SDR School Part One The Basics

[Music] [Applause] [Music] this is a spectrum recording I made on November 26 2014 it's not just an audio recording it's a recording of about 200 kilohertz of spectrum in the 10 meter band I can play the recording displaying the complete spectrum and waterfall and I can then tune any station within that spectrum or even to multiple frequencies 10 meters was hot that day and Europe is blooming in this is just one example of the many unique capabilities available in software-defined radios by now most hams have heard about SDRs or software-defined radios but it seems that many are hesitant to embrace this new technology that's too bad because it offers some new and exciting possibilities if that doesn't interest you how about better radios for less money so if you're a little overwhelmed with all this new stuff I'm here to help together we're going to explore this totally different way to make radios with micro processors running software instead of things like variable frequency oscillators mixers detectors filters all that stuff we've been using for a hundred years making a better radio for less money is what it's all about and for that reason I think SD ARS will soon replace the superheterodyne while the units have changed STRs are fully compatible with the aging superhead and with existing modes like am/fm sideband and CW and you can use those digital mode programs like ft 8 PSK 31 or slow scan TV or whatever without spending a hundred bucks or so on an interface like signaling STRs are already connected or can be connected to your computer digitally I'm assuming you have at least the basics required for the tech exam and have a reasonable handle on kilohertz megahertz and gigahertz and recognize the various amateur bands a lot has happened since flecks radio introduced us to the SDR 1015 years ago hard to believe it's been that long sdrs have improved dramatically and prices have dropped a lot it's not just a snazzy display on your computer screen it's the future of amateur radio and radio in general for that matter understanding a little about software-defined radios will help us make a more informed buying decision I'll try to provide a little enlightenment without getting too technical so from time to time I'll simply refer to some process as magic if you wish to know more you'll have to look elsewhere and I encourage you to do that these videos are just the beginning I'll be talking about HF radios specifically the receive function covering 160 through 10 meters usually providing general coverage from just a few kilohertz to 30 megahertz many will include 6 meters and some will offer options for 2 meters and higher there are some $20 dongles that offer pretty good performance for the price but I'll be talking about higher performance radios and mostly about the differences and receivers even though some are transceivers I won't be talking about specifications much either today's moderately price as dr's outperform even the best radios of yesteryear I think other differences like features and ease of use are more significant a quick look at Sherwood's receiver test shows us that today's receivers even the cheaper ones which were pretty much sdrs are way better than any radio made just a few years ago great performance just doesn't cost a bundle anymore even the thousand-dollar qrp SDR transceivers like the elad FDM duo or the experts son SDR 2 qrp perform on par with the most expensive radios while sometimes offering feature is not even available in those expensive units if you're more comfortable with the traditional look and feel the Ikon 7300 brought us an SDR with gangbuster performance at a gangbuster price that looks and acts like a traditional super hat to put all this in perspective we need to talk a little about those old fashioned super hat receivers then introduced the basic concepts of SDR s I'll be doing that here in part 1 in part 2 the hardware will look at the wide variety of SDR architectures they are not all the same we'll compare the advantages and disadvantages of the various approaches keeping a keen eye on cost the requirements for front-end pre-selector filters are different for SDRs and we'll talk about why we'll look at some specific designs and the radios that implement them while many see an SDR as a plain box attached to a computer all the computing can be done in the radio could even have knobs and buttons and but pretty much like a traditional radio in part three software puts it all together we'll look specifically at some of the differences that exist in the software from different manufacturers and how they do things differently sometimes in very unique ways granted this could be a little subjective but my objective is to get you to think about what's important to you sdrs offer many features not even available before and you might find some to be really useful the three parts should be viewed in order as each builds on the previous information if you feel you're getting left behind it may be worth some reviewing I first learned of SDR z-- at a presentation by Gerald Youngblood a flex radio at the Dayton ham vention probably in 2004 I was captivated by this completely different approach to building a radio and more specifically the receiver my first SDR was the Flex 3000 I later built a Peaberry kit which me a bit more about SD ours I heard about direct sampling and bought a Hermes board it was a big improvement in 2014 I bought an eel at FDM duo it became a little more aware of differences between the two and then in 2018 I bought an inexpensive receiver the SDR play SDR 1a then added the direct sampling Calibri nano Andy LED s1 receivers just to see how things are progressing I later bought the expert electronics son SDR to qrp transceiver these direct samplers all have excellent performance but they are not alike I'll be talking mostly about the above radios and their software because I'm familiar with them in order to understand a little about software-defined radios we need to go back a hundred years to the early superheterodyne days a local oscillator in the radio was mixed or heterodyned with incoming signals creating an intermediate frequency or if' on a.m. broadcast receivers that frequency was 455 kilohertz and the oscillator operated 455 kilohertz above the desired frequency so if you wanted to receive say 600 kilohertz the local oscillator ran at 600 plus 455 or 1055 kilohertz signals from the antenna and the local oscillator were fed to a mixer stage which created additional signals of both the sum and difference frequencies the problem was we wanted only the difference in this case 1055 minus 450 5 or 600 calories but what happens when we add 1055 to 455 another station at 15 10 kilohertz can also create a 455 kilohertz signal that passes right through the if' filters to solve this problem a.m. broadcast receivers added a tuned circuit that tracked 455 kilohertz below the oscillator and it did a good job rejecting that unwanted signal called an image a to gang variable capacitor tune both the resonant pre-selector and the local oscillator this concept did not work so well though when we tuned to higher frequency tracking the oscillator to the input tuning was difficult and the filters were not sharp enough to fully reject the image this led to double and triple conversion radios and a variety of filter designs over a period of about a hundred years many amateur radio designs of the 50s and 60s had a separately tunable pre-selector which needed to be peaked every time the frequency was changed much later pre-selector band pass filters for each of the amateur bands were automatically switched in when bands were changed eliminating the extra step of tuning the pre-selector unfortunately those mixers did some nasty things as well in addition to creating the desired frequency and its unwanted image additional sums and differences were created from sums and differences and sums and differences and so on undesired spurious signals were created software designed radios SBR for short use digital signal processing or DSP to replace most of the components in a traditional superheterodyne radio for years our radios have used DSP to add a few features like noise reduction and automatic knotch filters to the traditional super hat don't be confused the ultimate SDR is all DSP all those traditional tubes or transistors resistors capacitors inductors and more are not there anymore and something else that not there all those adjustments needed to align the radio no initial alignment after manufacturing and no alignment ever needed to correct for aging of components the basic concept of an SDR receiver is pretty simple convert incoming radio frequency signals to data manipulate the data with very complex arithmetic beyond the understanding of us mere mortals in other words magic then convert the data to sound to feed our speakers or headphones the great thing they can do all this with such perfection that the spurious signals can be practically eliminated let's take a look at the early SDR days of amateur radio when flex radio introduced the SDR 1000 around 2003 understanding what they did back then will help us understand what's happened since then so specifically what did flex do well first they combine some tried-and-true concepts to achieve some pretty remarkable things a process called direct conversion not to be confused with direct sampling had been used in the early CW days a local oscillator was mixed with the incoming RF to directly create an audio tone when the carrier was present notice there is no if' the only image would another nearby CW station producing a different audio tone careful tuning would change the pitch of each signal to allow easier CW copying since this eliminated the IAF there was no need for pre-selector filters to get rid of RF images there are no RF emma jizz flex combine direct conversion with a phasing technique that was used in the 1950s for eliminating the upper and lower side bands when creating a single sideband signal for transmission several amateur radio products using this method were introduced in the 1950s by central electronics the heath kid SB 10 the SSB exciter did that as well let's see how all these concepts might work in a modern SDR receiver as an example we could inject an oscillator into the incoming RF from the antenna maybe with some pre amplification say at fourteen point two megahertz for twenty meters this is all old-fashioned analog no digital yet let's imagine the strong injected oscillator is the carrier of an AM signal then let's imagine that all those signals above and below the so-called carrier are the upper and lower side bands very wide side bands 100 kilohertz or more since the signals above the 14 point 2 megahertz carrier are different from the signals below the 14 point 2 megahertz carrier the upper sideband is different from the lower sideband we then detect this sort of a.m. signal this gives us a very wide audio signal around 100 kilohertz in our example but now we have kind of a mess all the signals above 14 point 2 megahertz are commingled with all the signals below 14 point 2 but we can do something to get them apart again how we'll need to once again do exactly what we just did with one exception will shift the injected 14 point 2 mega by 90 degrees it's exactly the same frequency that shifted by 90 degrees the original signal will call I 4 in phase and will call the second signal Q 4 quadrature as 90 degrees is 1/4 of a complete 360 degree cycle quad like 4 ok now both I and Q are a mess because the upper and lower side bands are commingled the rest easy the magic is waiting we can then combine these two signals and like magics separate those signals above 14.2 megahertz from those below 14 point 2 megahertz and get rid of the carrier we introduced as well we now consume any signal within that spectrum as well as create a spectrum and waterfall display sometimes called a pan adapter notice that the data in I and Q and hence the separated side bands contain everything necessary to receive any signal within that spectrum in addition to creating the spectrum waterfall that highly detailed spectrum shows what's happening now and the waterfall shows what has happened it's that beautiful spectrum waterfall that first attracted me to st ours I said I wouldn't talk much about math but you might encounter a term FFT in SDR discussions FFT stands for fast Fourier transform and that's the math that creates that beautiful spectrum waterfall display the FFT was named for Joseph Fourier a French mathematician who lived 200 years ago the concept seemed so unusual that it was kind of ridiculed for a while there wasn't even electricity back then so it's a little surprising that today the FFT finds purpose not only in the spectrum display but the ubiquitous MPEG and JPEG compression for video sound in pictures and of course much more in the early flex designs most of the process I just discussed was performed in the analog domain the processing necessary to create I and queue signals was analog and then the I and Q which are kind of wideband audio signals remember were fed to the stereo sound card input of your PC the inq signals were then digitized as left and right stereo audio signals and your PC did the rest typical sound card sampling frequencies were 48 kilohertz 96 kilohertz or 192 kilohertz later digitizing took place in the radio but the use of sound card frequencies continued this digitizing process is known as analog to digital or aid to D conversion just to make sure we don't leave anyone behind let's take a quick look at the process of digitizing a signal it means taking a sample of a waveform periodically and creating numbers to represent that sample the more data bits we use to represent each sample and how often we take each sample determines how well the original signal is represented by the data in the early digital days an engineer named Harry Nyquist enlightened us with a bit of wisdom he said that our sample frequency must be at least twice the highest frequency that we wish to digitize in other words the bandwidth of spectrum to be digitized must be less than half the sampling frequency thus the so-called Nyquist frequency is half the sampling frequency flex limited the sampling frequency to a maximum of 192 kilohertz thus digitizing a maximum of 96 kilohertz and I and 96 kilohertz in cue combining I and Q then separated the signals that were in the 96 kilohertz spectrum above our 14 point two megahertz hypothetical carrier from those in the 96 kilohertz spectrum below 14 for two megahertz in the real world get a little less than the Nyquist frequencies so in this case we get a little less than 96 plus 96 or 192 kilohertz of spectrum remember this wide audio contains all the information necessary to create a spectrum and waterfall display as well as to listen to any signals within that spectrum multiple signals may be tuned in at the same time in effect giving us multiple receivers as an example the original flex power SDR software allowed us to tune two frequencies at once but they must both be within the spectrum range while this is the most common approach it's not the only method we'll discuss that more in parts two and three things stayed this way until the Dayton Ham vention in 2012 when flex introduced the new 6000 series which they touted as a game-changer and the improvement in performance was indeed a game-changer they brought direct sampling to the amateur sampling of frequencies of many megahertz instead of a couple hundred kilohertz previously the high-speed A to D converters were way too expensive and not to mention the cost of hyper speed processing needed to handle all that extra data that processing is typically handled in the radio with an FPGA or field programmable gate array high-speed A to D converter and FPGA prices started dropping around 2012 and the direct sampling SDR became affordable at least to the ham with $4,000 or so sitting around about the same time another group HP SDR high performance software-defined radio introduced a completely open-source direct sampling transceiver using the open source software power SDR already being used by flex radio in the earlier designs the HP SDR designs evolved into the Hermes board which is later manufactured by Apache labs in India launching their NN series of SDR s thankfully prices have continued to drop and direct sampling is the norm for all but the least expensive receivers if you're spending over $200 or so for a receiver today it should be direct sampling so what's so different about the direct sampling receiver in one sense not as much as you may think we end up doing about the same thing as the direct conversion receiver except we digitize the RF coming right off the antenna the down conversion is not accomplished with a variable frequency oscillator like the 14 point 2 megahertz in my previous example it now is all mathematics for that reason it's sometimes called digital down conversion or DDC in a transceiver the transmitted signal may use digital up conversion or D you see and the whole process labeled D DC do you see using digital down conversion fewer spurious signals are created and more importantly we can now create I and Q signals digitally remember before we directly converted the RF to I and Q in the analog domain and then use the computer sound card to digitize I and Q we then processed I and Q digitally in the computer but we created them as analog signals in order to completely separate the frequencies above our 14 point 2 megahertz example from those below that frequency I and Q must be exactly the same amplitude and exactly 90 degrees apart the analog creation of I and Q was unstable and left us with errors that could not be fully corrected those errors created a digital image from the unwanted sideband the math gives us perfection that does not drift or change with age I said earlier that this I and Q technique eliminated the need for a friend filter to reject those unwanted images created in the Superette that's true but now we have another problem remember the Nyquist frequency I mentioned earlier 1/2 the sampling frequency it turns out that if we allow any frequencies above the Nyquist frequency into our analog to digital converter it mistakes them for the lower frequencies ok any of that this is known as aliasing a certain frequency is alias to a lower frequency we must eliminate those signals above the Nyquist frequency with a low-pass filter it has to be an analog filter before the analog to digital conversion that filter is called not surprisingly an anti-aliasing filter now what I have just said is true but it's not the whole story it sounds crazy but we can actually digitize those signals above the Nyquist frequency in fact we can digitize signals above the sampling frequency so Harry Nyquist was not completely correct look at it this way imagine some fan fold paper like we used in the old dot-matrix printers then we'll call this first page Nyquist own 1 from 0 to 1/2 the sampling frequency or a Nyquist frequency the only Zone relevant for mr. Nyquist wisdom zone 2 goes from 1/2 the sampling frequency to the sampling frequency zone 3 goes from the sampling frequency to 1 and a half times the sampling frequency zone 4 from 1 and a half to two times the sampling frequency and this pattern continues in multiples of the Nyquist frequency with no filtering before the a to D converter all the signals in all these Nyquist zones will be digitized and will be alias to frequencies below the Nyquist frequency they can no longer be separated from each other or the frequencies we want so what can we do simply put a bandpass filter for each zone we want in front of the A to D converter of course we only need a low-pass filter for zone 1 all these are called anti-aliasing filters this can lead to some confusing terminal generally if nothing is mentioned you may assume that only signals below the Nyquist frequency that is zone 1 are digitized this might also be called sub Nyquist all the higher zones might be referred to as super Nyquist or simply under sampled just to add to the confusion another similar term over sampling has nothing to do with this previous discussion over sampling refers to the idea of sampling way beyond our needs this might seem to be a waste but the madness has a purpose the data representing the higher unneeded frequency spectrum can be exchanged for a higher bit depth of the spectrum we are using another example of math magic this process is called decimation I think a poorly chosen word which brings to mind something completely different hopefully by now you have a pretty good idea of how things can be done along with adding a few words to your vocabulary we learned about front-end filters I and Q a 2d conversion and DSP the Nyquist frequency under sampling over sampling direct sampling aliasing decimation FPGAs and a bit more understanding these concepts and the related terminology will help us understand the different SDR architectures we'll talk about that in part 2 the hardware

Channel: David Kennett

Views: 143,439

Keywords: Software Defined Radio

Id: ncxyycmSeWU

Length: 27min 53sec (1673 seconds)

Published: Mon Dec 31 2018