Homebrew software defined radio, Jeri Ellsworth style

Posted on Tuesday, June 14th, 2011 in RF by the machinegeek

Here’s Jeri Ellsworth’s vlog showing her early progress on a homebrew software defined radio. Anyone who has considered SDR prototyping knows how complex this can be. We can’t wait to see the finished product.

UPDATE: Jeri has posted updates Part 2 and Part 3 to the above vlog.

This entry was posted on Tuesday, June 14th, 2011 at 11:07 pm and is filed under RF. You can follow any responses to this entry through the RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.

12 Responses to “Homebrew software defined radio, Jeri Ellsworth style”

  1. Drone says:

    “Anyone who has considered SDR prototyping knows how complex this can be.”

    Not true… A direct-conversion SDR front-end (which is what this project is) is one of the simplest RF receiver projects. It is also very easy to do, if you do your homework first.

    This person is not doing it right at all. Her analog switches are poorly chosen, there’s no harmonic filtering on the front end, her local oscillator is exhibiting nightmarish phase noise, and the software she is using is out of date. I can say however, her USB sound card looks to be high quality, which is important for this type of project. But I strongly suspect this is by luck only.

    To see from simple to more complex examples of how to do this properly. Jump to here:

    A better, more up to date version of the software she is using can be found here:

    From the looks of it. Her effort to tack on some analog circuitry to make the receiver work stand-alone exhibits a clear lack of understanding of what is required. She should concentrate on getting the converter to work with the canned down-loadable software first, then crack some books before taking the next step. An example of a very high performance stand-alone capable SDR in development can be seen here:

    Regards, David

  2. Brian says:

    I am going to partially agree with David, this whole approach is going to have terrible phase noise compared to high Q narrow band IQ mixing. I find ultra low power analog a lot more interesting than the SDR stuff, which more or less is give me a lot of data so I can do it with math instead of well designed analog hardware.

  3. Vis1-0n says:

    Interesting subject, been googling it on the background these past few weeks.

    I guess she is decoding the signal on chip, so I guess that bit is a bit complex – no PC! The protoyping should not be, the software would.

    At the simplest end, $8 prop, . Apparently it can be that simple – a starter project to decode AM radio, forget about performance for now. Amazing chip based on specs and what has been achieved one way or another with it ( USB host even) – although I have not used one.

    SDRCube is an example of a pc-less design that uses the dspic33.

  4. Drone says:

    I”ll reiterate: Doing what the Author of the video referenced in this post is pretty simple, provided you do your homework before picking up a soldering iron. This was not done IMO.

    But going to a stand-alone receiver (and transmitter, DC-SDR radios are symmetrical in topology) is quite a bit more complex.


    I’m assuming the phase noise in the referred project is pretty bad from the WinRad display shown on the PC in the video and what seems to be a crystal-based digital clock oscillator used as an LO. But other things can cause lousy noise skirts in the spectral display such as poor power supply ripple (rule-of-thumb for this kind of DC-SDR receiver, use a well filtered analog supply or better yet a battery supply, for reference), ground loops, there should only be one ground source in the whole system, this includes the antenna ground, so the input RF filter(s) should be transformer isolated.

    The Softrock receivers referenced in my OP use either a low phase noise crystal oscillator or the very low phase noise frequency agile Si570 digital clock generator part. An Analog Devices DDS part is also suitable for an agile LO, provided you have a good anti-alias filter following it and understand how spurious products are generated from a DDS over the frequencies of interest. So in-summary, the requirements w.r.t. phase noise include a low phase-noise LO to start with, a single grounding point, and a very low noise power supply. All are easily achievable provided you choose a good LO and pay attention to the build layout and power.

    If you look at the Softrock Ensemble-II receiver schematics on the page I referenced. You will see that the Si570 LO is controlled via USB via an ATMEL ATtiny85 uC, and that the USB is ground isolated. So with the transformer coupled antenna, and assuming an analog (transformer) isolated 12VDC supply, the only ground point is between the SDR audio output and the PC’s sound card line input.

    For a stand alone receiver (DSP following I/Q audio output) you tune the LO with a fixed sampling rate offset from the LO frequency. The offset is either above or below the LO frequency. You fix the offset so you can optimize the amplitude and phase difference between the two I/Q outputs. This assumes you are going through a high quality (typically effective 24-bit resolution, >110dB S/N, 24kHz or greater sampling rate) A/D converter. A fairly decent PC sound card or USB sound card adapter will have these specifications.

    For analog only demodulation, root sum of squared I and Q for AM demod for-example, your follow-on audio chain will require a lot of gain (to compensate for the e.g., sound card gain) and have flat phase and amplitude response across the received bandwidth of interest (or is it twice the received BW?).

    This is often accomplished by an “all-pass” filter using all passives in a ladder configuration plus gain stages or distribted all-pass op-amp stages (“Weaver” demodulation comes to mind, but my memory isn’t so good right now). Either way, you get into very high component tolerances in order to get image rejection (above and below the offset frequency) down to acceptable levels.

    @ Vis1-0n

    She is not decoding (actually demodulating) the (I/Q) signals on her board. She is relying on the PC software application to do this using DSP. She alludes to further work to allow stand-alone demodulation, but only mentions an FPGA. Actually if she wants to go down this path, there are many other examples out there. Today there are some nice inexpensive development boards from the likes of TI or Freescale etc. that have good low noise A/D converters on them along with a DSP empowered uC and/or some programmable logic on-board. The fact that she’s attempting to receive WWV time signals via a dipole indicates the target is to get decent quality MF/HF reception. She will never achieve this with her current approach.

    I do know about the Propeller and the example code you link to. But the likes of the reception via the Propeller you refer to, although very basic I/Q demodulation, would never result in a really usable receiver like the Author of the video is attempting. But it is useful for demodulating strong MF AM signals or maybe even making a simple close-range propeller to propeller data link.

    I have looked at the SDRCube. It is a good attempt, but I feel the quality of the A/D converters is a weak point; at least from the last time I looked at that project, which was some time ago. This can be overcome to some degree using sampling techniques, but I didn’t see it when I looked at the project.

    Finally, almost all the DC-SDR hardware and software designs out there are open-source or similar. For the DSP stuff, there’s the Linrad and GNURadio project, or just look at the source for the multitudes of DC-SDR PC software applications. No need to re-invent the wheel.

    The simplest solution is to duplicate some or all of the work done before (e.g., the Softrocks for one) and move on from there.

  5. vis1-0n says:

    Yeah I know p8x32 decoder is very low end but I am always fascinated by the ultra basic projects using readily available components. Like the super regen – getting something for nearly nothing.
    My whole interest in electronics started from wanting a fm radio as a kid, and from there it went to building a crystal set – failure – & short story is it went to reading up on radio theory, mixers, plls, dds, mcu, the whole software and firmware side of things. And being fascinated by receiving the whole spectrum from 10khz to 100gabizillionhz…lol
    And receiving all modes, am, fm, nbfm, cw, ssb – but like the fm radio when I was a kid, it’s unaffordable. So sdr fascinates me, no exotic components and xtal/ceramic filters.Dspic that can be sampled cheaply from Microchip & codec from TI. I’ ve been sitting with the FST3253 and 74ac74 devices in my parts bin a couple pf years now, and I was even lucky enough to sample a few si570 – 4 couriered in a box big enough for many 1000 with two boxes inside marked as “empty box”!
    I have not built anything as yet but it has been a nice dream and education – all sparked by not being able to afford a $10 radio as a kid.

  6. Drone says:


    Take a look at these offerings for FM and AM/FM receivers on a chip from SiLabs. These are available from SparkFun.

    BusPirate is your friend when developing with devices like this.

    Both of these SiLabs chips are essentially direct-conversion software defined radio (DC/SDR) receivers with on-chip DSP processing to audio and a dedicated uC core.

    Be sure to read the comments and links to the SparkFun forum associated with these products for more info and links to projects data-sheets and app-notes.

    Breakout Board for Si4703 FM Tuner

    sku: BOB-10344 $19.95

    SI4735-C40-GU AM/FM Radio IC

    sku: COM-10227 $9.95

    The Si4703 is FM only it seems, but SparkFun offers a breakout board.

    The Si4735 is more capable; includes MF/SW AM receiption. But no breakout board from SparkFun.

    I am not affiliated with SparkFun.

    Regards, David

  7. vis1-0n says:

    I have got a SI4735 from them, sitting in my parts bin 4 months already… sigh. I really need to stop dreaming and start smelling the solder.

  8. rsdio says:

    Great comments! It’s nice to see so much experience summarized here.

    @ vis1-0n: I’m a radio novice, but my understanding is that you cannot receive “10kHz to 100gabizillionHz” from a single antenna configuration. I thought you pretty much needed a different size and perhaps even shape whenever your carrier changes by an order of magnitude or two. That said, if you attach the right antenna, then you should be able to use the same SDR electronics and programs to receive anything picked up by a particular antenna.

  9. HaIQu says:


    Your comments about Jeri seem to ignore her record of high achievement. I have no doubt that she can make the Spartan6 LX decode radio independent of the PC, and she’s well advanced on that already.

    As for your promotion of SiLabs chips, none are suitable for SDR since the DSP functionality is encapsulated and I/Q are not taken to pins on their chips. In short, they’re mass-market AM/FM chips regardless of how they decode signals internally.

    It’s one thing to have a theoretical appreciation of the subject and pontificate from afar, and quite another to build a working product. Let’s see your design, if you have one. And BTW, I wrote my own SDR software for one of the first batch of SDR-1000’s to leave FlexRadio in 2003, long before the current burst of interest in the subject.


  10. rsdio says:


    I’ve watched a lot of Jeri’s videos, and she often makes it clear that she doesn’t always know everything she’d like to know. Her accomplishments are impressive, but I would exactly call them a “record of high achievement.” Well, maybe they’re “high achievement” for someone who is self-taught and not fully involved in the profession. At any rate, I don’t think it’s very productive to argue that she’s necessarily going to solve this just because she’s completed other projects. I’ve always gotten the impression that Jeri’s goal is to learn and pass on knowledge, not necessarily to complete every project successfully.

  11. in respose to drone’s comment regarding “Jeri Ellsworth” approch to SDR I think drone should lay off and contribute somthing positive instead of criticism.

    In defence of Jerri’s approch I think it’s awsome (in the extreme) that Jerri has decided to make an attempt at building his SDR reciever & share those experiences, her high achievment shoulds be celebrated.

    Drone, before you dismiss & critique Jerri’s work or parts used just remember he has somthing that can be demonstrated as a working proof of concept. (no matter how rough the design is) so please dont ruin this person’s Voyage of discovery.

  12. Ducati-Man says:

    I second the above.
    You go for it, Jeri!
    And don’t be put off by negative comments from former drips under pressure (ex-spurts).
    Cheers, Ducky.

Leave a Reply

Notify me of followup comments via e-mail. You can also subscribe without commenting.

Recent Comments

  • Edward Mallon: A lot of us have ended up at this sensor / screen combination. But I couldn't afford the extravagance of six dedicated control lines on...
  • Andy: Me!!!
  • Bill F: Taking my place in line........
  • Mikee: 11
  • Clara: Thanks, Cinco!