Project update 1 of 1
Over the past six months, we’ve made significant progress on sSDR. Two intermediate board revisions have been built and tested, and the final production-revision samples are now in manufacturing.
Early measurements on the Rev1 board showed solid performance up to 8.5 GHz, but we wanted to extend the range into the X-band. This led us to redesign the RF matching network and move to components optimized for 11 GHz and above.
Rev1 used the LMS8001B (high-linearity variant). In Rev2, we evaluated both LMS8001A and LMS8001B. With its integrated LNA and PA, the LMS8001A lets sSDR pick up much weaker signals and deliver significantly higher transmit power.
Two graphs on these plots representing the LMS8002 up/down converter and the LMS8002 bypass mode.
We also discovered that some SBCs expose only a single PCIe lane (notably the Raspberry Pi 5) and pairing that with a 7-series FPGA limits PCIe Gen2 throughput to around ~52 MSPS MIMO. Since PCIe Gen3 is now common across most platforms, we upgraded to the modern UltraScale+ family with 12.5-GT/s transceivers. This upgrade not only unlocks higher PCIe bandwidth but also enables future support for 10GbE and even over-fiber operation.
The XCAU7P also offers roughly three times more logic resources than the XC7A35T while supporting full x4 PCIe connectivity. To take advantage of all four lanes, we switched to an M.2 Key M interface. The extra bandwidth benefits both complex FPGA processing and GPUDirect.
Initial Rev2 testing is complete, final design tweaks are in, and Rev3 is now in manufacturing.
In parallel, we’re preparing the sSDR crowdfunding campaign to go live soon.
Stay tuned—more updates are on the way!
sSDR is part of Lime SDR Accelerator
Lime Microsystems LMS7 RFIC · LMS7002M
Lime Microsystems LMS8 RFIC · LMS8001A
sSDR is part of Qorvo RF Accelerator
sSDR is part of AMD FPGA Playground