Updated 20110417

This is the story of building three Softrock based radios from 2007-2011. The history before 2007 was documented by M0KGK and WB5RVZ. I have basically done it all and worn the T-Shirt on this - so don’t participate in the Softrock Yahoo group much anymore.

The highest performance SDR converts a input signal directly to digital format. The best example is the HPSDR project. An alternative approach which is cheaper and simpler is a direct conversion SDR. This uses a mixer to to produce two signals (called I and Q) which are image cancelled with digital signal processing (DSP). That overcomes the image produced by direct conversion receivers, and the mixer (called a switching or H-mode mixer) has such a high performance, that the whole circuit becomes simpler.

The real key to DC-SDR is the intermediate frequency (IF) and demodulator stages are performed by a computer. Functions like panadaptors and variable width filters that are expensive or impossible in analog technology are now cheap to do in software. The result is a radio that offers remarkable facilities for the price of a laptop, which unlike commercial ham equipment is always falling in price.

The FlexRadio company sells this technology in ready made format. But the Softrock project offers similar performance, and also uses the PowerSDR software from Flex. Also everything from the software to the hardware is open source, though it’s recommended to start with a kit from the originator of the Softrock project KB9YIG (Tony Parks).

The original “Softrock” came out in 2005, and was a small board that plugged into a computer USB port. It didn’t do anything with the USB except draw power. The unit was a receiver that worked on 40m, crystal controlled, so that’s where the “rock” part came from. Later a transceiver version came out, first with separate boards, then with a single board. By 2007 the RxTx V6.2 kit was out, and I built one.

At the time I had no HF radio equipment apart from an old shortwave radio. I was determined not to spend thousands on a black-box. Any antenna at my house would be very small, and couldn’t do justice to such an expensive purchase. So I started out with a two band, 1.0W transceiver.

By the time the RxTx V6.2 was working, I still had no proper transmitting aerial. So going on the air was put off. But it did give a taste of how to set everything up, and how it would perform as a receiver. The main limitation was the coverage was only +/- 48kHz from the crystal frequencies.

Just as I looked at putting it into an enclosure, the SDR-Kits USB controlled synthesiser came out. This introduced the Si570 from Silicon Labs. A remarkable device, for generating a crystal accurate and low noise oscillator signal. In a 5x7mm size can, it was a revolution. The SDR-kits synth was quickly ordered. This freed the Softrock from the “rock”. It could now cover 9-15MHz continuous. All under USB control.

Over the past few years the Sun has been behaving oddly. There are few, or no sunspots. This makes HF communication difficult. I figured with 1W and a basic antenna, it would be difficult to  get any long distance (DX) contacts. So the idea was hatched to add an amplifier to boost the 1W output. This was to delay the project for several months. Development of the amplifier is described on my other page.

So six months after starting the Softrock project, it was still not on the air. Finally at Christmas 2008 the project and a suitable antenna were complete. The final radio was built into a diecast box for ruggedness. Using the “Rocky” software, I finally entered the realm of PSK-31 on the 30m and 20m bands. It was fantastic to be using my own made radio to make contacts all over Europe. The best DX was the Netherlands-Antilles, about 6000km away.

This is a picture of the completed Softrock Mk.1 project.

The Softrock itself is to the top-left, the Si570 (SDR-Kits) board bottom left. My own designed amplifier to the right is larger than the other two boards. The connections are I/Q audio from a sound card, USB, band switch, power, and antenna. The unit is 222x146x50mm overall.

So in the end, the system exceeded the functionality of the Softrock RxTx V6.2 alone. At the time I only had the Rocky software working. The next stage was to try MixW alongside PowerSDR. PowerSDR is an open source application, which means individuals can modify it under the GNU Public License. At the time, PowerSDR-SR40 was the most advanced version. I never got this software to work properly, and even by spring 2009 I was still using Rocky on PSK-31. I looked at other software, but events on the hardware side overtook the Mk.1 system. The rig was eventually sold on eBay.

I realised that though it performed well, the system lacked true multi-band capability. Also the small microcontroller in the SDR-Kits board could not provide automatic switching of the amplifier low-pass filters. Solutions to these limits were available by early 2009.

A group of amateurs designed a multi-band switching board. It became the MoBo (motherboard) V3.6. This combined with a new version (V6.3) of the Softrock RxTx covered all bands from 1.6MHz (top-band) to 29MHz (10m). The RxTx V6.3 had the Si570 device onboard. So it was obvious the way to go was get these two kits and build another system from scratch. That became the Mk.2.

As shown in the picture above and described on the 20W Amplifier page, the issue A linear amplifier board had some errors on the PCB. I had to make a Softrock V6.3 RxTx, the MoBo V3.6, and design the new issue B amplifier board at the same time. These three items formed the bulk of the Mk.2 system.

To control the three analog boards, I bought a PIC18F2550 board from Sparkfun. This could run the UBW controller software. I had to alter the software slightly to switch bands on my new amplifier automatically.

Lining up the whole new system took until July 2009. The main problems were the USB interface and the Firewire audio interface (Edirol FA-66) both crashed on key-down. Fitting ferrites over all leads and strapping the units together with braid solved the problem.

Because I wanted to use the system at field day stations, it was all designed to fit into another 222x146x50mm diecast aluminium box. Unlike the Mk.1 system, all connectors had to be panel mounted to make the unit as well sealed as possible.

The MoBo V3.6 covered top-band to 10m using only four switchable filters, so the filter bandwidths were wide. My amplifier board could only cover 6 bands using 4 low-pass filters, so I altered the MoBo coverage to go from 80m to 15m. Anything above 15m will be useless due to low solar activity, so a reasonable compromise. Alongside the hardware retuning, the UBW board firmware was rewritten to give the correct switch over points. Eventually the system entered service, and was a compact unit as seen here.

At the top of the picture is the MoBo V3.6, which is on top (and in front) of the Softrock RxTx V6.3 itself. I had to cut off the corner off the PCB to make it fit. Fortunately there are no tracks or components in the corner area! To the right is the 20W amplifier issue B first prototype. The first one used T68 size toroids rather than the T50 used in the production kits. A right angled BNC connector is seen in the centre of the photo, which goes to the antenna connector, a BNC seen above the (red) screwdriver handle.

Along the bottom of the unit is the UBW/Sparkfun PIC18F2550 board, going to the sealed mini-USB connector. The Sparkfun board sits on top of a little veroboard “motherboard”. The veroboard (also called stripboard or perfboard) holds an Analog Devices ADuM1250 isolator, which uses chip scale transformer technology. This device isolates the USB and controller circuit from the analog receiver/transmitter. I didn’t notice much reduction in noise from doing this, but theoretically it’s a very good idea. Also on the veroboard circuit is an antenna band switching control. The UBW controller has several spare pins to use for controlling my antenna here.

By August 2009 the Mk2 system was in its stride, making contacts on 40m PSK-31 easily throughout Europe. It was used alongside a Samsung laptop and Edirol FA-66 audio interface as my main HF rig until the end of 2009.

On the laptop, the system was driven by the excellent PowerSDR-IQ variant written by Christos (callsign SV1EIA). I decided to load the latest copy of Ham Radio Deluxe (HRD) and link them up. The links are from Virtual Audio Cable (VAC), and Com0Com for the internal serial data. These background programs are required to link PowerSDR to HRD. I also tried MixW, but having spent so much money on the rest of the system, decided to use the free HRD software!

Going back to the photo, the connections are audio input (Tx) and audio out (Rx) on the left. Also on the left, a 9V regulator to power the Edirol FA-66 audio interface.  The FA-66 comes with a fairly quiet switch-mode power supply, but I wanted to take any switching noise from that away. In the end the 9V regulator on the rig didn’t reduce any noise, but made the system tidier. Along the bottom are - antenna control output (DIN connector), USB (Seen with blue sealing cap in place), CW key (mini-jack), power supply (4mm red/black sockets), and the BNC antenna connector.

This project won the 2009 expert category constructor's award from the Horndean amateur radio club.

In the end the Softrock Mk.2 performed well. The performance on a small wire antenna on 40m digital modes was great. This type of receiver really excels with digi-modes. The UBW firmware was also developed further by 9V1AL, to provide editable filter switching points. I would be happy with just this unit and a suitable antenna for a while. But by the time it all entered service, things had moved on again...

LINK TO TOP OF PAGE

The Mk.3 system has similar philosophy to the Mk.2 in that it’s built in a rugged diecast box. Initial construction of the project took place in the early (UK) winter of 2009. As the kit was at the beta stage, there were some errors to overcome. Most notably the A/D converter was reversed on the builder’s notes, and being a very small AD7991 chip was difficult to put right. These things are now corrected, as is typical with the early stages of such complicated projects.

In the end the MoBo V4.3 itself was quite trouble free. The power-SWR board proved surprisingly awkward, with errors in it’s measurements. However again these things are largely sorted out. A set of eight filter boards, used on both Rx and Tx has also been constructed. Usefully the group found that SMD inductors could be used, saving a lot of coil winding for the lower bands. The SMD inductors have worse performance than hand-wound toroids, but useful for testing purposes, and I’m less interested in high performance on 180/80m because I don’t have room for an antenna on those bands.

THE FOLLOWING PICTURE AND TEXT REFLECT THE LATEST STATE OF THE MK.3 PROJECT:

The rig was rebuilt because of problems with ground loops, and having no space for a display. The 20W amplifier I designed is top left. The SWR/power measurement board on the left edge. The SDR-widget controller and audio interface to the bottom right. Note how the rig looks much “cleaner” than the Mk.2, with less wiring. At top right is a 20x4 LCD module. The centre of the box is a piece of veroboard which will eventually contain an I2C expander for general purpose I/O uses, such as antenna relay switching. The controller software is open source, so I can tweak it around the edges.

The goal of a one-box solution, plugging straight into a laptop via USB is now a reality. In a field day setup, the rig just needs USB, antenna, and power to operate. The picture of the case above shows the USB socket covered with a blue dust cap. I think the nylon coated black case and red LCD give it the look of a piece of military hardware!

The software on the widget is still being developed, but the major permanent limitation is a 48kHz bandwidth in Windows. The limit is caused by lack of support for the USB Audio Class 2 (UAC2) standard from Microsoft. Linux and Mac OSX support UAC2, but the SDR software on those platforms is some way behind PowerSDR in windows.

The original microcontroller on the MoBo V4.3 is disconnected. The widget firmware now controls the rig through an I2C cable. As with the MoBo firmware, automatic band/filter switching of my own 20W amplifier circuit is supported. Special versions of the widget firmware are available on request for any remaining amplifier builders.

My Softrock Mk.3 entered service at Christmas 2009, and has been on the air sporadically during development work. I regard the Mk.3 rig and the software running on my laptop as one entity. So here, let the whinging commence... is the current list of limitations and bugs:

1. 1.Limitation - 48KHz bandwidth in Windows. Microsoft does not support UAC2 (USB Audio Class V2)
   

2. 2.Limitation - Sensitivity on 12m and 10m is poor. Preamp will be required to improve it
   

3. 3.Limitation - The long awaited PSDR V2.0.22 does not support Si570 synthesiser tuning
   

4. 4.Problem - there are many “briar patches” of spurious spurs
   

5. 5.Bug - PowerSDR loops through audio on SSB transmit causing feedback when using 2 channel sound devices
   

6. 6.Bug - WBIR (WideBand Image Rejection) does not work if total signal power is below -50dBm
   

Of these I am addressing point (2) with a simple preamp design. Point (4) can be solved by using a key sequencing app, such as “autohotkeys”. The other part of my radio is the software on a Samsung Q320 laptop. The apps are easy to find by search engines, but for completeness, here are the links for anyone that wants to get them:

Windows 7 (you have to pay for this!)

PowerSDR-IQ V2.0 beta <OR> PowerSDR (official FlexRadio) V2.0.22

Com0Com virtual serial port

VAC Virtual Audio Cable (you have to pay for this!)

Ham Radio Deluxe DM780 V5 beta

Ham Radio Deluxe Logbook V5 beta

MiniDeluxe console

WSPR V2.1 (Weak Signal Propagation Reporter)

Voice Shaper

Hopefully this page is useful for those in the Softrock community to show the development of these rigs though the years, though the MoBo V4.3 is rather different to the present “ensemble” kits sold by KB9YIG. It should also demonstrate what can be done by the dedicated experimenter with the concept.

The SR Mk.3 won the 2010 Expert category project prize from my local amateur radio club.