Build A $360 Synthetic Aperture Radar With MIT’s OpenCourseware

A few profs from MIT’s Lincoln Lab are giving those poor MIT undergrads something to do over winter break: they’re teaching a three-week course on building a laptop-powered radar system capable of radar ranging, doppler, and synthetic aperture imaging. Interestingly, the radar system that teams will build for the class has a BOM totaling $360, and they’re also putting the entire class online if you’d like to follow along and build your own.

From the lecture notes from the course, the radio system is made out of an off-the-shelf  LNA, oscillator, and  splitter. By connecting two coffee can ‘cantennas’, it’s possible to record a .WAV file from the signal coming from the radar and use MATLAB to turn that audio signal into a doppler radar.

It’s a very ambitious project that goes deep down the rabbit hole of RF and analog design. One of the lecturers made a YouTube demo of the radar in ranging mode; you can check that out after the break.

[youtube=http://www.youtube.com/watch?v=x040ltNKDFo&w=470]

38 thoughts on “Build A $360 Synthetic Aperture Radar With MIT’s OpenCourseware

    1. If you can do it in matlab you can do it in any other language (meaning you could get hardware to do it, as well). It’s just algorithms.

      The hard part is understanding what matlab is doing, so you can reproduce it.

  1. I love how the instructions show you how to connect the threaded SMA’s and screw things down to the board, yet assumes using an o-scope to adjust the modulator up-ramp time and chirp magnitude and measure the roll off of the filter stage to be common knowledge.

    1. Yeah, that sort of thing sums up my experiences in 3.091 this semester fairly concisely. I believe that it has to do with the vast experience of the professors. They have a hard time remembering what is impenetrable for a newbie, so things slip through.

    1. Don’t forget the casual assumption of things like owning a network analyzer to trim the monopole antenna. You could get close with cheaper things like a resistive swr bridge but in all I think the equipment used costs far more than than the physical bits assembled together.

  2. I’m in the process of leading 8 teams of 6 people to build this now at work.

    Be warned: you essentially NEED access to a Network Analyzer to do this well (read: $60K). Otherwise, you need to measure the hell out of your monopole antenna and hope to god you got the length right. You also need a ‘scope for tuning, but I’ll assume you all have one.

    Be careful with the schematics – some of the resistors aren’t included in the BOM (I don’t remember which).

    Also, the op-amp isn’t made in a quad form-factor anymore – you need to order 2 dual op-amps.

    We’ve already built one completely and looking at data is pretty cool. We’re going add a bunch of stuff onto it once all the teams have the baseline radar working.

    1. I don’t think you NEED the network analyzer, although it would make it easier.

      Folks making cantennas for wifi use wireless adapters and the received signal strength indicator to experimentally tune their monopoles. By setting the adapters to the same frequency as the radar, you can perform the tuning with them, and transfer the cantennas to the radar when you’re done.

  3. This just in. Researchers make worlds most boring demo video.
    The project looks awesome, and I’m always interested in RF black magic so I’m going to spend some time going through all the details, but seriously that demo video was boring. This is one of those places where having the engineers spend some time with the marketing majors would probably be a good idea.

    Otherwise very nice.

  4. Ehh, looks a lot like work to me. This rig, while neat, is kind of going no-where without the processing end of it. But the IF is audio spectrum stuff, not hard and could be done without matlab. Measuring the ramp time and roll-off are trivial if you have ever used an o-scope. I bet you could tune it without the need for the spec-an as well. I mean, it’s faster to do things when you have access to a gazillion in test equipment, but this isn’t rocket science, just the telemetry end of it… And its important to focus on connecting the MW parts together the right way because they are the most expensive! And something tells me that the ending slide is not this actual set up, even though it is a ghostly blob without the background blended in to it, the resolution of a SAR at 2.4Ghz made out of a couple of coffee cans is not going to be that great.

    1. Well, the cans would make it directional, so the trees to the left won’t be a problem and judging by how he “dissapeared” off the graph when he got to the end of the field, those trees are simply out of range.

  5. If anyone is interested in hearing more from Greg Charvat, the guy who created this course, then there’s an awesome amp hour episode where he’s the guest. Lot’s of excellent discussion around RADAR.

  6. This project has a lot of potential for interesting applications.

    Will it work through walls? Can you detect someone on the other side of a wall/door by their breathing? Heartbeat?

    If you put an accelerometer/gyroscope on the breadboard, you can determine the direction and position of the horns. Can you use this and 3-d mapping software to make a radar map of the area?

    How far away will this system detect? Can you detect an airplane? Can you track an airplane using this system? (Or perhaps four?)

    1. No. No and um, No.

      Did you see the guy get about 10 paces away before he disapeared into the “grass”?

      Projects and courses like this are cool for the novelty aspect and playing around with MW, and most importantly learning the basics. But I have issue with the fact it gives people false idea’s and hope that it will be something it certainly is not. But you have to have some perspective on what a real radar is like I guess. But I can assure you, the military or your local airport does not have coffee cans scanning or tracking anything and if they thought this could be used for such things, MIT wouldn’t be making this so availiable to just anyone…

      But I still think it’s a neat idea.

  7. I’m working on a pcb kit version of this project. It will run at 5.8 GHz and is built as an Arduino shield . The first prototype is essentially finished, and this week I’m working on the rev. 2 PCB layout. http://reactancelabs.com/?p=272

    Also, Greg Charvat hosts a web forum for the discussion of people building these radars. http://glcharvat.com/tincan/ I’ve posted some samples of Doppler and range plots made using my prototype.

    A friend of mine is working on putting together a set of Python scripts for the processing end of it and my brother-in-law is working on a windows application.

    -Tony

  8. I’m working on a pcb kit version of this project. It will run at 5.8 GHz and is built as an Arduino shield . The first prototype is essentially finished, and this week I’m working on the rev. 2 PCB layout. http://reactancelabs.com/?p=272

    Also, Greg Charvat hosts a web forum for the discussion of people building these radars. http://glcharvat.com/tincan/ I’ve posted some samples of Doppler and range plots made using my prototype.

    A friend of mine is working on putting together a set of Python scripts for the processing end of it and my brother-in-law is working on a windows application.

    -Tony

  9. I wouldn’t mind building one of these, tweaking the frequency to something a bit more foliage-penetrating, then using it to find Mayan ruins through the jungle canopy of Belize.

    Anyone want to help? :D

    1. My thoughts exactly…connect up a GPS w/ choke ring antenna for millimeter accuracy, and you’d have one quick and dirty SAR-scanning aircraft…How about a helium-filled, solar-powered blimp for continuous surveillance? :D

  10. I have wondered in the past if, using custom software, a standard 802.11 radio could be used as a type of radar when connected to a 24dB grid antenna, transmitting packets and then timing the delay before receiving them back. Now I know what to do with the Anritsu 54177a and Agilent 8757d I have sitting on the floor in the corner…

      1. So, if I understand the description in your link(I cannot access the paywall article linked at the bottom), the system is reading the reflections of several incidentally placed access points, and determining location based upon remote transmissions, rather than reading responses from it’s own transmission?

  11. I actually built it back in 2011. Back then, they didn’t have the bill of materials online, and only one lecture note (but you could zoom into an image of the fully assembled radar kit, and read the labels of the RF modules).

    I definitely have to play with SAR imaging now when more lecture notes and matlab code is available. I only tried the more basic FMCW radar back then.

  12. I looked this over pretty closely and there are some confusing things. First, I have only used chirps short enough to correlation with the echos (optimal filter, pulse compression, goes by several names). This more common police radar type system of direct mixing and relating tones to distance is cool and maybe I am missing something. They say some things like aperture is as long as the track or an aircraft’s path. But it should be the width of the beam. In other words, the distance a plane travels in the time it takes a target to enter then exit the beam. Then there is some confusing talk of resolution. In SAR, the wider the beam, the better the resolution, and resolution is independent of range (the further the target, the more times you hit it – cancels exactly). I would expect better results. Perhaps the cans are too directional. Anyone looked at this?

Leave a Reply to Ryan7777Cancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.