Use Tx port on SDR for an LO?

Question

Designing a receiver for an experiment. The receiver will be in remote location and I will be deploying multiple in the field so have a cost-effective, self-contained system is a priority (is it ever not?).

I have an SDR I want to use that is ruggedized/weather proof and has an embedded linux OS, but its upper frequency limit is 6 GHz.

I plan to look at frequencies in the 6-9 GHz range and thus have some additional front end RF components ahead of the SDR's antenna port.

One thing of note is the SDR will have a Tx port that will be unused.

Instead of sourcing an LO from a company like Wenzel, which can be expensive, could I feasibly just use this Tx as my LO? Any drawbacks to this idea?

Answer 1

Instead of sourcing an LO from a company like Wenzel, which can be expensive, could I feasibly just use this Tx as my LO? Any drawbacks to this idea?

I regularly recommend that to customers of one of my clients, and it works. It definitely has the advantage that you have one clock source, so there's no mutual drift of frequencies, no reference oscillator phase noise adding up, and so on.

Now, it depends on what your external RF frontend needs as LO, but generally, you should be able to generate an LO that just mixes 6 to 9 GHz down to, say, 2 to 5 GHz by generating a 4 GHz tone, for example.

Two things that are advisable:

1. Since SDRs are not necessarily extremely clean in output spectrum, it's often necessary to add a bandpass filter to make sure you don't get harmonics and subharmonics as well as synthesizer spurs around your LO. That BPF doesn't need to be very narrow, though!
2. Make a sketch of the spectrums of your signal of interest, the RX LO + harmonics, and the TX LO plus its harmonics (doesn't have to be based on actual harmonic power, just to know the frequencies). Try mixing the harmonics with each other and add their difference and sum frequencies to the same diagram. Make sure you don't put intermodulation products unnecessarily close to your signal of interest (of course, the whole point is to put one specific intermodulation product at exactly the center of your band of interest, i.e. mix down with the sum of SDR RX LO and frontend LO). Sometimes a little change in both LOs can put spurs out of your band – for free, without any other effort.

Answer 2

Most SDR’s are HDX so it would require more ports and FDX capability with TCXO and excellent isolation and filtering in the mixer. Racks using this method with SDR’s significantly raises the cost.

But can you use two inexpensive SDR’s with one as the ultrastable LO with a step down IF converter but at what CNR, or Noise threshold and Stability for 30MHz BW at 9GHz is the question ?