Scratch-built Software-Defined Radio

[Ben] is showing off some results from his Software-Define Radio project. The board seen above, which he designed from the ground up, is receiving a WWV radio broadcast. This is the atomic clock signal from Fort Collins, Colorado. The audio heard in the clip after the break is a bit noisy, but since he’s about 2000 miles from the origin of the signal we think he’s done really well!

The seed for this build was planted in [Ben’s] head back in July when he saw [Jeri Ellsworth’s] SDR project. He’s posted some of the build details up in a forum post. The approach is similar to [Jeri’s] but there are several key differences. He’s using a DS1085 programmable oscillator where she chose an FPGA for that purpose. Once his hardware demodulates and filters the incoming signal, a PIC32 does the rest of the work and outputs a PWM signal to an Op-Amp to generate audio.

11 thoughts on “Scratch-built Software-Defined Radio

  1. Thanks for posting this!

    I just have a quick correction — we originally intended to use a DS1085, then were set straight by someone much more knowledgeable than us on the topic. We ended up using an Si570, which is the oscillator on the board in the video.


  2. Someone please correct me if I’m wrong, but it doesn’t sound like there’s much software in this supposedly software-defined radio. Okay, he does FFT and iFFT in the PIC32, but still uses lots of other analogue hardware to pre-process the signal, including “demodulation”. It sounds like it’s got some flexibility, but not as much as a full SDR i.e feeding IF into a high speed ADC and doing everything in a DSP. Or have I missed something?

  3. Until recently high speed adc chips were too costly for amateur sdr use, and two local oscillators in quadrature, i.e. ninety degrees out of phase, were used to down mix the incoming RF into an I (in phase) and Q (quadrature) pair of audio signals that could then be digitised by the adc in a stereo sound card and processed by software. This can make for a very cheap bare bones sdr, I.e the ‘softrocks’ by tony parkes, or the more upmarket flexradio sdr rigs, and also the self contained sdr-cube to be found at http://www.sdr-cube.comwhich uses a dsPIC to do all the heavy lifting with minimal latency or PC required.

    Some folks have been using a softrock to digitise the IF of an existing conventional analogue superhet radio transceiver and do things digitally from there (I.e. using powersdr-if software), and this is probably the closest thing to the setup posted.

    As adc chips have become cheaper, direct digitisation of the RF coming in is possible, such as with the high performance software defined radio (HPSDR @ and others, such as the universal software radio peripheral (usrp) offering bandwidth and modes limited only by the speed of the adc chip and CPU.

    Happily, most of the sdr work being done is open source and coders and hackers are most welcome to join in.

    Some of the most exciting stuff is in the realm of using the sdr as a server on the network and connecting to it with a client, I.e. use your TV as a remote head for your radio or another sdr someone has hooked up to the net, such as with the ghpsdr3 code currently being developed.

  4. It is mostly there, the input signal (at 20MHz in this case) is downconverted to a more reasonable frequency, then fed into the A/D converter. He clocked his a/d at 96KHz so there is about 50KHz of bandwidth available, which when combined with the initial downmixing stage lets him look at any 50KHz slice of spectrum up to 133MHz. As long as the signal of interest has less than 50KHz of bandwidth this radio should, in theory, be able to tune to it regardless of if it is modulated CW, AM, SSB, or even a digital mode.

  5. @Ian An absolute full SDR would be a antenna output going to ADC & then everything in SW (People have done it, atleast for simple MW/SW radio) However, more practically complete SDR would be RF preamp/LNA going to ADC. Another lower abstraction would be what you said, Antenna followed by RF followed by mixer/down converter (+ IF filtering + IF amp) going to ADC. Still all these are SDR. Even this could be an SDR although even at a lower abstraction.

  6. Hi,
    Is my understanding correct that fast A/D sampling rate is helping the development of SDR ?
    If so I suggest you look at the latest PIC24FxxxGCxxx and dsPIC33EP products which include a new type of ADC called “pipeline” A/D converter much faster than SAR converters able to reach 10 to 24Msps ;=)
    I am currently developing on the PIC24FJ128GC010 (beta samples) which has a 10MSPS / 12bits ADC ;=)

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.