On 4/16/2013 12:37 PM, Michael Hartje wrote:
As I notice the application FreeDV with CODEC2 is currently the major
application for digital voice on HF?!.
Coming soon to VHF/UHF using a GMSK modem which would work with
nonlinear amplification.
The older application FDMDV is currently not found in active operation.
The developer quit. I don't know why, but I think he helped us
understand his design so that we could carry on.
So far I have seen, the modem has been taken from a digital voice
application and was reprogrammed by David.
The previous version wasn't Open Source, so David Rowe re-wrote it.
The two pilot tones take a large part of the energy modulated with BPSK
It's really one pilot tone, you just see its two symbols on the display.
Would we use an OFDM, the bandwidth could be reduced - for example,
down to 70% of the present.
You are not the first to notice this, and suggest it. The main issue is
that OFDM requires better linearity than most Amateur transceivers and
amplifiers are normally capable of. Peter Martinez G3PLX (PSK31
<http://en.wikipedia.org/wiki/PSK31>), and Rick Muething KN6KB (WINMOR
<http://www.winlink.org/WINMOR>) have advised David in the modem
development, and of course Rick used QPSK in WINMOR.
SSB equipment is just good enough for voice, and the linearity of
equipment is not a specification used in current equipment reviews,
etc., so in general hams are not selecting equipment based upon how
linear it is. Class A amplification is an optional feature of the Yaesu
FT-5000 MP or D model if I remember correctly, but that's the only ham
equipment that I've seen advertised as having a super-linear mode. More
modern means than a class A amplifier are available in commercial
applications but have not made it to Amateur radio. Over time, the use
of digital modes will influence hams to ask for better linearity and
more headroom.
We'd welcome an OFDM modem if you'd like to develop one.
2. The energy for the bits in the pilot tones certainly helps to
reduce and assist the fast synchronization and help to get down the
crest factor.
We have theorized that we could invert the clock to reduce the crest
factor, and that a receive PLL would ignore this, but nobody's
implemented it. We're not clear what effect the pilot has upon the crest
factor at this time. David Rowe seems to feel that 12 dB PAPR is "good
enough" right now, but one of the strengths of Open Source is having
people challenge assumptions like that and prove their theory with
working software.
In the specifications I have found nothing about which information is
conveyed to these bits. - I could imagine, however, that there
particularly important information is transmitted or even the Varicode
text stream to facilitate compatibility with PSK31. As it is currently
set?
The pilot's just a clock, it carries no superimposed data. The varicode
stream is sent in an otherwise unused bit of the codec frame.
3. This raises the question of whether certain bits in the packet
information to the organization always included in the same place.
Instead of the energy in the pilot tones by previously calculated bits
the crest factor could be reduced. This could also facilitate the
synchronization and error correction.
Yes, you can figure out what carriers each bit of the codec frame are
carried in, and if I remember correctly the codec frame is sent in two
halves. There is more information on this in the blog at rowetel.com and
at http://freedv.org/tiki-index.php?page=FreeDV+Specification
4. For digital voice, we are also on shortwave still in the definition
phase. So from my perspective, it does not necessarily mean that we
are restricted to a bandwidth of 1.4 kHz. With a greater bandwidth, it
would be possible to introduce more error correcting bits and
particularly to skip the frequency selective fading better than today.
There are formats like Digital Radio Mondial which provide much more FEC
and resist fading, at the cost of a long delay. We felt that having
latency so short that the communication was not significantly
interrupted would be important.
5. The RF modem is still in the further development, but we can
probably rethink it again. If we thereby obtain a better method, we
would be rewarded for the effort. How do you think about that?
We are happy to have many modems. We have roughly established an API for
modems now, and the GMSK modem is coming in with that API.
Thanks for reading
Michael, DK5HH
------------------------------------------------------------------------------
Precog is a next-generation analytics platform capable of advanced
analytics on semi-structured data. The platform includes APIs for building
apps and a phenomenal toolset for data science. Developers can use
our toolset for easy data analysis & visualization. Get a free account!
http://www2.precog.com/precogplatform/slashdotnewsletter
_______________________________________________
Freetel-codec2 mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/freetel-codec2
------------------------------------------------------------------------------
Precog is a next-generation analytics platform capable of advanced
analytics on semi-structured data. The platform includes APIs for building
apps and a phenomenal toolset for data science. Developers can use
our toolset for easy data analysis & visualization. Get a free account!
http://www2.precog.com/precogplatform/slashdotnewsletter
_______________________________________________
Freetel-codec2 mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/freetel-codec2
