Sunday, July 26, 2020

A Ham Radio Series 9 – Software Defined Radio

Software Defined Radios (SDR) are kind of the “new hotness” in radio to hobbyists although they've been used commercially for a long time.  They almost have an air of magic around them.  One of the things that a few people wanted to read about was a primer on SDR.

At times like this, I like to start out with a definition of what we’re talking about just to make sure everyone is on the same page.  Several groups have suggested definitions, but I like the definition used by one of the biggest companies in the ham radio SDR world, Flex Radio Systems, who says an SDR is one:
where components that have typically been implemented in hardware (e.g. mixers, filters, amplifiers, oscillators, modulators/demodulators, detectors. etc.) are instead implemented using software running on a personal computer or other embedded computing devices.”  (Note: that last phrase could mean internal to the radio and not your computer) 
Those of you with some familiarity with SDRs might think of something like the RTL-SDRs, the seemingly magic little receivers that tune from around 24 MHz up to 1.8 Gigahertz and use software on your computer to do the many functions, “typically … implemented in hardware … modulators/demodulators, detectors...

Calling those little RTL-SDRs software defined is a distraction, though.  What makes those little SDRs capable of tuning 24 to 1800 MHz in something the size of a USB drive isn’t the software, it’s the hardware.  Dedicated hobbyist-software hackers (in the good sense) took a chip designed for TV reception and realized that they could control it and get outputs from it that they could use for whatever they thought of.  There are plenty of software packages to try to see what you like, things like SDR # (SDR Sharp) or HDSDR.  That previous link is your source for RTL-SDR information.

In the professional world I retired from, we had radios that met the basic SDR definitions 30 years ago.  These generally ran the Analog/Digital Converter (ADC) with a much lower clock frequency than the IF it was converting, called undersampling.  The first ones I worked on were in the early ‘90s, and that advanced with the hardware capabilities until the last radios I worked on (starting in about 2008) were close to having the ADC attached to the antenna.  These are called band samplers; they convert the entire desired band RF spectrum to digital and process the bits.  There's simply not much analog hardware at all.

The main radio in my ham shack has met the Flex Radio definition of an SDR since the early part of the ‘00s; I honestly don’t remember the year exactly but it was an Icom 746Pro.  In analog radios it was common to need to buy additional IF filters and you ordinarily only had a choice of only two filters for both voice (SSB) and Morse code (CW).  With the IC-746Pro, with filtering and demodulation done in software, I could have my choice of three filters for every mode and I could change the filter bandwidth while I was operating if I felt the need.

This is typical of the choice to buy an SDR over an “all analog” radio.  It may cost more but the cost buys you tons of convenience features; a choice of 3 IF widths for every mode (3SSB, 3CW, 3AM, 3FM and any other mode on the front panel) versus a couple of filters bought after the radio.  I believe the number talked about is at any given moment you have the equivalent of 15 IF filters in the radio and by changing them on the fly you have 41 IF filters available.  Plus, any other audio filtering or audio processing that can be done in Digital Signal Processing (DSP) software comes along with it.  Correlation cancelers (eliminate the annoying sound of guys tuning up on frequency; some systems remove more than one station tuning at a time), correlation enhancers (noise reduction), adjustable dual peak audio filters (for RTTY for example) and more operating conveniences.

It’s not that 15 filters couldn’t be done in hardware, but it’s terribly impractical; 15 filters would massively increase the size of the radio and add at least 14 more of the most expensive parts in the radio.  Additionally, the software allows the manufacturer to load the same software on successive models and have them behave the same way. 

To paraphrase something I said back in the Radio Sunday series,
The most important thing to point out is that an SDR isn't doing things that an analog radio can't do, but the SDR does them in more repeatable ways, and in ways that can be miniaturized more easily.
Today, the emphasis is on band sampling receivers, which buys you a tremendous advantage over most other radios: better linearity.  In a linear system, the only signals out of an amplifier stage are the ones you put into it.  A nonlinear stage adds signals because of mixing, a multiplication process.  The only intentionally nonlinear stages in an analog receiver are the mixers.  The incoming RF is multiplied by the Local Oscillator (LO) giving rise to more signals on the output than the input.

In a band sampling receiver more linearity is relatively easy to obtain.  Run more current in the RF amplifiers to increase their intercept points. In an analog receiver, there’s always a mixer (sometimes more than one) and those are intentionally nonlinear.  Making a more linear mixer is a more difficult design; there are off the shelf solutions for different levels of linearity, but since the intent of the part is to be nonlinear, it’s quite a balancing act.

A band sampling HF receiver looks like this.  It uses some filters to remove possible sources of signals that would combine into interference products,  has separate gain reduction circuits (Automatic Gain Control or AGC) for on channel and off channel signals, then samples them directly in 16 bit, high linearity analog to digital converter.

Naturally, you can't have a software defined radio without the software.  Flex Radio made software their emphasis, thinking that at some point within a few years anybody can do the hardware.  None of their radios had a front panel - until the advent of the Icom IC-7300 which took the ham radio world by storm.  There was a shocked reaction from the industry almost saying, "you mean people want an SDR with a front panel?"  Flex introduced their Maestro series of tablet computers that mount to the front of the radio and implement some knobs that can be assigned a function in software.  The major ham radio brands all have radios on the market which are SDRs with their software built in behind a front panel with tuning knobs, so that in some ways they behave much like radios have for decades but are already enhanced and can be enhanced more so with external software.  Icom, Yaesu, Kenwood.  I believe only the Icom is a band sampler.


  1. A very nice introduction to SDR. Some of the military SDR radios I was involved with from the '90s employed software to enhance bandwidth choices, signal processing and modulation types at a lower cost point. Interestingly in amateur circles the early radios with SDR were sold at higher prices than analog only models with similar features. Only now does the price "premium" seem to be gone.

    Analog only models have been around 100 years or more. Their weaknesses and flaws have been thoroughly explored, and under pressure from ARRL and experts like Sherwood Engineering significant improvements in performance have been achieved. We have yet to see similar pressure applied to SDR radios. For example, testing analog radios for rejection of image and IF frequencies was common. But testing of SDR radios at alias frequencies is not. And just as amplifiers are not perfectly linear devices, A/D and D/A stages are not either. I recall specs of early A/D converters such as "16 bits - 14 bits effective." The implications of these flaws in digital parts has not been wrung out, yet. When we finally do so, we will discover that SDR are not the magic they seem to be - while still being very useful.

    1. I think it's not revealing any secrets to say Rob Sherwood has that Icom SDR, the IC-7610, and talks about it like it's the radio he uses most.

      Aliases are less likely to be a problem in a band sampler than in an undersampling radio, like the ones you used in the 90s, because they're farther away in frequency so they can be filtered easier. I think it helps to put the ADC in a shielded box because they'll "hear" aliases out to several GHz, even with a relatively low clock speed. In a filter like the one in the block diagram, passing 35 MHz while attenuating 50MHz and above by 100 dB, it's a rare filter that can keep that rejection all the way "from DC to daylight". There's virtually always some parasitic response well above the cutoff.

      Still, the answer is filtering before things hit the ADC and the high end SDRs may have more and better filters than the undersampling radios did. They're not perfect though. Nothing is.

      The whole point of the post, and everything I've written about SDRs over the years, is that they're not magical and there's nothing they do that couldn't be done in analog hardware. They reduce inventory costs for the manufacturers, while allowing more features and operator convenience. That combination will keep them growing as a portion of the market.

  2. Just a receiver, not a transmitter, but this is what can be done these days.
    Sweep speed is 200GHz/second.

    Need better resolution and analysis?
    Granted, its sweep speed is only 24GHz/second. But with a 12.2KHz resolution.

  3. Don’t think FlexRadio had used the definition offered, as it is the first paragraph of Wikipedia -

    While I cannot speak for FlexRadio, they seem to focus on an SDR as a radio design that “ allows the personality, functionality and performance of the radio to be upgraded through simple software downloads.”

    As a participant in their external test team my NDA was in place two years prior to the ICOM 7300 announcement, so no the Maestro was not a response but was a planned evolution.

    The trickle down from commerical and military technology to amateur radio, including chip availability at an amateur price point, brought SDR to the forefront of ham radio development.

    Your right that ham radios had started defining portions of their functionality years back, I am thinking how you could change out the eprom chips to upgrade models for years, but were those really SDR?

    Led by TAPR the SDR redefined transceiver construction.



    1. Don’t think FlexRadio had used the definition offered, as it is the first paragraph of Wikipedia -

      I lifted the quote from a presentation I did at the local hamfest on SDRs back in 2011, because it has been my "go to" quote in SDRs since then. Unfortunately, I didn't do a bibliography so that I could find the source again but it was from the days of the Flex 3000 and quad sampling detectors (Tayloe detectors).

      The quote that the IC-7300 forced Flex in the direction of the Maestro system is strictly my own, based on some conversations with people who watch the industry closer than I do. I didn't write that well in that last paragraph (what I think was the last edit). I do recall seeing the Maestros in Dayton back in '16, right around the time that the 7300 was released in the States and they certainly couldn't have thought up a product and introduced it from scratch between the end of '15 and then. What the people I spoke with were saying is that Flex started emphasizing and selling the Maestro more.

    2. Whether ICOM or another, once the technology used to make a "radio-server" became affordable at the amateur equipment price price point, the neat evolution of hybrid technology was going to happen.

      The knobs/no-knobs issue predates the Maestro, as with the prior "Thick-Pipe" SDR technology (where the PC is part of the processing, rather than just a platform for the end user GUI) Flex had launch the FlexControl. I bought one in 2011 for my Flex-5000A Some on the Thick-Pipe vs Thin-Pipe differences:

      One area that has fascinated me is how the FPGA (Field Programmable Gate Array) went from exotica to an SDR norm. The technology was there, but the economics were against widespread amateur usage. When FlexRadio announced they would be using FPGAs in a bigger way I wrote:

      If an SDR is just a radio server, how you controlled it was an ongoing discussion. The current FRS ecosystem allows for a end-client GUI on normal computer hardware (Windows, MacOS, and iOS on both iPhones & iPads), the Maestro for the face-plate experience, and the M-models for those who want everything in a traditional transceiver packaging. In addition to the FlexControl other hardware has been made work to give more options.

      In my mind the real magic is the ability to remote, the ability to have essentially a "new radio" with a new Version of SmartSDR loaded into the radio, and in my case the amp/transceiver integration when viewed from the end-user GUI.

      While I am trying this comment I've made a dozen or more QSOs running digital mode to my home station via remote. I've my choice of a radio in the next building - a 6600M - or a radio at home - a 6700 - when I remote, more to do with antenna choices than much different in capabilities. I regularly had remoted while traveling, using an iPad and a noise-cancelling boomset.

      Since I bought my Flex-6700 I believe I have upgraded that radio over 50 times, which is more than the usual user experience being on the test team, but as a contrast few of my other radios have received more than a single feature upgrade while I've owned them. (I started with pre-release version 0.6ish to the latest 3.1.12 General Release - unfortunately I can't tell you anything about test versions though).

      While I typically run barefoot, I do have the FlexRadio/4o3a PGXL Amp, which base functions and metering are integrated in most versions of SmartSDR.

      Is it all perfect, no. But that is not a problem as feature-by-feature my system will evolve as the software upgrades drive feature improvements.

      Just as a counterpoint, I do keep a nice Collins setup for a traditional ham experience when I don't want to run all the gee-wiz stuff.