Sampling rate for a 200 kHz signal

Question

I'm using a rsp1 to sample the output of an underwater acoustic transducer which has a resonant frequency of 200 kHz (with a 500 Hz bandwidth) . The transmitted signal is a CW pulse which can vary from 0.1 ms to 5 ms.

I'm using the transducer fed into a balun I've made to correct for impedance matching. The transducer is 'listening' for the return signal from the transmitted pulse return (same as sonar, or the same concept as radar).

I want to be able to take a sample every approximate 1 cm of distance in the water, which (due to speed of sound in water) works out to be 75 kHz (so I'd either use 96 or 48 kHz due to soundcard output limits).

Am I correct to assume that I need to use the lowest bandwidth possible (200 Hz), tune the Centre frequency to 200 kHz, and set the output rate for the IQ data to either 48 or 96 kHz?

Or have I totally misunderstood the concept of the SDR?

---

Additional Info

Thanks for your replies and input. The RSP will be sampling at a very high rate, software is then used to digitally filter, mix (i.e. frequency down convert), demodulate (i.e. CW, FM) and provide frequency agility. The are a number of different transducers that can be connected to this device, each has a different frequency, but they all have very low bandwidth (typically 200-500hz). Frequencies are generally 28Khz, 38Khz, 50Khz, 75Khz, 100Khz, 120Khz, 200Khz.

Looking at the project from a time domain (like radar), I'd start sampling at the beginning of my High Voltage TX pulse (which is CW, 200Khz as an example), and capturing for a finite time (as an example 500ms). Until the cycle repeats. This allows me to capture any reflections (i.e. targets) from transmission to time X.

The output rate I'm referring to is the PROCESSED output rate after the RSP1 has been fed into an appropriate pcb like a RasPi, Beagle, PC etc. Cubic SDr software etc. offer IQ output, but only at rate applicable to sound cards i.e. 22K, 44K, 48K, 96K. The software cannot output at flexible or custom sampling rates.

If the signal is envelope detected, then the sampling rate only needs to be very low, as the signal is basically low frequency AF. But I'm not doing that, I'm sampling at high rate, digitally down converting, demodulating and hopefully outputting the demodulated IQ ADC samples.

I'll draw something up, hopefully it clears it up a bit. In the mean time here are some Oscilloscope samples of a TX pulse (approx 800vp-p) and the received signal of a target 2mtrs away in the water.

And to clarify, yes a 1ms sound pulse in water will be approx 1.5Mtr wavelength. 0.05ms gives you 7.5cm and so-on. So am i correct in saying if i am sampling every 13.333us (75Khz), that gives me a 2cm resolution?

Answer 1

I don't know the RSP1 well enough, but you probably can't set the output rate to something as low as 96 kHz (the data sheet suggests the sampling range is 2Msps to 10.66 Msps). That's not a problem, higher sampling rates will just represent the same signal with more sample points; the information content is the same and you can resample your received signal to your heart's delight in your PC.

Also, I'm not getting your 75 kHz of bandwidth argument: you say your transducer has a bandwidth of 500 Hz. So, that's the frequency range it can cover. Your transmitter hence only needs to change something in that bandwidth.

Your receiver gets a signal who has, seemingly, due to the physical setup, more useful bandwidth (I don't know what "sample every approx 1 cm" means, and I'm from a radar background!), 75 kHz you say. But: your soundcard has nothing to do with that. I also don't see where you have a 200 kHz signal! That only seems to be your transducer's resonant frequency, but you want to modulate that, so there's probably signal around 200 kHz. But then, 75 kHz makes even less sense.

So, I really can't answer your title-implied question. You're not giving us information that adds up to a helpful picture for that!

What I can do, and I hope that's more generally useful, is the advise on how to proceed:

You need to draw (sketch) spectra. For example, draw the PSD of the signal that should enter your transducer from the soundcard, and the spectrum that leaves it. Sketch it for where you observe it, and what goes into the RSP1. It might be very helpful to also make a sketch/drawing of your system (as block diagram? Or as physical setup?) to communicate it with others who have less experience with what you want to achieve (which is honestly not clear to me).

After you've done that, answering the question "which sampling rate do I need for the two involved sampling devices (soundcard & RSP1)" becomes easy.