SDR Sample Rates: How fast is fast enough?

Discussion in 'Amateur Radio News' started by VK6FLAB, May 11, 2019.

ad: L-HROutlet
ad: l-rl
ad: Left-2
ad: L-Geochron
ad: MessiPaoloni-1
ad: HRDLLC-2
ad: L-MFJ
ad: Left-3
ad: abrind-2
  1. VK6FLAB

    VK6FLAB Ham Member QRZ Page

    Foundations of Amateur Radio

    SDR Sample Rates: How fast is fast enough?

    If you measure the voltage at the base of an antenna and record the readings, you end up with a collection of numbers that represent the voltage over time. These numbers, or samples, can be used to represent the antenna signal inside a computer.

    An antenna system voltage is an example of an analogue signal, continuous over time, the recorded readings, the samples are an example of digital, discrete and intermittent.

    It's possible to reconstruct an analogue signal from digital samples and that's exactly what Software Defined Radio or SDR is all about.

    The process of sampling essentially converts a continuous signal into an intermittent one. As recording separate samples implies, there is loss of information in this conversion.

    For example, if you sample once a minute, you'd represent a continuous signal as 60 samples per hour, probably enough to reconstruct where you've driven in your car along the highway, but hardly enough to reconstruct the route through the middle of the city, let alone represent an antenna signal that varies millions of times per second.

    So, how often do you need to record a sample?

    Turns out that if you sample at least twice as fast as the highest frequency you're representing, you're good to go.

    So, for sound, the human ear can hear about 20 kHz, so more than twice that, explains some of why a CD is sampled at 44 kHz.

    If you want to represent the 20m band, up to 14.350 MHz, you need at least a sample rate that's double that, or 28.7 MHz.

    As an aside, there are other ways to look at this problem. If you managed to move the 20m band down to 0, then you'd only need at least a sample rate of 700 kHz to do this.

    Let me say that in a different way.

    The width of the 20m band is 350 kHz. So sampling it would require at least twice that, or at least 700 kHz. Moving frequencies around is something that we've been doing in traditional radios for a long time. The technique uses one or more frequency mixers. This means that combining some traditional radio tools with an SDR gives you even more options.

    Truth be told however, this idea of moving the band with one or more mixers is becoming less important as technology improves and there are plenty of reasons not to use this. I'll talk about that at another time.

    So, the first takeaway is that to sample a continuous signal and be able to represent that signal accurately requires a sample rate that's at least twice as high as the highest frequency in the continuous signal.

    Without going into the actual proof of this, consider a sine wave that oscillates at 1 Hz. If you sample it at anything less than 2 Hz, you'll end up with some cycles being sampled only once, which isn't enough to represent the sine wave. If you sample it at exactly 2 Hz, you'll have two samples on every cycle, but if you happen to sample when your signal is 0, all you'll ever measure is 0. By sampling at a rate greater than 2 Hz, you overcome that limitation.

    I'll make brief mention of another phenomenon, that of over sampling. An interesting thing happens if you sample twice, three times or more than the minimum sample rate. In short, the higher sample rate improves the dynamic range, noise performance and filtering, all very useful when you are processing radio signals. Cheaper and cheaper hardware are making this very attractive and it explains some of the reasons why SDR manufacturers are using sample rates that far exceed double the highest frequency being sampled, for example, the Flex-6600 samples at 245.76 Mega Samples Per Second, or Msps, even though the maximum receive frequency is between 30 kHz and 54 MHz.

    In case you're wondering, yes, I'm leaving out a lot of detail here, one thing at a time.

    The opposite, under sampling, has its uses as well, but I'll also leave those for another time.

    The second takeaway is that higher sample rates are used to reduce cost, increase performance and reduce component count.

    Some of what I've talked about can be explored with the popular RTL-SDR USB dongle which is actually a mass produced commodity digital television receiver, made in the millions and accessed directly thanks to the combined efforts of many different people. If you'd like to start to play, $25 should get you a dongle and most of the software you can start to experiment with is free. Check out to get started.

    If you'd like to get in touch, please do,

    I'm Onno VK6FLAB

    TL;DR This is the transcript of the weekly 'Foundations of Amateur Radio' podcast - for other episodes, see

Share This Page