About a month ago I was frothing over a datasheet for a (pretty expensive) PLL synthesizer by Analog Devices, the ADF5355. It's capable of spitting out signals from 50MHz to 13.6GHz - widest I've seen!. It's got a VCO parked between 3.4 and 6.8GHz and internal dividers & doublers. Naturally, I requested some free samples of the $100 AUD (!!) part, to evaluate it - might be useful in a future project.

I then designed a board to house it - for a few reasons. Firstly, great practice for RF routing. Also, a good opportunity to create part in my EDA library so it's easier to use later on. Lastly, the official evaluation board was way too expensive for my student budget (~$600 each). Especially as I want to experiment with more than one!

Here's what I came up with:

And the schematic:

This is essentially as much as I was able to distill down the component count such that the device still works. Once I got the board, I started populating it (see first picture at the top of the post) - unfortunately, as happens more often than I'd like - there were some errors.

Take a look at this:

Yeah, pretty bodgy. Somehow, I managed to swap the pins on the linear regulator ICs, and the LVDS termination for the reference oscillator was slightly wrong. Hacking the board using some modwire fixed it up; and after spending a lot of time writing some C code to bit-bang SPI into the device I managed to get a lock (red LED):

Here's a picture of the setup on a (very messy) bench:

Lab PSU powering the PLL, Tiva Cortex-M4 spitting SPI into the PLL, and RFB (6.GHz+) connected to my HP 8555A spectrum analyzer. Hard to see the spike in the above picture, but it does work!

I then tried a more substantial experiment:

This is just a directional coupler. I've got the PLL at the input (1), a (dodgy handmade in 2mins) monopole-like antenna at the output (2), the spectrum analyzer at the reflected port (3) and a 50 Ohm dummy on forward (4).

I wrote a driver that does a sweep with the PLL across the frequencies near where I calculated the dodgy monopole should resonate at. Putting the analyzer's CRT persistance all the way up I got:

Sweet! obvious resonance around 2.2GHz, and the PLL seemed to sweep great.

I've fixed the errors with the schematic & board made in revision 1, more experiments coming soon!

Recently I've been working on a project for which I needed a 14GHz double-balanced mixer, however I'm on an extremely tight budget.

Luckily, most manufacturers provide free silicon samples these days, and Linear Tech (or rather, Analog Devices these days) is no exception. The LTC5548 is a pretty awesome MMIC, DC-6GHz IF, 1-12GHz LO, and 2-14GHz RF. It also includes a buffer so that you can use LO powers around the 0dBm mark instead of an ordinary mixer which might require.

Unfortunately, the evaluation board for the chip costs hundreds of dollars (~$300!) and I need multiple for my experiments.

So, I've designed my own:

And a render of the PCB:

I'm about to send this off to manufacture. For now, the repo (& design files) is at https://github.com/schnommus/rf-mixer

Once I get it back, I'll make some measurements and post an update!