That's very impressive and I'm even more impressed if you can manage to sell tiles at that low price.

PA looks suspiciously similar to SE5004L. I just needed some for my own projects but every distributor is out of stock. I wonder if this is where all of them went?

That's Leela Zero (plays Go instead of Chess). It was good for its time (~2018) but it's quite outdated now. It also uses OpenCL instead of Cuda. I wrote a lot of that code including Winograd convolution routines.

Leela Chess Zero (https://github.com/LeelaChessZero/lc0) has much more optimized Cuda backend targeting modern GPU architectures and it's written by much more knowledgeable people than me. That would be a much better source to learn.

It was from JLCCNC.

There's not much software. PC calculates PLL register values for correct LO and source output frequencies, and transfers them to the FPGA which writes them to the hardware. FPGA accumulates the downconverted samples from the receiver ADCs. Samples are transferred to PC which calculates ratios of the receiver outputs to get uncorrected S-parameters. scikit-rf Python library is used for calibration and plotting.

Any possibility of seeing the code (on github)? Seems interesting.

Thanks for the submission. Author here if you have any questions.

Not a technical question, just want to say I really appreciate your projects! I work with a group of radar engineers and they were absolutely floored by your €800 C-band polarimetric SAR drone.

Any tips for working on hobby RF/microwave projects and staying motivated while doing them? I am an EE and have a long list of projects I want to do at home, but I can never seem to find the time to do them.

I have also had issues with motivation. I had a long pause from about 2019 to 2024 where I didn't work on personal electronics projects because I didn't have enough motivation. This and last year however I have been very productive and it's hard to say what exactly changed, just felt like working on them.

I'm surprised you're not using SMA edge connectors designed for 15G. The ones you're using, with the long center pin, aren't really 50 ohm above 6G. Higher freq SMAs have really tiny center pin for SMT soldering with low capacitance.

Makes sense

I work in retail but most 5G devices have these awful TS9 or similar connectors which we have to adapt to something usable like SMA

I literally have customers returning products as faulty because “the connector just will not sit correctly!” - even though it’s performing as intended

I just want to say thank you for your work. I bought one for the lab last year and I love it. Maybe even saved my job.

Do you have a link to the files on github? I assume https://github.com/Ttl/vna2 is the older version?

It's not on Github, but I do have schematic at the end of the blog post. vna2 is the previous version.

Can you please create a high speed oscilloscope next? :-)

Depending on what you mean by 'high speed', this project may be of interest: https://www.eevblog.com/forum/crowd-funded-projects/haasosco...!

The FPGA manufacturer has characterized all their package pin delays. It's possible to export a csv file with internal delays of the package for each pin from the FPGA design tool. With Xilinx Vivado it's just File -> Export I/O Ports.

Does KiCad allow to take these delays for each pin into account automatically or do you need to do all that manually?

Edit: it's using the pad property of "pad-to-die-length" if doing it manually, right?

No, it needs to be done manually. It wasn't as tedious as it sounds though. Most of the pins are very close in delay already and there were just few traces that I had to adjust a little.

How is kicad getting along with RF work in 2024, would you say?

I did some prototyping but never got the stage if actually fumbling around with PCBs a few years ago, things seemed to progressing quite well.

I actually didn't use any expensive test equipment, only oscilloscope and multimeter. Even design and simulation software was all open source.

Expensive signal analyzer or spectrum analyzer would have been useful, but they aren't absolutely necessary. It's possible to use the radar itself for many tests and debugging.

I have tried to limit the projects I do on my own to only the equipment that I have home and open source software.

Fantastic work! I enjoyed reading the previous posts as well, very interesting.

I built it to see if I could. I didn't have any particular use case in mind.

Would this be useful for another SAR setup?

Putting ground planes on top and bottom layers isn't usually done with high speed PCBs because components are there. There would need to be a cutout on the ground plane near every chip. High speed signals really need continuous ground plane and ICs, especially RF ICs, need short access to ground. Second layer is the best layer to minimize the distance from ICs to ground plane.

That's a mistake in the text. You're correct that layers 2 and 5 are ground planes.