Morse code qualifies as a digital mode, although organic brains are somewhat better at copying it than electronic ones. Ham radio operators that did “phone” (ham-talk for voice) started out with AM modulation. Sometime after World War II, there was widespread adoption of single side band or SSB. SSB takes up less bandwidth and is more reliable than AM modulation. On the digital side, hams turned to different and more sophisticated digital transmission types with computers pushing bandwidth down and reliability up. However, a recent trend has been to encode voice over ham radio–sort of VoIP with radio instead of Ethernet–using an open source program called freedv.
[AA6E] made a very informative video where he carries on a QSO (a conversation) with a distant station using freedv. What makes it interesting, is towards the end when the two stations switch to regular SSB. The difference is dramatic and really points out how even with less bandwidth (roughly 3 kHz for SSB vs 1.25 kHz), the digital mode is superior. The freedv software (available for Windows or Linux) compresses audio to 700-1600 bits per second and spreads it over 16 QPSK signals.
Even if you don’t have a ham radio license (and if not, why not?), it is interesting to see how transmitting a complex signal like voice using digital methods compares with the old analog-style transmissions.
FreeDV is also available for Mac.
Kind of. Latest build went up over the weekend, in readiness for the QSO party this weekend, but it is not yet tested on all versions of OSX YMMV. I don’t think the FreeDV site has been updated (give it a day or two) but a link can be found on areg.org.au
I’m going to be a rebel and sing happy birthday to someone.
Don’t worry, no one would confuse your singing with ‘music’ ;-)
Right now Digital Voice is fragments to heck and back. Icom wants to back their standard, Yaesu backs theirs, etc…
One of the advantages of the digital modes is repeaters are on the internet so I can connect to a Dstar repeater in Chicago, and talk to ham’s in LA from my car as I drive down the road. I can even have a group convo with friends in different states.
FreeDV is cool, but digital modes like WSPR are almost spooky. http://www.arrl.org/weak-signal-modes
Guys are using the moon as a cheap global communication satellite and pulling signal out of the noise floor.
FreeDV is good *precisely* because Icom wants their own standard and Yaesu wants theirs. With FreeDV anyone can use it, so everyone can use it. There is no need for proprietary standards.
One of the reasons that DStar did not gain traction is due to the proprietary nature of the AMBE codec. If FreeDV had existed at the time, DStar could have used that instead. As I see it DStar and other proprietary systems will go by the wayside as FreeDV becomes more widely-known and popular.
Actually, I don’t think it will matter whether the codec in use is open source or not; most hams at this time have little interest in the digital voice modes, compared to using FM, even though digital voice modes tend to be more spectrally efficient than FM or SSB are. Let’s not forget that FreeDV isn’t the first attempt at a narrow bandwidth digiatl voice modem on HF.
I can’t say the previous attempts in the 1980 and 90’s were very good, though. They used a lot more bandwidth and depended on excellent conditions. We have a lot more computer power today, which has helped. We can beat SSB in poor conditions with the 700B mode (which isn’t in this video). Actually working better than SSB is probably a tipping-point for HF digital.
@bruceperens – I’m not sure but the first attempt at digitizing voice communications over HF radio for encryption purposes was during WW2 in 1940’s. The NSA museum has an exhibit called X System, Project X, Ciphony I, the Green Hornet or just SIGSALY. It was a successful method of encrypting voice 2-way radio communications via companded PCM for the POTUS to secretly speak in full duplex with UKPM from Pentagon to Downing St (via Oxford St and a hardline to Downing) and I think a military vessel at sea for the high ranking general to speak securely with POTUS. There were 12 all together installed worldwide. The NAZIs where very aware of it but could not break it like we broke their ENIGMA system.
I disagree. The patents, proprietary standards and fragmentation are the difference between needing a $400 Icom radio and a sub-$40 Baofeng or any other brand. The actual processing requirements for digital voice modes are tiny by modern standards, and every one of those dirt cheap Baofeng radios is already SDR-based because it’s such a cheap way to do things, they’re just ROM programmed with an FM-only decoder as that’s the only widely-used voice standard that’s free to implement. If digital voice was that cheap and easy to get into, I have a feeling a lot more hams would use it.
DStar didn’t gain traction? Tell that to all the people using it here in my area! It does seem to have quite a bit of traction here.. much to my chagrin. Now the local club would have to give up an expensive repeater and a lot of individuals would have to give up their expensive radios for there to be a switch to an open codec.
WSPR can be a lot of fun. I used it to work a guy in Perth, Australia from New England, and neither one of us was using more than 5 watts of RF output power. Spooky? Maybe, but definitely interesting.
I really doubt the digital modulation is better. Yes, it uses less bandwidth and yes the audio was better but as the signal strength decreases (and the SSB portion in this video was a weaker signal, no doubt) the demodulation fuckups will increase until the voice is so garbled up one can’t even understand it properly.
Also, the digital modulation uses a lot more power than SSB due to its “always on” nature. There always has to be a bytestream while the microphone is keyed. I wonder how susceptible the receiver is to varying symbol rates!?
I also noticed a sinewave like modulation in the spectrum. Can somebody elaborate on that? Is it a technique to track the signal better by cnvoluting it with a moving sine spectrum or was that an almost-DC part of the signal coming through the audio? Inaudible, but still modulated on the baseband.
That ‘modulation’ you’re seeing is frequency selective fading, which is characteristic of the multi-path nature of ionospheric propagation. It’s one of the main problems of the ionospheric channel which HF modems need to overcome.
The modem in use in the video is the 1600 bit/s FreeDV mode – yes, with low SNR the bit errors will garble speech, and it doesn’t work at the SNR levels which SSB can still be usable.
There is a 700bit/s mode currently released (though still in constant development) which performs much better in low SNR conditions, and uses a somewhat different modem. There are some good comparisons of the different modes and how they handle fading and low SNR here: http://www.rowetel.com/blog/?p=4291
Thanks for pointing that out! I didn’t think about frequency selective fading especially with such a small bandwidth. I’m used to UHF/VHF transmissions where the ionosphere does not come into play.
That rowetel.com article also looks very interesting! the simulations carried out by Matlab/Octave are good engineering practice! :)
As the signal strength decreases we will indeed reach a point where the voice can no longer be understood.
Well done! With such insight I can see how you managed to acquire your ham licence.
Without body language I can’t tell if your “Well done!” is sarcasm or not. But I guess that doesn’t really matter :)
The symbol rate thing in relation to sound quality bugs me a lot. We have made measurements using 4-PSK and changed the symbol rate by just 1Hz (I think we had 1 or 2kHz symbolrates) which completely destroyed the constellation diagram:
(Good lock on the source signal) http://abload.de/img/goodf6kt8.jpg
(Bad lock on source) http://abload.de/img/notgoodz9k08.jpg
While your modem obviously had problems, we’ve dealt with this (and many other issues) already. We can handle significant differences in the symbol clock frequency at the transmitter and receiver. The mode shown can handle sound cards that are 200 Hz out of sync at 8 KHz.
So the mode is able to recover the symbol rate – Awesome! We were investigating the effects of symbol rate errors on the bitrate error and how we can avoid them. Obviously a PLL was the way to recover the SR :)
If anyone else is interested, I was reading an interesting GNU Radio tutorial the other day that explained how to simulate problems like frequency and clock offsets using it, demonstrating the effect they have on QPSK signals and one way to deal with them: http://gnuradio.org/redmine/projects/gnuradio/wiki/Guided_Tutorial_PSK_Demodulation
The sine signal appearing on the modulation is short time fading. This is what you also hear when listening to a short wave broadcast sometimes, when the signal goes away and comes back. It happens because there is more than one path through the ionosphere. These paths can interfere destructively or constructively. For ODFM type modulations it does not matter, since the pilot tones will have the same distortion, so it cancels all out.
I’m not talking about fading here! Look closely at the received signal within the filters bandwidth: you will see there’s a kind of sine shaped amplitude distribution which wanders to the left :)
Screw my comment. I didn’t think about frequency selective fading. I’m a herpderp :)
This is actually due to multiple reflection altitudes in the ionosphere. It happens at your antenna where the multiple paths combine. There are a lot of papers about it. The best scheme to mitigate it is probably polarization diversity with two receivers and parallel decode. The cheaper way is circularly-polarized receive antennas.
FYI, That peak in the middle of the spectrum is a pair of BPSK pilot tones used to sync the modem.
The latest modem he’s designed is a coherent QPSK and the pilots have been moved to the other carriers. It has 14 carriers (7 carriers + 7 copies as a diversity channel). With a 700 bit/s vocoder, it isn’t designed for FM quality, but instead is designed to allow the ears to pull out the speech. It has been tested by many down to the noise, and seems to be a valid design path. My uncle is playing with a dissected version on github: https://github.com/k5okc/DVModem
How much processing power does it take? Would it be realistic to port it to run on a microcontroller?
It has been ported to run on a STM32F:
http://www.rowetel.com/blog/?p=3125
The FreeDV 1600 mode used above has been ported to a microcontroller. Right now it’s running on the STM34F407, with a 168 Mhz arm cortex m4f core. Most of the code uses floating point math for the DSP, so it’s either got to have hardware floating point or be fast enough for software floating point to work.
Seems a bit excessive, 168MHz m4f, but maybe I’m wrong, it just feels a lot.I mean the f means FPU and it still needs that speed?
The development board is $15 at Mouser. It’s not the speed, it is availability and inexpensive. The CPU has more than enough horsepower, but memory is limited.
I’m not sure about RX, but TX only uses around 25% of cpu cycles.
That’s running the modem as well as the codec. In theory this could be re-done in fixed-point, in practice the CPU is so cheap that nobody cares, and we’re more interested in developing software than optimizing.
FreeDV is awesome! I’ve been smitten by everything PSK. I really want to play with this. Does anyone know if you are using a SSB mode or AM mode when using FreeDV in transmit mode? Also does anyone have an example of how the raw signal sounds on a normal non-FreeDV-equipped SW receiver? Or can anyone describe what it sounds like?
Use SSB. To me it sounds like a motor running. Modulate to about 1/3 the audio level you use for voice.
Here’s a sample of the modem sending 700 bit/s using Coherent PSK: https://goo.gl/JxC54V
Wow it does sound like a motor running. I wonder if it changes with the voice pattern or is it a constant steady data signal. Can a person identify it from Packet, RTTY. AMTOR, etc?
I guess this could be used for EME communications. Just find the moon and shoot it with your Yagi-Uda. Probably no need for 1KW ERP?
It’s pure data, the voice is just binary bits at this point (2 bits per QPSK carrier). You can see the carriers if you use audacity to look at the audio spectrum. But it is on 100% so you will have to reduce power or your heat sink will start glowing red hot :-O
From what I gather from the ARRL web site is that this thing can be run at QRP (via EME?) and still pull out the sig. Not like the old EME days of running a 1KW ERP to moon-bounce. So even running MAYBE 12W SSB (ERP) should do the trick? I wonder how much better tropo and cloud bouncing would be. Imagine modulating light with this using telescopes LOS. Maybe achieve a lot further distance than a hundred miles. Of course because of the horizon you’d have to aim for high mountain peaks to do better than 100 miles with a Blue/Green LED, t-scopes, Fresnel Lens, and photo transistors. Blue/Green to help penetrate precipitation in the air (i.e. USN discovery in 1970’s). Forget LASERs. Too finicky to aim.
Many thanks to Peter Martinez G3PLX (father of AMTOR) who invented PSK-31 Dec-98 (based on Paweł Jałocha SP9VRC original idea).
Or use a bigger cooling fan! :)
I am scheduled to take my ham test in a month, so overwhelmed where to start. is there a quick guide i can read to orient me better? All I understand is FM and AM, playing with my little SDR dongle I see lots of these modes but don’t understand what is happening. Do these guys know each other in the audio or did they just run into one another? How do they know to keep talking and when to say their handles? Do HAMs mostly just talk about their equipment?:
Im confused
There’s all kinds of hams – those that talk all night, those that only come out for contests, others just use a license to experiment with other hams. A ham license gives experimenters an outlet for radio communications at high power. I would recommend downloading the Technician study guide PDF, and every day read only the “right” answers. After a couple of weeks you won’t know the question, but you’ll know the answers. Later you can learn what you want at your own pace (with a license).
@Tyler – No they don’t normally know each other. They can look up each other’s home address in Call Sign books, CDs, or web sites.
FM and AM are just a small selection of modulation schemes. There are MANY!
You need to get or make a good reception antenna for your SDR. The bandwidth is large with those things and one antenna does not normally fit all. More expensive Discone antennas however kinda’ sorta’ do though. Transmitters require much more exactitude in electrical length than do receivers.
Hams NEVER have any trouble in knowing how to keep talking (LOL). Some would call that “rag chewing” but there is a reason for it. Amateur Radio was designed for emergency communications during times of disaster etc. Any high-tech experimentation between those times is key to the development of the craft. So any long-winded convos tend to help others to tune VFO’s etc. Sometimes they are trying to over come fading due to the ionosphere. Repeating yourself or stretching out sentences allows you to tweak your rig or others to copy what you said during a severe signal fade out.
The FCC requires that you identify your station call sign periodically. It’s in the rule books. Hams never use “handles” only call signs and REAL names. No made up fictitious names. And yes the techy convos are usually about their radio shack and what’s in them and the antenna (not really a prideful thing – just helps others know the specs of the rig their are listening to). But at times they talk about friends and families. But during emergencies it’s only serious business like passing emergency messages to disaster victims families, first responders, military, etc. Rag chewing is suspended for the calmer periods.
The contests are not just for fun. It is to hone their skills and demonstrate that they can do the job of communicating with the world when others can not. Your cell phone or landline will not always work (i.e. Hurricane Katrina?). HAM radio can. They even sometimes have their own solar, wind, hydro, or generator power. They can also sometimes repair their own stuff like a wind damaged antenna. Your local police department has to call someone to do that. Not HAMs, as they usually are the ones called as they sometimes work at radio electronics service companies too.
Have you ever listened to an phone autopatch on 2-meters FM? Notice it’s not a toy. It is to help with public safety by allowing a HAM to use it for telephonic landline communications in emergencies. However, a little bit of family matters do crop up “Honey I’ll be late for dinner…” but I don’t hear much of that lately as there are club rules about autopatch usage limitations. Also you might be fascinated about watching SLOWSCAN and ATV videos over the air. PSK-31 allows for free-software real-time RF chat sessions using just your PC or MAC and a authorized RF transceiver (one you are authorized to use?). It just uses the audio channel (mic and loudspeaker).
All you hard-core HAMS give me a break. I’m just simplifying the concept to be less confusing for Tyler (et al)… No criticisms please (LOL)
1. The Author of this post sounds like a Digital Voice (DV) “Fanboy”. Analog has a distinct advantage for small signal work (low signal-to-noise ratio). As soon as you replace the brain’s ability to “pull” signals out of noice with a “relatively” hard-decision codec, you lose the small signal reception ability.
2. The evolution of DV is being over-run by commercial interests. Even if you try to introduce a free and open digital voice codec, the large equipment manufacturers will do everything they can to put you under the table. They want to OWN the DV codec and force you to pay outrageous prices to use it. This is a GOOD reason for using open-source DV. Something the Author of this post does not emphasize at all. The ultimate stupidity of the big equipment manufacturers is that in the end, if they backed an open DV codec, they would end up selling far more radios.
My only comment, is that the short SSB trial right at the end, was to my ears very poorly setup, with way too much bass (using up most of the available TX power) and not enough top (where the “information” is) as evidenced by the spectrum at the time. If that had been sorted out, the redability would be even better, much better as “analogue” SSB.
Not to say FreeDV is bad, it appears stunningly good at withstanding the selective fading you could see sweeping through the spectrum, and is obviously the way things will go for “appliance operator” (non ham) HF radio in the future. Codan among others are already going there…
I did try to install and trial FreeDV after seeing this video, but for whatever reason, it decides it cant access or use the internal soundcard in my PC! (Win7 Home Premium, 32 bit. AMD CPU, Nvidia GPU, AC97 audio.) The external iMic device that the rig (currently) is hooked up to was found and selected, but even though the internal speakers were found and selected, the test tone (or whatever) was not played.
Will have to see if I can contact the FreeDV boys for advice.
73