The SDR-Cube is a brand new Software Defined Radio which is able to process and (de)modulated RF signals without a PC, but with all the benefits of an SDR. The SDR-Cube was introduced to the public in September during the TAPR Digital Communications Conference (DCC). After going through the presentation I was immediately thrilled by this tiny little cube. From my perspective this device is so important that it desired greater explanation. I’m very happy that the founder George, N2APB volunteered to record an interview with me. The interview lasts about 75 minutes and is full of valuable information, mainly but not only about the SDR-cube. Don’t forget to have a look into the **shownotes **below!
Listen or Download this interview
Picture of the SDR Cube
This is the schematic of the SDR-Cube which George explains in the interview
The mentioned Links
- SDR-Cube Website – SDR-Cube Presentation form TAPR/ARRL Digital Communications Conference (DCC) 2010 (.pdf) – George Heron, N2APB – Juha Niinikoski, OH2NLT – Microchip dsPic33 Development on a USB Stick – Microchip MPLab free Integrated Development Environment (IDE) All pictures in this post are copyrighted by George, N2APB and used with his permission.
DH1TW: Hello everyone I am Tobias, Delta Hotel One Tango Whiskey, and I’m your host tonight. Most of you might know me from my blog at www.dh1tw.de where I’m writing on SDR and contesting related topics.
Today I’m very happy to have George Heron, November Two Alpha Papa Bravo with me on the line. George recently caused some buzz in the blogosphere with his presentation of the SDRQ, a stand alone SDR transceiver which he developed together with Juha Niinikoski, Oscar Hotel Two November Lima Tango.
So, George, thanks for joining me tonight.
N2APB: Hi Tobias. Thanks an awful lot for having me. This is a fun project and I’m really pleased to be able to talk about it with your audience. DH1TW: Awesome. So I think we should start with a short introduction of yourself.
N2APB: OK, well good. So again, I’m N2APB, Alpha Papa Bravo.
I’ve been a ham for, oh gosh, I’ve not counted it up recently, oh, since 1968, I believe when I was in High School. Always been interested in electronics and I went to college, or university, and got my degree in electrical engineering, so it’s always been a love relationship with hardware, with ham radios, and building my own equipment and over the years that’s really lead me to stay strongly into technology in the homebrewing area.
So, I love building my equipment and experimenting with new and different kinds of technology and as time went forward and technology went forward, I really enjoyed the microcontrollers and microprocessors and the things you could do with small electronics and radios and the combinations of the two, and building, my love of homebrew.
Putting my own equipment together always stayed with me and throughout the years I’ve been associated with various local and national ham radio clubs that have people that are also really interested in homebrewing, for example the New Jersey QRP Club, is a local regional club here in the US on the Eastern Seaboard. I was doing that with a partner for maybe the last ten, fifteen years, and then we formed the American QRP club, more of a national type of club. Again associated with homebrewing, we would design small projects, put them out there as kits for guys to build.
I published a magazine called QRP Homebrewer, and then it turned into Homebrewing Magazine for the last seven years. So just getting all of the different ways you can use homebrewing tools, common techniques you can use to produce QRP types of gear, measurement equipment, tools and instruments all built by yourself to get on the air. That’s been the main interest that I’ve had over the years.
DH1TW: Oh, that’s cool. So, how did you get into the SDR business? N2APB: Well, SDR, kind of like the latest craze, at least from a technology standpoint, and the SoftRock is, we’ll talk more about this going forward, but SoftRock was kind of introduced by Tony Parks, KB9YIQ, some five, six years ago. It was a way to easily put together a relatively high quality station with a very minimal set of parts and he demonstrated, for like $29, you could put together some circuits to produce some raw signals, some quadrature signals that with some computing power you could actually decode and encode any modulation scheme that you would want, whether it’s AM, sideband, CW and digital. So this flexibility of modern technology called SDR is really what caught my attention back then and then I coupled that with kind of a goal for a motivation that I’ve always had, is taking my equipment out into the field. I love taking my radios out up on a mountain top, out on hikes, in the backyard, operating field day and you can’t easily take your Yeasu, Kenwood, you know your big old standard transceivers out into the field. Power supply, portability is just not there so the combination of QRP and homebrewing really comes together well with SDR, in my opinion, because you can take it outside. DH1TW: Ok, so now I think I know where you want to go, since I also build up already a couple of SoftRock or SoftRock clones, we have some more popular kits here in Germany. It’s always been a hassle wearing around the computer and I suppose this was one of the motivation reasons why you started with the SDR Cube, wasn’t it?
N2APB: Oh, absolutely, you zeroed right in on the problem. I love the PC, I do software for a living, I do cyber security as my day job, so I’ve been in security and software for a long time so I know the power of the PC and indeed, a laptop and even today’s Netbooks are very popular and very low cost. You can take them anywhere but you still can’t easily, as easily, take them out into the field and look at the screen in bright daylight, it does eat up a lot of power, and when you’re out in the field and portable you want to be sure you don’t use up your batteries too much.
I don’t know about you, but taking even a $300 Netbook, it’s cheap by a PC perspective, but it’s still expensive from a radio perspective. So, I don’t want to take my laptop and throw it in my backpack and treat it rough and have the morning dew settle on it and so I’ve always been in search of finding a way to do these kinds of things. You have the power of the PC but without having to take the PC along, and my first step was actually when it came some three years ago with my design partner, W8NUE, we designed the new PSK digital modem.
Typically when Hams today are using the digital modes they again connect the PC in order to do the digital decoding of the tones coming across the air, so you take your HF rig, you plug it into the sound card and your PC and beautiful! The screen displays that are able to take care of decoding all of the digital modes. Again, I wanted to take my digital mode operation to the field and I didn’t want to take my PC, so we developed, then, this was again three years ago, we developed a DSPIC controller, a small microcontroller that goes in a box. I have it here in my hand and what it does with the display on it, it is actually able to connect to a transceiver like the FT817, and actually display the digital mode. So taking this the next step has been the goal for about the last year.
A solid year has been going into my thoughts as far as what can we do along the same lines as putting the computing power in a small box. Instead of using the PC, we have it in a small box that we can take to the field. So, a number of Hams have been doing experimentation in the recent years along this line. It’s curious that nothing has come out until now, and all of the sudden now there are some self contained SDR transceivers. I think my Cube is one of the first to be as self contained and all the controls and some extra features that we’ll get into.
I found a design partner, another design partner for this project, the SDR Cube, in a fellow Ham who had similar interests in wanting to, first of all push technology, try to find a solution, who has desire for taking ham radio to the field and as you mentioned during the introduction Juha Niinikoski, OH2NLT, and I met up about a year ago, or a little bit more than a year ago, met up because he had been working on a project that he called Cheap DSP. He had originally taken a lesser capable processor, called the DSPIC30, and he built some experimental circuits around it in order to decode I and Q. DH1TW: So, just for clarification, DSPIC is the digital signal processor from the company Microchip, but besides those DSPIC’s and the Microchip products you could also use the DSP from Atmel or from Texas Instruments right?
N2APB: Oh yes, there are some. There’s Atmel, TI makes a version of the 56000 DSP. We did actually a lot of experimentation leading up to this point.
It’s quite amazing the path that we have taken or that we and actually some other developers. The Austin QRP Group, from Austin Texas here in the States, they call themselves AQRP. That group has been doing some really good experimentation along the way too, and I’ve been working with them on some of the projects. We’ve been experimenting with that TI DSP, we’ve experimented with, oh gosh, I’m not going to remember, but there’s another one that comes along on a small stick, a USB stick, a processor, a DSP processor on a stick called EZDSP.
But in each case we kind of came to a roadblock because either it could not be easily homebrewed or that it didn’t, like the demonstration boards, didn’t have the necessary IO that we needed to run a transceiver. So kind of simultaneously that group and Juha and myself and, oh gosh, there was another fellow who recently published in QRP Quarterly Magazine, a DSPIC solution. Juha and I coupled up and we took his Cheap DSP approach and really amplified it and built onto it and strengthened it, added many, many more features. We went to a higher capability DSPIC, DSPIC33F, with lots of IO and that’s given us what we have today, and it really quite a unique solution we think, because it brings together the flexibility that SDR offers, we got a fully software defined radio with all of the software and capabilities that SDR brings to the table. That is, the ability that one has to demodulate and modulate any mode that you wish. If you have I&Q, you can conquer the world. That’s kind of like a little manta that we have, because with those two quadrature signals, the mathematics is such that you can compute in order to demodulate any signal that’s coming across. That gives the radio, like any SDR, tremendous flexibility, because when a new mode is invented, for example a digital mode, there are often new modes coming along, you can just download new software to your processor and crunch those I and Q signals in a different way in order to decode the signals properly.
Now of course with the smaller processor in the SDR Cube, it’s not possible to have all of the capabilities of some of the really good SDR radios such as Flexradio, has a number of good products out and other SDR solutions that plug into the computer of course have the entire computing power of a laptop at its disposal, a Pentium. We don’t have a Pentium processor in our SDR Cube, but we have enough, and enough to really do a good job for what it was designed to do and that is to take it to the field.
DH1TW: Yea, I’m coming a little bit more from the contesting scene and there we have similar problems. It’s not the problem that we have to take it into the field, in our case it can be more like the big box on the table or below the table, but one of the crucial requirements in contesting is reliability and if you’re using a computer you always have the certain multitasking unreliability and this caused a lot already to me and other fellow contesters, it already caused a lot of frustration and it just starts if you loose the focus of your SDR software and you press the button and it doesn’t transmit anymore or if the computer crashes, you’re just off the air and you’re loosing the frequency, you’re loosing QSO’s and therefore SDR, the technology has to involve a little bit, and I’m very happy to see now finally the real first stand alone transceiver which I suppose will come for a very affordable price.
We have seen now one or two, I think the ADT-200 from Switzerland, which has put more of the higher priced segment, and I’m very happy to see now that there is coming activity into the direct demodulation and independence of the computer. I wouldn’t say that it is unnecessary, I suppose that at a later stage we will have both worlds. We will have the important, the time critical part in a microcontroller, but then having the advantages for example, of CW Skimmer, which is just decoding the whole band of the CW signals at the same time in the computer, but which is not such a time critical task if the call sign on 7042 appears a second later or a second earlier, no problem. But if you’re running QSO’s you want to be sure that you have real-time or almost real-time.
So, coming back to the SDR and its development, since we are having this conversation with webcams on both sides, I can see your great big shack with all these measurement devices such as a network analyzer and over there I can see a spectrum analyzer. If one of my listeners is interested in getting started with the development an SDR device, are these measurement devices a prerequisite? What do they actually need on a hardware, but also on a software side to get started with SDR development?
N2APB: That’s a great question Tobias. You don’t need a full blown RF shack or an RF lab bench in order to do some of the things that we are talking about here. As you’ll learn as we talk a little bit more about the design and architecture of the SDR Cube, most of the technology that Juha and I developed has been in the digital world.
We depend on a SoftRock for the RF front end, so right there that limits us to needing to design and develop on the digital side mostly at first, and then this sort of, as you might think to because radio today really involves a lot of digitizing the signal, computing values and then converting back into analog space in order to hear the signals that we are receiving. I think the bare minimum that one would need, at least taking the path that we have taken is a software development environment that mates up with the DSP processor that we are using. As we mentioned, what we are using is the microchip DSPIC33.
Now it just so happens that Microchip makes a very, very powerful, and free, development environment that can run on a PC and it connects to the target board, the Cube, by means of a cable, such that we can program the new software releases into the DSPIC and actually step through it in a debug fashion in order to make all the software work. So, right there, there’s a free development environment which produces, which gives you an editor, it gives you the compiler, the program is written mostly in C and therefore it’s a very easy, it’s a very comfortable language and very easy to read and most programmers today know how to program in C and our software is what we call open so we are making it available to anybody who wants to take the software and experiment with it and they could take this and put it on their own PC in order to change things to be more suitable for them download it to their Cube and they would have their own capabilities right then and there.
There’s something like a small pod, I can show you here on the pod/video cast, your listeners can’t see it of course. This is a small what they call a Hockey Puck. Microchip makes a small programming pod that interfaces your PC to the target board, it provides signal translation and timing. That is like, less than $100, and there’s even cheaper ways to do that same thing.
So it really doesn’t cost too much from the development stance, that capability. Of course an Oscilloscope is always helpful on the bench and when you’re debugging either digital signals or RF signals an Oscilloscope, a DVM, a digital volt meter, a multi-meter is very essential, and a nice strong power supply is good. Beyond that you don’t need other types of equipment, too much, unless you start to get into some of the more specialized areas of your design.
If you want to monitor your transition quality, the signal levels at the RF, then you need some other things, but as I indicated, and we will talk more about it, is that with the SoftRock as the RF front end, we are depending on the designers of the SoftRock to have done all of that. We are just taking that module, putting it inside the box and connecting our electronics to it. They are significant electronics in order to do what we are doing, but by depending on the SoftRock for all of the RF, it takes an awful lot of that load off of our shoulders and enables us to produce a box that can be used with any of the some 11,000 SoftRocks in the field today. 11,000 SoftRocks! So we designed the Cube to work with any of them.
So if you have a SoftRock, you mentioned that you have a couple of SoftRocks, there is an option that we are planning in our kit, that you can buy the kit without the SoftRock and you can use it with your SoftRock. So all you do is connect to a connector on the back of the Cube, connect the signals down to your SoftRock, I imagine it is in some kind of chassis or some kind of a small enclosure, or maybe not, but still that is all you need in order to have your SDR transceiver.
DH1TW: Oh, that’s great. So if I’m a novice and I want to get started in SDR, well what I need from Tony, a SoftRock head, and on my side, the MP lab, which is the integrated editor and compiler which comes free from Microchip, right?
DH1TW: And I would need the end circuit debugger, the hockey puck, in order to get the software which I have written into the microcontroller itself. How much is the microcontroller? Just a few bucks I suppose, right?
N2APB: Yea, just a few bucks, maybe $3.50.
DH1TW: Great. A power supply and then this might be the basic setup in order to get started, right?
N2APB: Oh, it’s even easier than that. You can buy, there is just recently that came out, there’s a small stick, it’s about one inch by three inches and it has a USB plug on one end so you can plug it into your computer for power and I/Q processing, but that stick contains the DSPIC, it contains the processor and it’s available for maybe $45, I think. It’s already built and you can immediately start debugging and using the software and maybe modifying it to have a smaller scale type of SDR Cube available on that small stick.
DH1TW: Do you know the name?
N2APB: I think and I don’t know, you can put it on a link on your page perhaps, I’ve got one on order myself just for playing around with.
DH1TW: Ok, so now we know which kind of microcontroller to use, and how to load the software into the microcontroller. But what about writing the software? I mean, digital signal processing is inevitably connected with complex mathematics. I remember that back at university, I learned all of the equations and I remembered that I knew how to solve them on a piece of paper. Later in university, I used Mathlab in order to generate the code for me. Are you also using any kind of simulation software or did you have DSP libraries in C available which you just needed to call with the right set of parameters?
N2APB: Ok, this is a great question. And I think in general what I will state is that any software developer these days, or ever, has always taken and borrowed modules from somebody else, either to start with or as functional building blocks.
For example, let’s do it this way: Juha had developed or is the main software development arm for this project and we built upon his Cheap DSP, and I’ll give you the link for that for your website. As a starting point he had taken application notes from Microchip and other public domain sources that actually preformed an FIR filter, which is a filtering mechanism implemented in DSP, and IIR filters and various other types of functions even for non-DSP functions for control of an encoder, for example. So we take these libraries, these different subroutines, if you will, and we collect them from all different sources into our target, our brand new SDR Cube project.
At the start there was no functionality at all in our Cube, but we accumulated these various different functions, they didn’t all operate exactly as we wanted them or as we have them operating now but it was a starting point. So you take from others, other public domain sources, application notes, other projects that you or your friends have done and you bring them together in MPLab. Then you start weaving, tying them together. In software you’re able to compile them all together, get your variables such that it compiles without error and then you’re actually able to turn a dial on your rotary encoder and see something change in your system. Maybe it’s a display.
For example, we took as a starting point the display driver from the new PSK modem, which again is shareware software and what we did in order to use the nice graphic display that we were able to put on the new PSK digital modem, we were able to use that same graphic display as a starting point on the SDR Cube. Then, once it was all working in a basic building block sense, we then stared to add to it and change its behavior in order to make it do more of what we wanted it to do.
For example, on the SDR Cube we have a very unique function and a very unique feature that is in our opinion that is very attractive and powerful and that is that it displays live spectrum on the display. So as you’re tuning along you’re able to see the, it’s not a waterfall but it’s a live spectrum of peaks indicating different signals of QSO’s and progress. DH1TW: Is that similar to, sorry for interrupting you, would that be similar to the Icom Radios which have the big screen?
N2APB: Of course, exactly, that’s exactly right in fact it’s similar to the SDR versions that are run on a PC. Not quite as many frills and options, we don’t have a waterfall, yet. We don’t have a, what do they call it, there is a tuning indicator, a vectorscope. We don’t have a vectorscope indication, yet. But these are all things we can put on top of the basic FFT processing from the block that we put in there that started as the application notes from Microchip. So with these building blocks in place we can put in these specialized features like I’m describing here and it is a beautiful blue screen, a blue graphic screen, tune in across the dial and see the live spectrum. In our case we have an 8 kHz swath of RF being shown, of RF spectrum being shown at the top of the display so we all of the different signals right on the screen and dial over to where you want, in the same way as you do on your Icom and Kenwood types of expensive transceivers.
DH1TW: Ah, that’s cool. So now I have a picture in front of me of the SDR Cube and I can see a frequency knob, where I suppose in the middle I change the frequency, can tune the frequency as I want to have it. What are the buttons to the left and to the right?
N2APB: Oh good. Let me give you a tour of the front panel. Looking at the Cube, right in the center bottom is the frequency dial, as you indicated. That’s the main tuning dial. All of our buttons, well the controls have two functions, either to press it short or to press and hold it a longer period of time in order to get it’s second function.
Earlier in our discussion you hit upon a very important point for me and one of my major goals for this transceiver, the SDR Cube, was to have it be very transceiver-like. Frankly, and his might sound strange, I do not like operating a radio with a mouse, I do not like operating on a screen because if you hit something it really goes haywire and you don’t know why, just because of some accidental push. I wanted a front panel that looked like a radio, and I like knobs, so this is kind of like how it developed here.
So, in the center upper part we’ve got a character display, I’m sorry it’s a graphic display, 128 pixels wide by 64 pixels down and that allows us to put in text, we can put in graphics, we can put lines, and again, the spectrum. So in the center, as you pointed out is the frequency dial, to the right of it actually we’ve got a row of push buttons and on the bottom we’ve got a row of potentiometers. So, again, in typical fashion it’s nice to have a potentiometer for volume. In the lower left hand corner is the AF Gainer volume.
Next to it is the Keyer potentiometer, it’s nice, in my opinion, even though many radios, minimalistic radios put many layers of menus to get to certain things and you can adjust settings like keyer speed with push buttons, I don’t like that, so what we did was put a potentiometer there to adjust the keyer speed, and it’s a full iambic keyer, either mode A or mode B or straight key and we can adjust the keyer speed. On the right half of the display, at the bottom are two potentiometers. We have one for the filter, we have an audio filter that takes and produces a very nice brick wall filter at the standard audio frequencies that you might expect to see in a transceiver such as 2.3 kHz or SSB, and then we produce a number of other filters that are suitable for CW, 500 Hz and 300 Hz, so this allows us to really zero in on a signal and perhaps we can get an audio test later on and that might come through your podcast.
To the right of that is the last potentiometer and that is called RF attenuator. We produce the manual control for attenuating the front end RF signal to prevent overloading and more easily keep the signals within range of the DSP. Little bit of technical detail, but perhaps if we have time we can get into is that we do not yet have ALC, or AGC actually, built into the system, it’s a difficult thing to do in a small computerized system like this. There some five or six gain points throughout the entire signal chain and it’s important to make certain that none of those exceeds the limits and saturates or overloads the system. So we provided an ability to control the RF attenuation with a pod.
The next row up if your looking at the picture would be the white buttons, push buttons. From the right is a menu, every radio has to have a menu and this is our menu that allows us to select a lot of things that are not always adjusted but just occasionally. Beneath that if you press an hold it long, it says “lock”, so we are able to lock the frequency display such that if you accidentally bumped it you would not change frequency.
The push button next to it is the Rate and in order to easily move around the band we provide three tuning rates, we can tune at 100 Hz rates, or we can tune at 1 kHz rates or we can tune 100 kHz rates if we really want to go from the CW part of the band up to the voice part of the band for example. Tuning side band signals is convenient with a 1 kHz tuning rate so we just tap it once to get it and then on the screen is an indication that we are in the 1kHz or what we call the medium tuning rate. If we are tying to fine tune, get that voice side band just sounding right or get the CW tone into the pass band of a narrow filer we can tap it again and it would then be tuning at the slow tuning rate. We can very slowly tune it and it becomes very easy to tune the radio once you have done it a couple of times.
On the left hand side the push buttons are Mode, so we CW mode to sideband and underneath that is persistent or if you press and hold that button, we can actually save all of the system settings so if you set some of the settings here that you like, the Mode, the CW speed, the tuning rate, the frequency, you can press an hold that save button and the next time when you turn the transceiver on it comes back to that exact same setting.
And then lastly is the VFO button on the left hand side. So we have an ability to select an A or B VFO. Again, very similar to the kind models we are familiar with on expensive transceivers. An A or B VFO on any given memory position and then beneath that button if you press and hold that, we can turn on RIT or XIT for receiver incremental tuning or transmit incremental tuning. So to keep your receive or transmit signal stable while you tune around to follow the station who’s drifting.
And the last major function on the front panel is the memory banks, right now we have four memory banks, but actually we plan on increasing those greatly, such that if you have a favorite band or frequency you can press and hold the dial and that would indicate the memory bank that you were using on the front panel and you can turn the dial to turn to a different memory bank.
So it’s a frequency agile type of transceiver in a very conventional manner that many of us are used to whether it’s with the Icoms or my favorite radio is the Elecraft product line. K2 is my other favorite radio and it’s just very elegant from a user prospective and intuitive and that’s what we had a challenge with such a small front panel. But if you want to take it to the field, if you want low power consumption we have 90 mA. 90 Milliamps is the power consumption, the current consumption. DH1TW: Including the display?
N2APB: Including the display, but not including the SoftRock because the SoftRock adds more, a little bit more, about 100 mA in receive and 200 mA in transmit but just the SDR Cube alone with out the SoftRock is 90 mA.
DH1TW: Boy you really included a lot of functionality, I mean looking onto the front panel of the SDR Cube and the few buttons there it is not that obvious that there is so much functionality already behind it.
N2APB: It becomes an exercise in fun, actually in putting more and more features in we have to hold ourselves back from trying to do too much at one time. DH1TW: Probably next challenge would be the Apple approach, you just have one single button, you control the whole radio with the one single button.
N2APB: There you go, somebody is going to say, well I can do that on my Iphone or I can do that on my… whatever the new phones are these days and yah, you probably could but we decided to go this way. DH1TW: Great. So if we open now the box, what will we find inside? N2APB: You would find three boards that comprise the SDR Cube and they are arranged in kind of an interesting way that we, both Juha and I decided that we wanted to have this be a very, very flexible radio for homebrewers, for kit builders for people to so with whatever they wanted to do with it. So we broke the functions, the major functions into three pieces. One is the front panel, we’ve gone through that, and that’s a four inch by four inch board that sits right up behind the front panel.
The second board is the DSP board and the DSP board is about a little bit smaller, maybe four inches by three inches and it sits along the side of the radio, so if I were to be holding the radio here in front of me the…
DH1TW: I have to say for all of our European listeners, four inches is about 10 centimeters and I can see the box now and it looks freaking small! For such a cool radio it looks really small.
N2APB: (Laughs) It fits in the palm of your hand, we haven’t done the weight tests and other performance tests yet but it’s just really nice, put it in the backpack along with a battery and an antenna that you throw up into a tree.
On the side of the Cube, on the inside of course, but on the side at a 90 degree angle to the front panel is the DSP board. And the third board is 90 degrees to that on the back panel and it provides the connections to the outside world as far as microphone, the panel, your headset, your headphones, the on/off switch, and the power, and it also contains a very special connector called New PSK. Now, what we did at first since we did not have the time or processing, we didn’t have the software ready yet to perform digital mode decoding and encoding here in the software in the SDR Cube, so what we are doing at first is leveraging or providing a connection to the new PSK digital modem by this connector on the back. So what we do is send the signals, the digital signals, from the SDR Cube, which is in essence an SSB transceiver, down to, and I showed you this earlier Tobias, the digital modem. So right now we can take a digital modem connect it to the Cube and we’ve got not just AM and SSB and CW but you’ve got the digital mode.
Downstream we will take the software from the digital modem and put inside the processor fro the SDR Cube and have everything all in one so that is the evolutionary path. The upgrade that will happen maybe within a year, we’ll try, it takes time.
So coming back to your boards, you know, what’s inside the Cube are three boards. They’re kind of arranged in a U shape, the front panel, the DSP on the side and the IO board on the back that provides connections to the outside world. And then, in the other corner in a shielded aluminum enclosed area of the Cube is the area that we designate for the SoftRock. Remember that I said that anybody could take their existing SoftRocks and connect it to the SDR Cube, if they can take their SoftRock and fit in into that area of the Cube, which is about four inches by three inches by two inches, I forgot the exact dimensions, they can actually put it in there and wire it in there with the wiring harness that we have.
Actually what we designed it in, we designed the Cube to hold, the very most popular at the time SoftRock, which is called the RXTX 6.3, and the RXTX 6.3 fits really nicely in there and what we did was acquired a whole bunch of the 6.3 SoftRocks which are no longer in production, but we can be producing them for the SDR-Cube, so we can provide that as an option for those people that want to buy the kit so that they could ultimately buy the bare circuit board they could buy the full kit, they could buy the kit plus the SoftRock 6.3 or they could just use their own SoftRock externally to the SDR-Cube.
DH1TW: This provides a really a big amount of flexibility.
N2APB: Flexibility was a very big factor in our design. DH1TW: Great. So lets go a little bit into the design of the SDR Cube. Could you describe the path which a signal takes coming from the antenna and then going out on the end onto the earphones?
N2APB: Sure, Ok. I don’t know, maybe on you websit you dould include a block diagram that would correspond to this but, the antenna, the RF signal coming to you would come into the antenna which brings the signal to the SoftRock. Remember that I said that the SoftRock is the front end. So the SoftRock provides the initial mixing that brings the RF signal down to baseband much as a direct conversion receiver or transceiver would do. However, the thing with SDR, the way that SDR works most often is that it’s a new version of an old way of demodulating SSB signals called phasing. What the old time radios used to do and we do it now with DSP’s is the mixing of the RF signal to bring it down to base band produces two signals. One is called I, for in-phase and the other signal is called Q for quadrature or 90 degree out of phase. It’s the same as that I signal but it’s 90 degrees shifted or phase shifted such that ultimately the computer is going to take those signals and demodulate anything with it with the mathematics of the DSP.
So, again, the RF signal comes in, it gets mixed down to base band to produce an I and a Q signal. The I and Q signal then goes into the codec and in our case here we are using the TI codec, a Texas Instruments codec called TLV320AIC. It’s an older one but it’s around and it’s cheap and frankly it does the job pretty well so we decided to use that at least in our first versions of the DSP board. DH1TW: What is the functionality of the codec exactly? What is the codec doing? He’s receiving the I/Q signal, and then?
N2APB: Actually, it’s even simpler than that Tobias. A codec is a co-der, “co” is the coder and the “dec” of the codec is the decoder. It’s an A to D converter on the input and it’s a D to A converter on the output. That’s all it is.
Now the chips these days are getting very, very smart and powerful and they provide amplifiers in there and compensation and extra amplifiers for microphone, which is exactly what we have here, but in it’s most simplest sense a codec only converts from analog to digital on the input, so it takes those I and Q audio signals coming from the SoftRock and it converts it into a bit stream, a digital bit stream, that is input by the DSP in the next block. So the DSP just sits there and it takes in the data constantly that is coming in from the RF, from the antenna and getting mixed down.
Now it’s important to note the I and Q signals are being mixed to audio frequencies, it’s the audio baseband, so that determines the sample rate that we already have in the DSP that’s able to sample those I and Q signals in order to accurately represent and manipulate the signals that are coming in.
So, again, the signals are coming in from the codec, get changed from analog into digital. Then there is filtering that happens there too, and that’s actually some of the magic of the DSP. There’s a filter called the Hilbert Transform, on one hand it’s very simple and it’s as old as the hills, but it’s a very important on that provides a shifting of the phase and a recombination of the I and the Q. Which ultimately, once you recombine the I and Q after the Hilbert Transform inside the DSP, you have actual demodulated transmitted signal. So whatever was transmitted to you was able to be decoded essentially after that Hilbert transform and some other mathematics that happen in the DSP. Another audio filter is applied still in the digital world and that is sent out to the headphones and that is what you hear.
So let us take for example a CW that you are having. The person on the other side is sending you a continuous tone at 70 or 0 kHz and that comes into the SoftRock, gets mixed to audio frequencies, maybe 800 Hz and that 800 Hz of both I and Q are fed into the DSP. I might have skipped a step but it gets turned into digital by the codec and the DSP takes that I and Q and it manipulates it, it filters it with the Hilbert Transform and then recombines the 1 and the Q to form an actual CW tone that you can hear with all of its capabilities and puts it out to the headphones.
It’s easy for CW, it’s more complex for sideband and it’s even more complex for digital modes, but that is the basic process on the receive side.
On the transmit side, you would be, lets take your talk, and we’re doing CW, you would key your keyer, your paddle. It instructs the DSP because the PTT, the Push to Talk line going active, instructs the DSP to create a tone and an I and a Q are created of that tone, that audio tone, again the same tone but it’s put 90 degrees out of phase with respect to eachother, I and Q. That I and Q signal is fed over to the codec where the digital signal is turned into analog and you actually have an analog 800 Hz tone on the I and an 800 Hz tone on the Q that is fed to the SoftRock. The SoftRock mixes that at 800 Hz with a local oscillator, that is controlled by the DSP by the way. There is a clock that the DSP sends over to the SoftRock at 7040 kHz, and it’s actually a little bit higher tan hat but for simplicities sake it sits generating the clock for that frequency, and that LO, that local oscillator of 7040 is used to mix that 800 Hz I and Q analog signals up into the RF spectrum at 7040 kHz and it gets transmitted out to the antenna.
In the same fashion, voice processed by the system. If you key the mic, the audio is sampled first by the codec, it turns your voice signal, your analog signal of the voice into digital signals that the DSP is constantly reading. The DSP takes that, it ultimately produces an I and Q version of those audio signals. It sends it to the codec which then turns the digital into analog and sends those analog I and Q over to the SoftRock, which mixes, lets use 7300 or 14300 for an SSP signal, frequency and then it transmits it out at the 14300 frequency.
That’s the basic process, it’s really quite straightforward as the SoftRock bringing the RF down to audio, creating an I/Q signal which is read or ultimately input by the codec and the DSP, process and then put to your headphones in the same, with the D to A converter. DH1TW: So the term “processing” actually causes some fears in my ears on all of my PC based SDR solutions I’m constantly trying to minimize the delays which are introduced due to buffering and the nature of multitasking. But I also remember from university that FIR and IIR filters especially with a high order can take up quite some processing power. So have you actually measured the total processing time, the total signal delay of the SDR Cube in comparison to an analog transceiver?
N2APB: We haven’t done it but we think, we haven’t done those particular measurements yet but we think that it’s really not much of a factor at all because of the way that we are accomplishing CW is really the most problematic area because for full full QSK it’s really important to have almost no delay of course. So what we do is, and Juha came up with this technique, he is brilliant by the way, Juha, OH2NLT is a brilliant ham and among other features he is able to pump the side tone into the headphones while receiving, while your transmitting such that when you are hearing a signal and you want to respond immediately you start using the paddle and that automatically starts putting a tone into your headphones, that is the tone of your 800 Hz or whatever it is set at and that is ultimately what is going up, out the antenna.
So in essence there is no delay between what you are hearing and what you are going to respond. When I’m operating this I could turn it on just real briefly and go to CW portion of the band, I see that nobody is here in this area, maybe you can hear what (Sound of radio)… So I’ll put this in CW mode… and… (series of Morse Code beeps) That’s just to be legal, I ask if the frequency is in use… So if there were a CW coming in that I wanted to contact with, even though I’ve got the headphone signal going up to speakers such as you can hear it, it’s whenever I touch the paddle is whenever I’m hearing the side tone that is injected into the audio and I’m not depending on the audio delays that are incurred, that ultimately would be incurred going out to the transmitter that would be heard through any other kind of CW monitoring technique. So as soon as I press that paddle, I hear that side tone and that either masks or obviates, it gets rid of any of the concerns for the delay.
There are other architectures that have a more difficult time handling this because they approach it in a slightly different manner, but this worked well for us an it is certainly adequate for the kind of radio that the SDR Cube is. DH1TW: Another issue we are facing with the SoftRock type of transceivers is I/Q balancing, since analog components never have exactly the same values they always introduce a smaller or greater imbalance between the IMQ signal and phase and amplitude. This imbalance leads in a poor side band rejection, poor side band rejection causes imaging signals appearing on the bands. How did you tackle this issue in the SDR Cube?
N2APB: Side band rejection is the main issue that comes about if you don’t have… Let’s do it this way, if you don’t have good side band rejection, you really need to have your eye on the Q balance and you’re right on the money relative to the importance of that, that is one of the all important items in an SDR radio. Since everything comes down to the I and the Q, how well you handle the phase relationship of the I and the Q is all important. The definition of I and Q says it all, it’s a 90 degree phase shift between the two signals and it is imperative that the 90 degree phase shift is maintained throughout the entire chain of the signal processing in order for the integrity of that processing to be present for you to produce as good a side band rejection as possible. There are other factors that come about from not having good I/Q balance, but that is a major one.
There are many factors, there are at least several important factors that contribute to not having the right I and Q balancing, the signal path through the electronics sometimes, you might think of it as curious or not, but two identical circuits can have different phase relationships. A signal that starts off 90 degrees relative to each other and they go through identical circuits can end up with different than 90 degree separation when it ultimately gets out to your antenna or to the mixing part of the SoftRock. Similarly, on the input, there are many different stages of electronics and software that the signal goes through, each of which has the potential of making it not 90.000 but maybe 90.05. This can actually be measured, but more easily it can be empirically determined, what is a good setting. In other words, one can adjust either the I or the Q or both such that the end result, that is the opposite side band attenuation can be determined.
For example, we’ve got the bare SoftRock, it’s electronics has two out bands, in each path there an op amp, in the I path an op amp and the Q path, those op amps, even thought they are identical and have identical resistors there are saw told in the component values and the capacitor values and the reactance’s at different frequencies, even audio frequencies, and certainly different response of the op amp which respect to each other, that has to be tuned out.
In the SDR Cube, what we are able to do is essentially adjust the delay and more importantly the amplitude, equally important is the amplitude of the I and the Q signals with respect to each other, such that by adjusting them to be just right we can maximize the opposite side band rejection. We can make that opposite side band signal go as low as it needs to be. Right now, we bring it to 25 dB down, relative to the other side band. With the bare SoftRock, with no other types of processing, that’s about 21 dB, so this DSP in here and ostensibly the DSP in a sound card, if you were to be using a PC, is able to null out the opposite side band, and that is a different way of saying balancing the I and the Q signals
Further quality can be achieved by using different electronics in the SoftRock and some different techniques in the SoftRock, but again our design is limited to the use of the SoftRock right now and being able to use each of the 11,000 different SoftRocks in the field, we live with using the SoftRock in the front end. As it turns out only a purist, I think, is going to really determining or be able to tell that maybe his SoftRock is not as good as a Flexradio, you and I operating the SDR Cube for example, would have a heck of a good time digging into a contesting mix op and be able to pick out signals and filter out and actually use it as a nice transceiver.
But ultimately, the better that you have I and Q balanced, as we have been speaking here, the better type of signal you are going to receive and the better opposite side band rejection you are going to have.
DH1TW: Ok, so currently the software supports 25 dB image rejection, is that right? As a value of thumb?
N2APB: That’s about right. DH1TW: And is that sufficient or do you hear in the daily usage still some image signals on the band?
N2APB: They’re still there, I have several Cubes here in the lab and I have them on everyday, all the time, even while I’m doing my day job and listening to whether it’s to ARRL broadcasts on the CW or just some rag chewing or the Maritime Mobile band up in side band. An interesting thing with the Cube is that we are able to see, as I said before, the signals on what we call the band scope, so the live spectrum and with zero as always in the center of our display, upper side band frequencies are shown to the right of zero, lower side bands to the left, so when we’re in typical voice communications on 20 meters, we are on upper side band and I’m able to see very strong signals but they would be appearing as a low, as very low signals to the left of zero. That would be an indication where with better electronics, something more sophisticated than SoftRocks, for example, we would be able to null out the opposite side band better.
But if you’re asking me if it is good enough for now, it certainly is the signals that are coming through to my headset when I am listening, I’m not able to hear that opposite side band, it is so low, but I can see it. So a nice feature of the band scope is actually able to tell you that maybe some better I/Q balancing could happen in the system if you had the right electronics for it. But is it good enough? It certainly is and you have to use it yourself in order to determine for yourself and when we put it through some lab type of measurements we will be able to get some quantitative results that everybody will be very much looking for and we will be doing that. But for me as a user of it every day, I find it entirely enjoyable. DH1TW: Great, thank you so much for providing us this overview and giving us the insight on the SDR Cube. So now I would like to wrap up a little bit the technical part and go to a few more organizational questions. Let me start with problems. What was the biggest problem you encountered during the development phase of the SDR Cube and how did you solve it?
N2APB: Not being smart enough! At first. I don’t know if you have ever experienced this or maybe your listeners have in their own way. When you have a hard problem in front of you, and you conquer it, and you look back and you say “Jeez, that should have been an easy problem to solve but it took weeks and weeks to solve it.”
That was the case with the SDR Cube, Juha on the software side a lot, me a lot on the hardware side, we encountered some problems of, for example an FFT processing, Juha was pulling his hair out relative to not getting it right and it wasn’t working right and he had not done the FFT processing himself before. As I said, we took the application note from Microchip and there happened to be a bug in it, there was an error in the software that not even they, well they didn’t publish it, well a fix for it for sure.
Well, Juha was able, after many, many sleepless nights, literally. We have a tremendous time difference between us and then he was working till all hours of his night, like three o’clock in the morning his time in Finland, in order to solve some of these problems and he would say “George, I have success, we now have band scope,” or something. But it was only after a lot of effort to solve some of these problems that later seemed kind of easy.
DH1TW: Boy, that’s a pretty nasty one, if there is even a mistake in the Windows documentation, you wouldn’t really expect that.
N2APB: Oh yea, amplifier gain on the hardware side was a hard thing to kind of get, as you will see in some of the slides that we have done in other presentations, there are about five gain stages in a system, both on a receive and a transmit and it’s important to get the right gain setting for each of the stages because if you overload one or don’t get enough at another the system is not going to perform well. So getting that set right as a good combination between hardware and software was a challenge while we were getting to it. I don’t think we’ve got all the problems solved.
There are lots of problems, still in the design, there always will be but the more we work at it and the farther we go along the more smooth the operation becomes.
DH1TW: It sounds like a lot of time you spend on the project. Do you have a rough estimate, how many hours in total the two of you spent in the project? N2APB: A very definitive answer is no, but I can tell you although we have been working at this for a year we only really engaged full time on it back in May. So from May until now, October: May, June, July, August, September, October, so six months. Six moths of, this was the only project on my bench, other than our day jobs and it was the only project on Juha’s bench other than his day job. So we spent a lot of straight nights working on this between May and now.
It’s been a tremendous effort in order to bring it to this point. We had certain goals. We wanted to introduce it at the Tapper Digital Communications Conference, which we did and we wanted to get it out before Christmas time for people to be working on during the winter and have that kit in their hands, and so we were motivated to get it out so we spent a lot of time on it. DH1TW: This leads already to my next question which is, you already mentioned it, the kits. What is the idea? How are you going to bring now the joy of the SDR Cube to the world?
N2APB: Yea, I mentioned, I want to do the model that I’d done previously with the digital modem and that is to on one hand provide a whole range of kit options. And what I mean by that is of course, bare boards, or a full kit of parts, an option to get the enclosure or not, an option to get the SoftRock, or not. You can take any of those combinations depending on what you want to do with it. You are getting the flexibility of having the three boards separated by connectors as I was getting to before. You can put it into a different kind of enclosure, you could put it behind a R390 Receiver front panel, which is an old boat anchor, if you’re not into it, and have lots of knobs on it if you wish that way. Or you could just put it into the enclosure that we’ve got here. So there are a lot of different options from a kit builders perspective. And of course the joy of that, or one of the joys is that you can get into it at any price point that you want. From the low price with the bare boards and get the parts yourself and populate it yourself.
Or the second major path that we’ve taken that I’ve used pretty successfully with the digital modem, is to provide a fully assembled, tested box. Not everybody can put surface mount parts onto a PCB. Now we chose to use what we call 1206 size surface mount parts. They are the larger kind but still there’s a couple of big IC’s on there that are kind of dense, they are kind of hard to put on. So for the guys who really want to experiment with it from an operation perspective, or take it to the field to use it, but they don’t want to be bothered or the don’t have the time or they no longer have the capabilities or steady hands to build such the small electronics, we are going to a built op.
The next question you are probably going to ask me Tobias, is how much are all these different kit options, fully assembled version, how much is this going to cost? Well we are still collecting parts to get the kitting together and I’m working with what we call a contract manufacturer in order to automatically assemble the circuit boards and all of that, so I don’t have the pricing set yet. However, my gut is telling me it’s going to be on the low end compared to some of the other SDR solutions that are on the market now today. If I were to say around $200 US for the kit and maybe around $300 for the assembled and tested SDR Cube, that is probably the ballpark, it might be more, it might be less, but I think that’s the range we are talking about. There’s a lot of little parts involved and there is some custom machining for the cabinet and cable harnesses, and things that will make it easy for the homebrewer. But again if the homebrewer wants’ to get a bare set or maybe a bare set of parts with the boards he’s more than welcome to get in an even lower, much lower price point than what I’ve said.
DH1TW: Wow, that’s what I call a flexible pricing model so now we know how to obtain the hardware, the platform itself, but what about the software? What are your plans in the reach back for the software? Are you going to provide them with it as open source? What are your plans regarding the kind of license if you are going to provide it as open source?
N2APB: We are making everything available. We’re putting this under the open software license, under the GNU Public License, GPL, such that anybody can take this and for their own purpose, non-profit, they can take it an modify it for whatever they wish. So we will soon have all of the source code posted on our website. We will soon likewise, once all the patches are done and everything documented, we will have all the schematics up there as well, such that you can actually see the path we were speaking of before from the antenna, through the codec, through the DSP to the headset, so all of that is going to be on the website, as well.
DH1TW: And you’re going to make the software and the hardware kits available around about Christmas this year?
N2APB: Oh, oh definitely this year. Yea, we are intending to provide first availability of the kits as soon as within four weeks. So that would put it in mid November. So this is going to be very soon that we are going to be having this available to those who are willing to dive in with us.
DH1TW: Oh great! Where can my listeners find more information about the DSR Cube? Do you have already set up a webpage where they might be able to preregister for a kit?
N2APB: We do. It is www.SDR-cube.com. So it’s SDR Cube dot com, with a dash in between. That’s the website. We also have a Yahoo group, an email discussion group. The name of the Yahoo group is, SDR-cube. So that information, too, is on the website an it will point you over to the Yahoo group if you wish to join that one. And there is a lot of discussion on there already about the various features, capabilities, our design approach, why this architecture, why not that architecture both Juha and I love answering these kind of questions and being totally open about the approach that we have taken and how we have decided to solve this particular problem.
Right about now, I’m about ready to start taking a list of pre-orders for those who want to buy it as soon as we establish the price, we will start taking a list of pre orders but certainly everybody on the SDR-cube email group is going to be very much aware of when we are able to start shipping and such.
DH1TW: That’s definitely a good resource. Well, so now we are basically almost at the interview. Well, I have finally a last question and this is regarding the future of the SDR Cube. Now since you’re making more the transition from engineering into production, you’re actually preparing the kits for shipping, which will be available ‘round about Christmas, what about the future? What about the roadmap of the SDR-Cube once you have sold the kits, once you have now finished the development of the SDR, are you going to move on project or are you going to still stick to the SDR-Cube and provide future improvements?
N2APB: Oh, definitely further improvements. This is the kind of project, especially with an open software licsence and everybody’s ideas and suggestions and what about this and what about that, and a lot of software developers out there are willing and want to contribute and add in a new feature, a new capability for the display, a new mode to be supported within the software of the cube. For example, a really definitive, I mentioned it earlier, the cube needs to natively have digital mode support, that’s a big thing that we want to get to, in fact, that is a driving factor for us choosing the character of the graphic display. So not only showing all of the characters, but all of the graphic symbols that symbolize the spectrum activity is ideal for digital mode support. Right now we provide a plug for using the digital modem, an external digital modem in a very optimized way, but one of the next major features to the cube will be to upgrade the software and maybe a little bit of the hardware in order to provide digital mode capabilities within the SDR Cube. So you can actually connect a keyboard to the back panel as well, and actually do digital mode within that same box. We don’t need no stinkin’ PC, we don’t need no stinkin’ any other kind of transceiver, it’s all in the cube. So maybe a little bit of an antenna tuner like the Elecraft T1, or there’s ATU that’s very popular with the 817’s, but just a little bit of an antenna tuner and a long wire and you’re really suited well to take it out to the field.
DH1TW: Awesome! Well George, so thank you very, very much for coming up to the show tonight. Thank you so much for your contribution to Ham Radio. By making the SDR Cube with all the schematics and the entire sources available to the public, you’re definitely making a great contribution to the Ham Community.
I hope that the SDR Cube will evolve in a great crowd sourcing project with a lot of participants and maybe also worldwide. Getting rid of the PC is definitely the next step in the evolution of software defined radios, and I’m already excited to see the further improvements of the SDR Cube.
So, thanks again for joining me tonight George, it was a pleasure having you.
N2APB: Oh, very good. Well thanks very much Tobias, for the chance to come on here meet and chat with you and through you, with your different listeners, and feel free yourself and anybody that’s listening to contact me, my information will be on your website, or on your blog posting, and feel free to contact me with any questions, join in to the discussion on the Yahoo group, help give us some ideas for features and capabilities that you want in this trail friendly type of radio, using the really cool type of technology, SDR, and we will see if we can get it plugged in with you and for you and alongside you. DH1TW: Thanks George. 73 and enjoy SDR!