Ross Whenmouth
2016-03-17 09:53:20 UTC
Hi,
Regarding "2400B" for use with unmodified FM transceivers, and Brady's
post on "FMFSK" or "MEFSK". This expired patent for the "Xerxes"
Run-Length-Limited line code might be an interesting alternative to
Manchester coding:
https://www.google.com/patents/US4437086
Quoting: "/a binary input data stream of rate 1/T bits per second is
encoded into a binary waveform having a minimum interval between
transitions of T seconds, a maximum between transitions of 2.5 T
seconds, no DC content, and a maximum value for the running integral of
the waveform of 1.5 T seconds times half the magnitude of a transition./"
So, for a hypothetical 4k8 bps, Xerxes coded, ignoring sidebands, the
minimum interval between transitions of 1/4800 seconds = max frequency
of 2400 Hz and a maximum interval between transitions of 2.5/4800
seconds = min frequency of 960 Hz, which should fit through the audio
passband of an unmodified FM voice transceiver (much like 2k4
manchester, which should occupy audio spectrum from something like 1200
Hz to 2400 Hz + modulation sidebands).
I wonder which would perform better over an FM voice channel;
say a 2k4 stream containing Codec2, plus short interleaving and 1/2 rate
k=7 convolutional coding, resulting in 4k8 bps fed into a xerxes line coder,
or say a 2k4 stream containing Codec2, fed into a manchester line coder?
An observation of real world over the air 1200 bps AFSK (APRS, packet,
etc) is that many hams appear to be overdriving the audio input of their
FM transceivers, which due to pre-emphasis, limiting, FM transmission,
and then de-emphasis in the FM receiver, results in "twist" where the
2200 Hz tone is attenuated relative to the 1200 Hz tone. I would expect
that exactly the same situation will occur with the 2400B mode for
Codec2 - many hams will inadvertently "overmodulate" their FM transceiver.
It would be nice if the 2400B mode could, at least to a certain extent,
be forgiving of overmodulation (ie twisted receive audio), because I
fear that if a deviation meter or oscilloscope is absolutely required to
set the TX audio level, many hams that might otherwise have "had a go",
may decide that it is too difficult, too technical, not worth the
effort, etc and not even bother.
As good as David's implementation of OFDM for Codec2 is, I have noticed
that some of the big boys on HF (eg users of modems according to the
STANAG protocols) appear to have moved away from technology such as a
39-parallel tone OFDM modem to what they refer to as "serial tone"
modems which use adaptive multipath channel equalisation and send a
single modulated carrier (eg STANAG 4285 & 4529). The idea appears to be
that channel equalisation will reduce intersymbol interference to an
acceptable level (even though multipath delays are much greater than
symbol durations) and that "serial tone" waveforms allow a given PA to
be operated at a higher average transmit power due to the lower
peak-to-average power compared to OFDM. Please note that I have not had
any experience with any STANAG type equipment - my knowledge is limited
to publicly available documents such as:
http://www.n2ckh.com/MARS_ALE_FORUM/MIL-STD_MODEM_PRIMER.pdf
On that subject, this paper is interesting:
http://tnrc.ajou.ac.kr/down/milcom2008/Milcom08/pdfs/499.pdf
73 ZL2WRW
Ross Whenmouth
Regarding "2400B" for use with unmodified FM transceivers, and Brady's
post on "FMFSK" or "MEFSK". This expired patent for the "Xerxes"
Run-Length-Limited line code might be an interesting alternative to
Manchester coding:
https://www.google.com/patents/US4437086
Quoting: "/a binary input data stream of rate 1/T bits per second is
encoded into a binary waveform having a minimum interval between
transitions of T seconds, a maximum between transitions of 2.5 T
seconds, no DC content, and a maximum value for the running integral of
the waveform of 1.5 T seconds times half the magnitude of a transition./"
So, for a hypothetical 4k8 bps, Xerxes coded, ignoring sidebands, the
minimum interval between transitions of 1/4800 seconds = max frequency
of 2400 Hz and a maximum interval between transitions of 2.5/4800
seconds = min frequency of 960 Hz, which should fit through the audio
passband of an unmodified FM voice transceiver (much like 2k4
manchester, which should occupy audio spectrum from something like 1200
Hz to 2400 Hz + modulation sidebands).
I wonder which would perform better over an FM voice channel;
say a 2k4 stream containing Codec2, plus short interleaving and 1/2 rate
k=7 convolutional coding, resulting in 4k8 bps fed into a xerxes line coder,
or say a 2k4 stream containing Codec2, fed into a manchester line coder?
An observation of real world over the air 1200 bps AFSK (APRS, packet,
etc) is that many hams appear to be overdriving the audio input of their
FM transceivers, which due to pre-emphasis, limiting, FM transmission,
and then de-emphasis in the FM receiver, results in "twist" where the
2200 Hz tone is attenuated relative to the 1200 Hz tone. I would expect
that exactly the same situation will occur with the 2400B mode for
Codec2 - many hams will inadvertently "overmodulate" their FM transceiver.
It would be nice if the 2400B mode could, at least to a certain extent,
be forgiving of overmodulation (ie twisted receive audio), because I
fear that if a deviation meter or oscilloscope is absolutely required to
set the TX audio level, many hams that might otherwise have "had a go",
may decide that it is too difficult, too technical, not worth the
effort, etc and not even bother.
As good as David's implementation of OFDM for Codec2 is, I have noticed
that some of the big boys on HF (eg users of modems according to the
STANAG protocols) appear to have moved away from technology such as a
39-parallel tone OFDM modem to what they refer to as "serial tone"
modems which use adaptive multipath channel equalisation and send a
single modulated carrier (eg STANAG 4285 & 4529). The idea appears to be
that channel equalisation will reduce intersymbol interference to an
acceptable level (even though multipath delays are much greater than
symbol durations) and that "serial tone" waveforms allow a given PA to
be operated at a higher average transmit power due to the lower
peak-to-average power compared to OFDM. Please note that I have not had
any experience with any STANAG type equipment - my knowledge is limited
to publicly available documents such as:
http://www.n2ckh.com/MARS_ALE_FORUM/MIL-STD_MODEM_PRIMER.pdf
On that subject, this paper is interesting:
http://tnrc.ajou.ac.kr/down/milcom2008/Milcom08/pdfs/499.pdf
73 ZL2WRW
Ross Whenmouth