About Me

My Photo

I previously worked on Virtual Reality and other hardware at Valve.  I currently work at Google[x].

Prior to starting at Valve, I built computer peripherals such as keyboards, mice, and joysticks that were designed to be used inside MRI machines.  My company, Mag Design and Engineering, sold these devices directly to researchers at academic institutions who used them to publish scientific papers in peer-reviewed journals.

After work, I spend time on many different types of projects that usually involve circuit design, machining, material selection, and general fabrication/hacking.  My favorite place to be is my home workshop.

ben dot krasnow at gmail




Saturday, May 7, 2011

DIY Scanning Electron Microscope - Image Quality Improvements 3

A generous benefactor donated some proper SEM apertures to my project. These are much thinner than my brass plate in which I drilled a hole. This should increase resolution as there will be less scattering from the edges of my original, thick aperture plate

The main problem that I am having is that the oscilloscope X and Y amplifiers do not provide enough range or offset to easily control where the scan pattern hits the sample. I knew this would be a problem, so I included small mechanical X and Y stages for the specimen in the microscope. I even designed a vacuum-safe rotary passthrough into the vaccum chamber. The biggest problem is just connecting the rotary passthrough to the stage itself with a right-angle and/or flexible shaft arrangement. Space is very tight and the shaft must move with the stage, making for a difficult mechanical design.


  1. Stainless steel bellows from vacuum valves (ebay).
    You could affix such a bellow on your base plate, with a viton joint, providing a passtrough for a 3D mechanical movement. Rods in the vacuum environment could transmit the movement to the stage.

    A bellow for NW16 valve is flexible enough to provide about 8mm of lateral displacement, and +25mm of vertical movement. Of course, your mechanical arrangement must resist to the athmospheric pressure.

    Alternatively, I bought on ebay two very tiny stepper linear actuators (Ø8mm, 6mm stroke), to move the sample under vacuum. Not tried yet. Probably some degassing...

  2. Another idea :
    A tube through the vacuum chamber, open to athmosphere at both ends. A cable or a string move a magnet inside the tube. A iron ring, inside the chamber, slides around the tube, following the magnet.

    Transmission rods to the stage are necessary, to avoid magnetic perturbations of the electron beam.

  3. Have you considered using a frame store to scan the image into? Then you could read it out as video into a monitor. That way you could have effectively infinite persistence.

  4. Hey Ben,

    I write for Popular Science magazine and would love to talk to you about the microscope project. Any interest? Let me know.



  5. Greg, sure that sounds great. Please send me an email: ben at magconcept dot com.

  6. I thought about the FPGA stuff
    you should get one with static ram (SRAM), because its much easier to program than dynamic ram(DRAM)

    an vga port is not a must for the board, as you can build an greyscale/green only 8 bit vga adapter quite easyly with a resistor ladder (i have sample code/a build instruction for an altera cyclone board), but if you want true color, you should get one with a video DAC (108mhz for 1280x1024) onboard.

    some boards have fast enough (5mhz per channel minimum x 10 bit or more) general purpose analog outputs, but they might be on a i2c or SPI bus, which is not trivial to program.

    you might as well build your own DAC boards with an parallel interface which you can plug into the expansion headers, which most boards have (3.3v i/o)

  7. bricodada, thanks for the suggestions. I will let you know what I end up doing.

  8. This is terribly interesting! I second the proposition that you upgrade your scan frequencies to 60 Hz / 15.something kHz and make a B&W video signal. Oscilloscopes are good for signals but most implementations I've seen of scope TVs are pretty terrible.
    If that proves too high you should try to run it at the SSTV (ham radio) frequencies, there will probably be some equipment on ebay to convert that to normal frequencies.

    Also not sure if anyone commencted but the reason you got black lines in the picture is because the 60Hz AC/30Hz vertical frequency beat together (probably the electron flow stopped during zero-crossings).
    It's something TV designers struggled with a lot in the early days and also the reason NTSC is actually 59.97Hz.

  9. no holm, the black lines are simply printed on the osci screen.

  10. I was referring to an earlier post where he discussed AC/DC for the filament supply.

    I thought about it and it would be possible to use an AC filament provided the vertical scan is synchronized with the mains and ran at 60Hz or a multiple thereof.

  11. @holm i didnt get the part with the ham radio frequency? why? to stay in the legal band with hf emissions ?

  12. It was mentioned in the second image quality post that he believes image quality can be improved by scanning slower, SSTV transmissions are a good starting point since designs for scan-conversion to NTSC/other higher quality display devices should be available.

    That way he could use a TV or a capture card to record pictures rather than the oscilloscope.

  13. After Maker Faire, I plan to get started with FPGA programming. My first project will be to build a scan generator and digital acquisition device to capture images from the microscope directly to the computer at any scan rate. This way, I can use a fast scan rate to focus the image, and a slower rate to capture higher resolution.

  14. The reason that image quality can be improved by scanning slower is because signal-to-noise ratio is improved - if you dwell on a pixel for a longer time, and then average the signal over that time frame, any random noise averages out. In this manner, you can use a smaller spot size and get higher resolution without sacrificing SNR. I usually dwell on a pixel for 100us for stuff that I want to show people.

  15. btw. are you using linux?

  16. Simon, no I'm still stuck in a Windows world.

  17. Ben - This has Open Source project written all over it!! I'm sure I'm like half of these guys who looked into building/buying an SEM but never took it much further than that...

    Would be very interesting to split this into its major components and let the community create an Open Source SEM similar to what has been done with 3D printing machines and Open Source.

  18. Hash, yes providing documentation is definitely my intention. The cheapest SEMs are currently $70k, but I think they could be built for much less, with some compromise in quality. I am currently working with some people on creating full, and complete documention. Stay tuned for a major update.