Open Hardware for small, largely-software-defined FMCW microwave radar altimeters

[Luke Weston]

Hi guys and girls,

Matjaz Vidmar's radar altimeters (http://lea.hamradio.si/~s53mv/avnr/amodules.html) are extremely cool, very interesting, but also really challenging to reproduce since there are no CAD/EDA files, it uses discrete microwave GaAs HEMTs (Agilent ATF35176s, to be specific) which are near-impossible to source, and it uses distributed (microstrip) elements in the RF design for the mixer, VCO etc. which will require the use of the same Teflon PCB laminate with the same Er.

This looks interesting, but it's immature and not documented in a completely friendly, happy fashion. But it's still understandable, if you know what you're looking at, though.

https://github.com/loxodes/mr_radar

The IF output from the main 'Ferret' radar board is then plugged into the AFE board for amplification, filtering and AD conversion, and then the digital side of the ADC is plugged into the STM32. And the STM32 is also plugged into the digital side of the DAC on the main board that generates the VCO sweep voltage. At least, as far as I can tell, that's how it's supposed to end up working in the end.

The microwave stage uses discrete off-the-shelf "blocks" for the directional coupler, gain blocks (amplifiers), the mixer, VCO etc, instead of using distributed microstrip elements and discrete active components. This has the advantage that the RF stage can just be fabricated on a cheap, common 1.6 mm thick FR-4 fibreglass PCB and still work happily.  The fact that the board was designed to work successfully on standard FR4 makes things much easier and cheaper to reproduce compared to annoying PTFE or alumina substrates. :)

It might be practical to substitute the GALI-55 microwave amplifier ICs on the board with MGA30989s, which would be appealing since they're available at Digi-Key, and having every component known to be available through a single major distributor like Digi-Key sounds like an appealing choice for easy, convenient access to one-shop components for reproducibility.

However, the MGA30989s have a fair bit less gain... which might limit the transmission power, signal-to-noise ratio and resolution

Greg Charvat's MIT IAP2011 radar class project is kind of cool... very simple and easy for anybody to reproduce since you're mainly just screwing together pre-made modules for the whole RF stage. But those MiniCircuits connectorised modules come at a relatively high price, they're relatively bulky, and it would be nice to be able to extract simple information such as an altitude using only an embedded microcontroller without the need for a whole PC.

http://ocw.mit.edu/resources/res-ll-003-build-a-small-radar-system-capable-of-sensing-range-doppler-and-synthetic-aperture-radar-imaging-january-iap-2011/projects/

But still, if you've got a few hundreds of dollars spare for hardware, you just want lab-bench demonstration electronic hardware not rugged spaceflight hardware and you can figure out how to replace the functionality of MATLAB using code you've written with free software then you can follow the above instructions and put together your own flexible software-defined radar in a weekend, which is pretty fun.

Also, the open-source friendliness of Charvat's design is let down a little by the need for MATLAB in my opinion... it would be really nice to have the same functionality reimplemented in some free software so experimenters who don't have the luxury of a university's site license aren't left with little choice aside from software piracy.

Both the above designs use a 2.4 GHz local oscillator - this relatively low frequency probably isn't optimal, since it will require relatively large antennas and during development and testing it's likely to cause severe interference to 802.11 and every other common thing that uses 2.4 GHz radios.

It could be - probably should be - changed to a higher frequency, by changing the VCO, changing the filters to the appropriate centre frequency, and ensuring that the amplifiers and the mixer are capable of operating at the higher frequency - it is likely that the latter couple of devices can be kept the same for operation across a range of different frequencies, and the amplifiers and mixer will function across a wide range. For example, the Minicircuits MAC-60+ mixer will operate from 1.6 GHz to 6 GHz.

This one is also very interesting and promising... although they haven't shared any schematics or anything openly, at least not yet.

http://reactancelabs.com/?p=293

Personally, I'd lose the Arduino and integrate the AVR into a single board, on the same board as the microwave stage but away from it.
This looks really good, especially since they're sweeping from 5.4 GHz to 6.0 GHz, which means it is less likely to cause as much interference with other communications compared to the 2.4 GHz spectrum. Eventually though, when prototype working hardware is built, I suspect I'll just ultimately need to get an advanced amateur license or something. :)

I'm really waiting for the above guys to actually post their promised open hardware, component details, PCB layouts and schematics with some real detail, though, so I'm not feeling around it in the dark making educated guesses. :) They promise me it's coming soon. :)

I've also been working on drawing these up (no PCB artwork yet):

https://github.com/lukeweston/RadarAltimeter/blob/master/RadarAltimeterRFboard-schematic.pdf

https://github.com/lukeweston/RadarAltimeter/blob/master/USBRadarBaseband-schematic.pdf

The USB baseband board plugs into the IF output from the RF stage and amplifies and filters it, and digitizes it into a PCM2902 sound class device chipset - so you just plug it into a PC and it appears as a USB sound card in other words, with everything conveniently integrated into two boards, with good driver support for that audio chipset across platforms, and proven Linux support :) The whole thing is probably able to be powered just off the USB port too.

Of course, this is currently just built for 2.4 GHz (it's derived from the "ferret" board linked above) and it relies on a PC to perform the digital signal processing and visualization - it doesn't have an embedded microcontroller capable of finding a vehicle's altitude autonomously, which is clearly where we really want to go.

These schematics are highly immature, rough, new stuff that really needs a whole lot of review and refinement - keep that in mind! :)

Anyway, this is just a dump of things I've been working on, thinking about, or other research I've been reading :)

Cheers,
  Luke