In the last blog post, describing my journey in building an 8 element phased array antenna system, I found the hardware and software ecosystem that Analog Devices provides is well suited to a crawl->walk->run prototyping experience. Start small, and then scale up to production. This blog post is about the final phase of that development: Run!
Why Do We Need a Phaser?
After tinkering with my phased array system, I was now up to 8 antenna elements. It’s not a 1000 element array, but at least I was “walking”! By this time, others were also interested in building a phased array prototype starting small and scaling as their needs develop.
Analog Devices’ Customer Office Solutions (COS) team, led by Mark Thoren and Trisha Cabildo, took on the task to turn our earlier prototypes into a manufacturable phased array exploration platform. The goal was to provide an affordable vehicle to learn and experiment with beamforming--as well as a springboard for developing larger and more complex systems.
Back to X-Microwave
In the previous blog post, I used X-Microwave RF prototyping hardware to build the 8 element array. We want to use those same blocks but configure them differently to build a prototype of the Phaser. The hope is that a working prototype, very close to the final solution, would save us the time and money of having to revise the final production board.
All the X-Microwave blocks could simply be unbolted and reused in our new setup. Rearranging those blocks, and further upgrading some of the hardware, gave this:
For those that like to see block diagrams, here’s what that looked like:
This was a mockup of what would become the Phaser. It is only a 4 receive element array, but it added in a transmit path and a programmable ramping chirp generator. The transmit and chirp generator allow the Phaser to perform FMCW radar – something we’ll cover in a future blog!
Did I change the software? No! I mean not really. We added the ADF4159 to the device tree (that’s a configuration file that tells Linux systems what hardware is connected to it). And then pointed the program to use the ADF4159 driver instead of the ADF5356 that was used previously. But these were easily done thanks to the IIO ecosystem. So, we again had a new phased array prototyping working. But now it was time to optimize performance—and that meant more changes!
Phaser’s Performance Goal
Up until now, we hadn’t really considered the overall performance of the system. It was working well enough for simple antenna beam steering demos. But radar is a bit more demanding, and we wanted to be able to offer something in the order of 50 dBc of spurious free dynamic range (SFDR). That means we didn’t want any other tone within 50 dB of the signal. 50 dB is a power gain of 100,000x. So, any unwanted tones would need to be made very small!
Frequency Planning
There’s a lot of stuff in any RF design that will give you unwanted signals much greater than 1/100,000 of the main signal, primarily mixer spurs, RF leakage, and component non-linearities. And we had some of those! Here’s what the signal spectrum looked like on our current design:
The only signal we wanted is that one in green. But we got all those other spurs too… We needed to attenuate them until they are 50 dB down from that 10 GHz tone. You can always filter to get just the signal you want. But that only works if there is enough separation between the signal you want and the signals you don’t want. And we didn’t have that space. So, the first step was to create a frequency plan that pushes these unwanted signals far away from our desired signal.
These unwanted signals fall at predictable places, but there are a lot of them. And they move for each combination of input, output, and mixer LO frequencies. I followed the design advice found here, and arrived at a frequency plan consisting of:
RF Output: 10 GHz
LO Frequency: 12.2 GHz
IF Frequency: 2.2 GHz
And now I had this:
The number of spurs didn’t reduce by much, but now all the spurs were far enough away from my signal of interest that I could use low cost filters to attenuate them.
Filtering
We had to select and place filters in key spots throughout the transmit and receive chain. This is another great reason to use the X-Microwave prototyping system. Using it, we could easily try out a variety of filters, move them around in the signal path, measure the results, and make changes. A normal RF circuit board doesn’t allow you such flexibility! Ordering a handful of filters, we bolted them in, tested, and repeated. After an afternoon of measurements, we arrived at a filtering combination that gave this:
So that main tone is more than 50 dB (almost 60 dB) above any other tone! And this is the frequency plan, and filtering scheme, that we used for Phaser.
Conclusion
We’ve now manufactured more than 100 Phaser kits. And they all use the IIO software and hardware design results from this prototyping journey. There’s much to say and many to thank, but that’s in a future post. My hope is that this story presented a way for you to crawl->walk->run to production on your next project.