I describe electron beam optics and deflection as they are employed in scanning electron microscopy. web.jfet.org/vclk/ - Deflection amplifier www.johngineer.... - CRT Christmas tree www.fei.com/ - FEI Company
Ben - you can use a parallel port drive into a simple R2R Ladder DAC to be able to speed up your scan dramatically. I did a video on this. Skip ahead through the R2R tutorial and start watching at about 11 minutes in: Basics of R-2R Resistor Networks for Digital Analog Conversion, Tutorial DAC Thevenin Superposition This will be MUCH faster than the PWM outputs. I show an XY example where a more complex XY drawn figure is flicker-free on my scope.
People say you can't build an electron microscope at home. But Ben didn't know and just build it. You are the man! I love your videos I get a lot of inspiration from them. keep up the good work.
Awesome. Makes me wonder if a TV tube electronics could be converted. It'd have to but put in a vacuum chamber of course. Now I'll have to look back at your previous electron microscope videos.
i guess im randomly asking but does anyone know of a way to get back into an Instagram account?? I was stupid lost the account password. I love any tricks you can offer me.
@Edwin Leroy Thanks for your reply. I got to the site thru google and I'm trying it out atm. Seems to take a while so I will reply here later when my account password hopefully is recovered.
If there was a metric of average cool new things I learn per video, your channel is pretty much right at the top of the list. I've never seen, heard of or even contemplated the possibility of spinning vanes fast enough to literally bat individual molecules out of a region of space. Looking into that, I learned the term 'mean free path', which is something that I had wondered about but didn't have the vocabulary to look up. At the pressures you are achieving, the air molecules are likely colliding with nothing but glass and steel 90% of the time. I'm sure the pump is very well balanced, but i can hear it spinning up. Do you have to do any vibration damping or isolation to avoid affecting the microscope? I imagine that new toy is going to do great things for your sputtering projects as well! Thanks again for an awesome video...
While I definitely enjoy the compliments, remember that there are lots of qualified and talented hardware engineers in the world -- many of whom are far more accomplished than I am -- I just happen to make videos about my work.
Applied Science but communicating this science and technology is sometimes just as important also, would you be available as a guest on my little podcast?
JRoque250 "simple terms" :) While I am able to follow him when he talks about things like PWM and voltage follower, I doubt the average person would call it "simple". Regardless, he does do a fantastic job of explaining everything.
Applied Science humility is good but "lots" is a relative number. Not so many pple can design and build this kind of thing on their time and money. Now the mandatory compliment is done, can you please tell me how you machined the pinholes ?
I've watched almost all your videos now. You are a very clever dude! Keep up the great work. It's the people like you that make youtube great. Thank you.
Great video and great topics. It is fascinating me how this hightech equipment like an SEM can be constructed by hacking together some simple equipment and your skills to do this. Just amazing! But I have some other questions: Why do you turn on the turbomolecular pump so late? Systems with build in turbos just turn it on with the rough pump as far as I know. Do you think this waiting is expanding the lifetime of the turbo pump significantly?
You could use ADV7123 DAC chip from Analog Devices. It is very fast 3 channel 10 bit video DAC. Hook it up to your FPGA board with some soft core CPU and you will have fast and configurable XY(Z) deflection signal generator (of course if you are ok with 300 MHz bandwidth :). These chips are not that expensive. Or you can find them on many AMD graphic cards.
An FPGA demo board with a fast DAC will give you fast and precise timing (no jitter). Digilent has a selection of such boards, with S/W communication libraries and some demo HDL code. Down load Xilinx's ISE webpack for free and write the code in Verilog. I don't work for either company but I use their products.
I'd turn the circuit into a pseudo-differential amplifier with two separate common emitter power stages, each driven by an op amp that corrects for nonlinearity.
High information density on this one. I will have to re-watch a time or two to parse your explanation. This is a feature not a fault. thank you for your detailed explanation that is just at the right level.
Ben, removing the top transistors will greatly reduce the bandwidth and slew rate. Scopes use a cascode circuit where the top transistors operate in common base mode to eliminate miller capacitance feedback.
This is a good point. I didn't know the cascode would help bandwidth so much. The gain in this system is about 100, so the Miller capacitance does become a noticeable problem. I guess adding another transistor is so low cost, and appears to improve performance in every axis, why not do it?!
Why not upgrade the electron gun with one from a CRT like the common 0.28mm dot pitch monitors? And you even get 3 guns per CRT for the RGB dots or stripes. Maybe even saving most of the CRT neck and using the existing monitor scanning for scanning a sample using the deflection coils? Some even have dynamic focus for the slightly different distance across the tube face. Just greatly reduce the picture size and add some current sensing for the sample.
you dont need a uC with a true DAC to make good stuff, you could use a good uC with a high resolution PWM (usually with a very speedy clock and lots of bits). i work with power electronics, and have been successfully implementing power converters with digital control and very high resolution PWM that gives me fine control of variables with the TIVA TM4C123G uC from TI. it is exceptional piece of hardware (80 MHz, 32 bit, floating-point single-precision) and very easy to program... im sure you would like it!
You can probably get more speed and accuracy from your driver circuit by using an opamp to feedback sense the real voltage on the output and correct for it. Oh and if you are planing to do digital acquisition this is a good place to use a FPGA where you can easily grab data from a high speed ADC at 10 to 200 megapixels per second, buffer it and then directly show the live high res image on a VGA monitor or freeze it and send the image to a PC (You can get few Mbit/s links with some COM port emulation chips to transfer the large image in a reasonable amount of time). I can offer help with the FPGA side of things.
Keep it simple - use a low cost CPLD or FPGA with a R2R DAC followed by a output amplifier. With a simple Xilinx CPLD you can make a very high speed binary counter - simply vary the clock speed for the beam scan raster frequency. You'd need one more electrode near your electron gun to modulate the beam though if you want to draw images .A XL9500 family chip will go to 50Mhz easily for only a few $$. Using a microcontroller will potentially limit scan speeds due to processing time.
do you remember those old video cameras that used vidicon tubes?....i do. try using the magnetic deflection yoke from one of those? sometimes they even include a focus coil. also, they're spesifically built to provide a clean magnetic feild.
There are some STM32 micro-controllers with a built in DAC, they aren't the friendliest to use, but very fast! (for instance the Cortex M4 series) There should be at least a few Arduino compatible micro-controller boards that have a built in DAC, or you could possibly use a very fast IC that works through SPI (or maybe a parallel interface, depends on the precision you need)
Nice. But i hope the magnifercation will greatly improve as last time you didnt get as high as of that of a light microscope which made use of this project pointless as a traditional microscope could have done the same job.. But as a technical exercise and a continualy improving project i must admit its looking good thus far.
Could you keep the beam unfocused and scan a shadow mask. Baird., tempted to make one, optical scanning with UV ? I tried to make an SEM using a high definition CRT TV electron gun and had an interesting introduction with vacuum . had to shelve it . this was when we had CRT tv's ..sold the diff pump and pirani gauges but just got a turb mole with controller under £100. and now need gauges. Its a wonderful world.
hey, very nice build.....looks almost like one i saw in the archive when i was a kid. it was an alvorst lmp-4500.....about 175 (+\- 25) years old, but a good instrument in ti's day. although, your sensor is a bit more complex.....it's that phosphored disk that's a little over complicated. the problem is this....ambient light will cause interference. the lmp-4500 never had that problem...no phosphor disk! try this.....take the photocathode off that photomultiplier tube and feed the dynode stack direct.....trust me, it'll work a total treat! although you could use that one with the black-coated dynodes, it should give better performance than that phosphor disk rig you've got right now. if i could send some drawings i would.....my great grandad's been at this stuff since long before my dad was even born (and i'll be 52 this april!) l8er!....;)
You need something to isolate vibrations from the turbo pump, in this respect the dif pump was better. I would also use 32bit IC dacs for beam control.
Nice! I didn't think you'd ever do another SEM Video This is super helpful, I was going to use triangle wave generators and some amplifiers to try and get a nice scan. That turbo pump is epic, 500 mtorr to 4e-10! Do you notice any differences in the behavior inside the chamber vs the oil vapor present with a baffled diffusion pump setup?
I'm planning to make the acquisition all digital. I'll use a microcontroller to generate the raster, and collect data with the micro's ADC, and have the whole thing controlled over USB. The bandwidth is a concern, so it will probably have to be high-speed mode for realtime video (focusing would be difficult without at least 15 fps video). I haven't noticed effects of diffusion pump vapor -- it may take a while to build up. Someone said that the deposits in my Penning gauge were caused by pump oil, but I'm not sure. I can see why the industry moved to using turbomolecular pumps in general. They are much better in every axis -- except cost!
Applied Science If you can generate a 15kHz horizontal raster, you can make the beam timing equivalent to NTSC or PAL video, then you just need an old analog video capture card.
I also heard that with the BT878 chipset, it captures also the front and back porch, and the vertical blank, in that case you could run the vertical and horizontal traces at some division of normal TV50/TV60 rates and delivering the full optical field spread out over multiple sequential capture-fields.
Could you use a similar concept to steer ionized molecules in your sputtering chamber (not sure if they are naturally ionized by the process or not)? Maybe pass through a small aperture to provide a manageable beam size? It occurred to me that if you could focus and steer that material you could 'print' shapes without using a stencil (at the cost of time most likely)
What size deflection are you aiming for? I would guess 10mm square would be plenty. Your sputtering experience will be useful for coating non-metallic samples. I want to see an ant or fly!
When you move on to microscopy, how will you capture the return from the sample? Will you use a phosphor+PMT or is there some more direct way of acquiring an image?
The traditional approach that I've already used is the Everhart-Thornley detector. It uses a phosphor screen that is mounted onto a photomultiplier. This allows the phosphor screen to be highly charged to accelerate electrons toward it, and then allows the PMT to operate with an isolated electrical system. There are also other vague reasons about why a phosphor screen coupled to a PMT is better. It definitely works, and is still state-of-the-art, however I want to try using a direct electron multiplier to detect the signal. This would allow the machine to operate without a cover over the bell jar in full room lighting.
Hello, do you know of DC/DC converter component that could output more than 10 kV DC from input of typical table top power supply (0-20 VDC). EMCO CB101 goes only up to 10 kV DC.
Hey Ben, I was trying to do a similar project but I can't seem to find a p-47 or YAB scintillator that's under like $110. Do you know of any place I could get a cheaper phosphor, or if there's a way I could buy the phosphor material itself to coat a piece of glass with?
You need some parallel DACs like R-2R resistor ladders. I created a detailed video on this achieving a sampling rate of about 3MHz: DAC using R-2R resistor ladder Important is to write 8Bit ports that are completely mapped to pins of the Arduino directly not using arduino digitalWrite. Unfortunetelly the Uno hasn't two complete 8 Ports. Candidates on the Arduino Mega are PORTA, PORTF, PORTK and PORTC. You can see the mapping here: arduino.cc/en/Hacking/PinMapping2560 A demonstration how I have used 3 ports (X/Y + sound) a once from the Mega can be seen in this video (code can be taken from the description): ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-nTC3vJzc0MY.htmlm9s But in difference to my synth I would rather use use opamps like in the vid above to get best precision. Disconnecting the Arduino from USB reduces the noise significantly. Looking forward to see your results
R-2R dacs are very fast, I can get VGA output out of them, but beyond 6 bits they no longer work OK. Also they tend not to be too linear, even with precision resistors.
Daedronus the only slight problems is saw were with values close to 0 when not using opamp. you can clearly see in the first video clear distinguishable steps on the oscilloscope even on 8bits. I think you could even use more than that. Also to get clean output you need reasonable decoupling. the more complex your circuit is. the more noise is induced. But let's see what solution Ben will find :)
Hi Ben; I heard about redpitaya dot com recently and thought this might help in your build. Still seems to be in prototype stage, but once available could be pretty amazing. HTH
How do you even make an electron beam? I know it might sound like a stupid question but I have no clue how I would go about it. Maybe make a plasma and try and access the loose electrons that way. Secondly, why are there spark plugs in the vacuum chamber?. Once again I apologise that they may be stupid questions but it would be great if someone could help me out in trying to understand this stuff. ;)
Ben, there is some good information about DAC interfacing at uChobby.com www.uchobby.com/index.php/2008/01/08/arduino-audio-dac-options One thing you can do is change the PWM frequency to something much higher. I used 60KHz PWM to do an audio range DAC. With the higher frequency you can set your PWM to analog filter for about 60KHz and sweep around at maybe 30 or 40KHz.
Hay Ben, maybe you could think about using the Red Pitaya (redpitaya.com/) for analog control/data acquisition for the electron microscope. I think it might be a good fit for high speed control of it.
Actually, yes. I wanted (and still) want to make a linear accelerator, and one of the interesting things that can be done with it is make a nuclear "energy amplifier". It doesn't require the fissile material to be critical -- it relies on protons from the beam to keep the fission reaction running. I think thorium is a good material that will drop below criticality if the beam is shut down. Unfortunately, the accelerator itself is already pretty absurdly hazardous, and the activated thorium is even worse, so the whole project is pretty extreme. The easiest way to get nuclear power in a small package is a radioisotope thermoelectric generator. So then, all I would need is some plutonium. Got any sources?
