FreeDV - open source digital voice

I had been sceptical of digital voice modes as the standards supported by the major manufactures are all proprietary. This will lead to competing 'standards' along the lines of Betamax vs VHS where early adopters are the losers.

Recently I became aware of an open source standard which will work on existing pc's called FreeDV. I have also realised that digital voice on HF is no different to RTTY, PSK or JT65 for communicating because those modes are also only able to be decoded by computer. Given the woeful state of my CW I may as well add that to the list as well! 

My computer runs Ubuntu 14.04 with the xfce windows manager. There is a FreeDv package available from the FreeDV link above which works with this configuration. I did try to install it on Linux Mint 16 but had problems.

The trickiest part of configuring FreeDV is setting up the sound cards. Two are needed. One can be the standard sound card built into the computer and the other can be the sound card that is built into a USB headset (headphones and microphone).

To configure the 4 inputs and outputs I needed to use alsamixer and pavucontrol (both available from the Ubuntu Software Centre).

alsamixer setup for the internal sound card used for the audio input from radio to computer (using the computer line in socket) (1) Note that 'Input source' is set to the input you are using to get audio from the radio and 'Capture' is used to set the level for that input source.

 alsamixer setup for the Logitech headset headphones (2)

alsamixer setup for the Logitech headset microphone (3)

alsamixer setup for the internal sound card used for the audio output from computer to radio (to be honest I don't know which setting is used here but the socket on the computer is audio out) (4)

 

This is the FreeDV audio configuration screen corresponding to screenshots 1 & 2 above

This is the FreeDV audio configuration screen corresponding to screenshots 3 & 4 above

I also had to use pavucontrol (PulseAudio) to adjust the transmitted audio so it isn't over driving the radio, although I haven't experimented enough to see if I could also do this using alsamixer. I think I should be able to.

I have a Yaesu FT-920 and the audio in/out from the computer is connected to the data port along with the USB-serial CAT cable for PTT control.

What the FreeDV waterfall looks like without a digital signal being received.

That's it for now, there are more instructions for setting up and using FreeDV here.

If you would like to set up a sked please let me know, my email is on the top right of this blog or on QRZ.com (G0FCU).

Arduino Antenna Analyser

Inverted-L SWR plot

The alternative title for this post could be "Even more fun with antennas"! 

I know that the inverted-L that I erected recently resonates on the bands that I wanted it too (i.e. 10/14/18Mhz) but tuning wire antennas to resonance always involves a lot of running between ends of the antenna and rig, taking notes of SWR achieved at specific frequencies and working out whether to shorten or lengthen the wire. 

An antenna analyser can help reduce some of this effort as you get a graph of SWR plotted against frequency. However, these cost upwards of £200. Now bring to the table the ubiquitous Arduino, my favourite 'toy' tool of the moment.

I found several designs for antenna analysers based on Arduino's but I think it was someone on Twitter who pointed me to the design by K6BEZ. If you have an Arduino and a few resistors this is really cheap to make. I had to buy the diodes and the op-amp, but these cost me less than £10 on e-bay with an op-amp and about 10 diodes to spare.

Fortunately I also found Matt Gumbley's website and Matt has designed a Veroboard layout for the K6BEZ circuit. This made building the circuit very simple and it worked first time for me. I didn't have any 50 ohm or 648 ohm resistors so I used a 27R and 22R in series (49R) in place of the 50R resistors and a 470R & 180R in series (650R) in place of the 648R resistors.

Setting the frequency range to scan involves entering the commands below via the Arduino serial monitor. The key commands in the Arduino sketch are:

• frequencyA - set the frequency to start the scan from, in Hz.
• frequencyB - set the frequency to end the scan at, in Hz.
• stepsN - set the number of steps for the scan
• ? - displays the parameters that have been set
• S or s - start the scan

e.g. 12000000A sets the start frequency to 12Mhz.

For a 1Mhz to 30Mhz scan I use 2000 steps.

The results are returned to the serial monitor and in the default sketch are:

• frequency,0,swr,forward voltage, reverse voltage

To get the graphs shown here I have simply copied the data into a spreadsheet, converted the frequency to MHz by dividing by 1000000, and divided the SWR by 1000. I have then used gnuplot to create the graphs.

The gnuplot commands I used are:

set term canvas size 600,400

set output "g5rv.html"

set linetype 1 lw 1 lc rgb "blue" pointtype 0

set xlabel 'Frequency (MHz)'

set ylabel 'SWR'

set xtics scale 2,1

set mxtics 5

plot 'data' smooth bezier title 'G5RV antenna'

Line 2 sets the output filename  to 'g5rv.html'.
Line 8 loads the data from a file called 'data' (frequency and swr, one pair of data values per line, separated by a space) and give the output the title 'G5RV antenna'.

G5RV SWR plot

5 MHz dipole SWR plot

What this shows me is that my G5RV resonates slightly higher in frequency than the 80m & 40m amateur bands but that my 5 MHz dipole and the Inverted-L resonate exactly where I want them to.

Even better is that I have got an antenna analyser for less than the price of a pint of beer. Cheers!

WSPR comparison between Inverted-L and G5RV

WSPR spots using 5W on 20m with my Inverted-L

I managed to get wsprx working in receive mode on my new installation of Linux Mint 16. The realtime waterfall makes it much easier to ensure that your radio is accurately set to receive in the wspr sub-band. I found that my FT920 reads about 40Hz low as a result. However, it crashes when coming out of transmit mode. Consequently, I installed the older version wspr 2.0. This also had a problem with one of the Python libraries but there was a fix documented on the wsprnet.org website.

The waterfall used in wsprx

This means that I can now run wspr with a little more power than previously, my broadband HF amplifier only produces about 0.5W. I have been wspr'ing with 5W and the results using my new inverted-L you can see in the screenshot at the top of the page. I am amazed at how far 5W will reach. However, it is a little too easy, on my first transmission my signal was received by KK7UQ on the west coast of the USA.

After an hour wspr'ing on the inverted-L I switched to my G5RV and the difference is considerable. Although the G5RV is higher than the inverted-L (around 12m versus 6m high) one leg runs E-W and the other leg NW-SE. The inverted-L runs SE-NW pointing directly at north America. The resulting wspr spots from the G5RV are below.

WSPR spots using 5W on 20m with my G5RV

Interestingly I received VK3HAD equally well on both antennas and KK7UQ received me equally well until dusk fell when a few more US stations heard me and were heard by me. 

Along the greyline I was heard by ZL3DMH at 19016km distance.

This isn't a scientific test, by any stretch of the imagination but very interesting none the less. However, what it has shown me is that 5W and wire antennas equals QRO in wspr terms!

Inverted L and results from WSPR

Good results with my new Inverted-L overnight. Running 500mW I got multiple decodes from the stations you can see on the map above.

The tx is the same homebrew amplifier that I have used previously so the antenna is the only item I have changed. With my other antenna, a G5RV, I had only been received once in the USA.

Overall this confirms my initial findings that the Inverted-L pointing at north America is performing very well.

4 metres contest

Screenshot of my QSO's in the 70Mhz Cumulative contest on  25 May 2014

Many years ago I participated in many VHF and UHF contests with the G4VAT contest group so I sort of got drawn into this 70Mhz contest today. I have described my new 70Mhz elsewhere on this blog and this was a good opportunity to give it a good try out with my homebuilt Moxon Rectangle antenna.

The result was that I worked all the stations that I heard, except one who was a long 230 miles away. He heard me on an aircraft reflection but it didn't last long enough for a full exchange.

I was pleased to note that most signal reports I received were reciprocal to those that I gave out. There was deep QSB on most signals and comments made by stations who have submitted logs also note the QSB and generally poor conditions.

Here is a link to my contest QSO's in an interactive Google map

I have just submitted my log to the RSGB contest committee site and I have claimed the grand total of 825 points!

Spring fun with antennas

The part built 4:1 balun

The May weather was very warm at the weekend and so I took the opportunity to put another antenna up. I spent some time trying to decide what to build but eventually decided that anything was better than nothing and just got on with it.

For some time I had thought about erecting the Inverted-L described here cut for the 30/20/17m bands. First though I needed a 4:1balun. I built one described on the website of M0ZPK

I then retrieved from the shed a roll of 32 gauge wire given to me by a work colleague some years ago. I suspect this wire is quite old as it is manufactured by WL Gore, who are now rather more famous as the manufacturer of the waterproof material GoreTex.

This wire is so thin as to render it almost invisible, which makes it difficult to erect but pleasing on the eye in the longer term. I am sure it won't last a winter but it was free so is great for experimentation.

The inverted L was hoisted about 7m into a tree in one corner of the garden with the top running down to a smaller tree at the front of our property. Here I had to bend the final few metres (of 37 metres) to make it fit. The antenna was immediately a good match on 15/12/10m, which was annoying as I was trying for 30/20/17m.

I fiddled around with it for a while (and actually shortened it so there is now no bend in it, but it still didn't resonate on the desired bands) before leaving it for the night. The next day I relocated the balun closer to the ground and the stream that runs along the back of the garden. Hey presto, a 1.1:1 match on 30/20/17m was achieved. Obviously the antenna elves had been out during the night and tuned it to resonance for me! I can only think that changing the angle of the more vertical part of the antenna had a significant effect.

The result is an antenna that is directed towards north America and I can hear Canadian and US stations on it that are several S points down on my more elevated G5RV.

So I am classifying this as a great success, and it has caused me to have more QSO's which is even better. Here are a couple of videos showing the difference between the inverted L and the G5RV, the G5RV is antenna A and the inverted L is antenna B. The antenna input selected is shown to the top right of the frequency display. OK, so you cannot make out the A and B on the video, but take it from me that the inv-L is the stronger signal as I switch between the two.

OK, here is one, I'll add the other later.

70Mhz transverter & Moxon Rectangle beam finished

At long last I have completed my 70Mhz transverter, which I first wrote about here. Since I took that video I have rebuilt the transmit stages using Manhatten style construction. I have also replaced the tx/rx switching circuit with a much simpler arrangement. 

Whilst it is probably not the most sensitive receiver around I am proud of the fact that I built it from scratch. It is the design by EI9GQ published in Radcom in June & July 2009.

The transmit output is now 100mW. To bring this up to a level to drive my Pye A200 amplifier I bought a ready constructed amplifier from SP2DMB For 100mW input this gives 8W output. Using this to drive the A200 gives well in excess of 50W but I have turned the A200 down to avoid over-driving it so it only outputs 50W.

I have learnt so much from this project about receivers, transmitters, construction techniques and fault finding. Unfortunately I have also found, along with my optician, that my close up vision is now very different between each eye. This makes soldering difficult as I find it difficult to judge depth of field, so any sort of SMD construction is definitely not for me. Another visit to the optician is due and I'll try to get a pair of glasses that makes soldering easier.

To augment my horizontal dipole I have made a Moxon Rectangle antenna for 4m using this design. Now this is on the mast I have found that I can receive the GB3BUX, GB3RAL and GW3MHW beacons all the time rather than only by aircraft scatter.

The distance between my QTH and GB3BUX is 153 miles. The path profile is shown below:

My QTH is on the left, the first piece of high ground is the North Downs in Surrey, the second is the Chilterns around High Wycombe and the next piece of high ground is the Peak District south of Buxton.

I have left the dipole on the mast attached to my FunCube Dongle SDR. I'll use this for spotting Sporadic-E openings. I noticed last year that some of the tones I receive around 70.230 - 70.240 Mhz get much, much stronger when there is Sp-E around, so they are a useful indicator of propagation.

Now, roll on the Sporadic-E season!