Comments on: HackRF SuperCluster

By: Oleg Kutkov

Oleg Kutkov — Sat, 07 Oct 2023 09:15:19 +0000

In reply to Baraka. RFSoC Ultrascale+ or even USRP radio is way out of the budget of the hobby project.

By: Baraka

Baraka — Fri, 06 Oct 2023 23:39:39 +0000

If you knew that you could have achieved all of this without much of the hassle using the Xilinx RFSoC Ultrascale+ FPGA, would you consider it? If so, how would you do it? BTW, in case you are not aware this RFSoC would have removed all synchronization issues and given you 8 high speed ADCs and 8 high speed DACs, plus high speed PCIe, which would have allowed you to demodulate your signals like you wanted to do.

By: Oleg Kutkov

Oleg Kutkov — Sun, 03 Apr 2022 10:20:34 +0000

In reply to KG5DXG.

Yeah. I thought about the Lime, but it is more expensive.
I bought my HackRf boards for $90, and I already had some boards at the start of the project. That’s why I stuck to these boards.
Maybe I will switch to the Lime after the war.

By: KG5DXG

KG5DXG — Sun, 03 Apr 2022 07:25:11 +0000

LimeSDR might be another SDR board to try in this configuration, as it has USB 3 and much much higher bandwidth than the HackRF for the same price (61.44MHz + USB 3 for $300 vs 20MHz + USB 2 for $300).

By: Oleg Kutkov

Oleg Kutkov — Tue, 30 Nov 2021 20:01:51 +0000

In reply to Andrew.

Hi Andrew,

I will check the qspectrumanalyzer algorithms, it sounds interesting.
Thanks!

By: Andrew

Andrew — Tue, 30 Nov 2021 19:45:01 +0000

This is really amazing work! Have you looked at all into the interleaved sampling algorithm software like qspectrumanalyzer uses to combine multiple sweeps into one wideband spectrum? It splits the bandwidth of each sweep step into 3 parts (+ guard bands). Working up in frequency you’d have:
the lower band edge (thrown away)
the lower piece (used)
the middle piece (around the DC spike) – thrown away
the upper piece (used)
the upper band edge (thrown away)

the lower piece, middle piece, and upper piece are all the same bandwidth. Then it interleaves the tuning frequencies so that on the next step, the lower piece covers the center piece of the previous step that was thrown away (to eliminate the DC spike), the middle piece of that scan (which is thrown away) lands on the upper piece of the previous scan, and then you get one new slice. Somewhere in the documentation, it goes into a good explanation of how the sampling works, though I had to read it a couple of times to understand what they were getting at.

Obviously here rather than tuning steps on the same radio, you could do the same thing with each hackrf in parallel

Also, the HackRF has a hardware sweep mode that allows it to sweep a large frequency range without individual tuning commands being sent at each step which makes it’s sweeps really fast – great for simple spectrum analysis, but this is clearly superior (a bunch of synchronized hackrf’s in parallel) to do signal analysis.

Really interested in following the future updates!

By: Ivan

Ivan — Tue, 30 Nov 2021 13:09:32 +0000

Cool project! I repeat it at the moment. I have already bought everything, it remains only to configure! Thank you so much for the work done and for being open to other users!