?

    ReplyDelete
  35. Ravaner, I used about 7KV for the ET detector acceleration voltage and 0-500V for the cover screen.

    ReplyDelete