| |
PA2AGA > HDDIG 17.07.00 16:40l 161 Lines 6789 Bytes #-9389 (0) @ EU
BID : HD_2000_191B
Read: GUEST
Subj: HamDigitalDigest 2000/191B
Path: DB0AAB<DB0SL<DB0RGB<OK0PPL<OK0PPR<OK0PJI<OM0PBB<SR9ZAA<EC1I<EA7URC<
PE1NMB<PI8HGL
Sent: 000717/0135Z @:PI8HGL.#ZH1.NLD.EU #:61854 [Den Haag] FBB $:HD_2000_191B
From: PA2AGA@PI8HGL.#ZH1.NLD.EU
To : HDDIG@EU
Date: Mon, 17 Jul 00 01:30:53 MET
Message-Id: <hd_2000_191B>
From: pa2aga@pe1mvx.ampr.org
To: hd_broadcast@pa2aga.ampr.org
X-BBS-Msg-Type: B
coded modulation! With coded modulation it is often impossible to separate
the channel coder from the modulator since they are intrinsicaly bound up
together. Neat idea! Thanks Gord and Gottfried! Gordon invented "lattice
coded modulation" in the late 1970's [1976 - 1978] at Motorola in Canada and
designed and built a Leech Lattice modem. The first coded modulation
system. The Leech Lattice is a "natural" lattice or polytope in
24-dimensional space which has a sphere packing density very close to the
maximum possilble. John Leech, now deceased was a Canadian mathematician
employed at University of Waterloo. Later in a paper published in 1982 and
in patents granted later Gottfried Ungerboeck who was then working for IBM
at its' Zurich Switzeland laboratory invented "trellis coded modulation" in
which a covolutional channel coder was married to a phase shift modulator.
Late Gottfreid's trellis coded modulation was generalized by many others and
is now the most widely used coded modulation system [Or probably the most
widely used modulation system period!] in use. All modern telephone modems
such as the ones we are using as we type here on the NG use trellis coded
modulation. V.32, V.34, and V.90 modems all use trellis coded modulation!
Back to CW and its' amazingly efficient but simple to implement
capabilities.
I believe that in the conventional sense CW as a channel coding system is
"ternary". It has three and only three symbols, the dot, the space, and the
dash. Call them A, B and C if you wish. This is independent of time and
unit intervals etc... it is strictly "channel coding".
Then the classical modulation method for CW is straight forward
on-off-keying [OOK] of a "continuous" carrier. Of course the ternary coded
channel coder output could be used to "key" many different modulators, like
PSK, or FSK, or QAM, or whatever...
So I say CW channel coding is ternary, not binary.
Peter K1PO
"Steve Sampson" <ssampson@usa-site.net> wrote in message
news:smnlfcfand6116@corp.supernews.com...
> That sounds like the most reasonable description to me...
>
> It looks like pulse radar with poor leading and trailing pulse edges :-)
>
> "Klein Gilhousen" wrote
> >
> [snip]
> > By the way, I think that Morse CW is a binary modulation system (not
> > ternary or whatever). There are two signalling elements: transmission
> > of a dot length pulse, and a dot length silence.
>
>
------------------------------
Date: Sat, 15 Jul 2000 21:00:43 -0500
From: "Peter O. Brackett" <ab4bc@ix.netcom.com>
Subject: CW versus hi speed digital etc.
Ed:
What you are discussing here is approximate "matched filtering".
Matched filtering sounds exotic, but its' really simple! [In theory a
perfect matched filter is impossible to achieve since it requires that the
filter be anti-causal, that is the output comes out before the input
arrives. A good time domain description of the impulse response of an
ideal matched filter is that its' impulse response must emulate or produce
the time reversed image of the desired signal!] In practice what digital
receiver designers do is to approximate the ideal matched filter in some
sense. This approximation is usually done in the frequency domain. When a
matched filter is approximated in the frequency domain it of course will
have a causal impulse response and does not respond before its' inputs
arrive, and to that extent it deviates from and is only an approximation to
a matched filter. Now if you look at the frequency response of the matched
filter and "eyeballed" it and also "eyeballed" the frequecy response of the
desired signal you will immediately see thaty their frequency responses are
approximately the same! Simple idea but arrived at by abstract mathematical
reasoning. This is just the sort of filter any unsophisticate ham would
have used! Look at the transmitted signal, and then use a receive filter
approximately the same shape! What? Simple, eh.
Now one needs to ask how close to the ideal unrealizable matched filter do
actual implemented matched filters become and how would you measure the
deviation? Measure it how you wish in terms of mean squared error or
deviations in dB of the frequency response averaged over frequency bands of
interest or whatever...
You will find that good transceiver CW filters are often much more closely
"matched" to CW signals than most modern digital system matched filters
are... essentially those modern digital matched filter designers "skimp" on
their designs. And so if you operate CW and use your transceiver's CW
filters, don't worry... for all practical purposes you are using matched
filtering! Period, end of story.
When you can't hear a CW signal in a 2 kHz band and you cut in your
transceiver's 250 Hz or 100 Hz CW filter and it magically pops up out of the
noise...
Well that's "matched filtering".
CW can use matched filtering just as any modern digital system can, and
what's more...
Most CW filters in modern ham transceivers are better matched filters than
the so called "matched filters" in digital systems. i.e. the matching is
closer!
Ed, I hope this helps, and it helps some of the undereducated digital bigots
to appreciate matched filtering a bit more.
Peter K1PO
"Ed Hare, W1RFI" <w1rfi@arrl.net> wrote in message
news:sIMb5.317$c5.773901@news.ntplx.net...
> Klein Gilhousen <kleing@NOSPAMqualcomm.com> wrote:
>
> > In principal, OOK (CW) ought to be poorer in performance than PSK in a
> > Gaussian channel by a few dB. But we have anecdotes claiming that
> > human operators can match the performance of some of the digital
> > modes.
>
> Actually, most of the anecdotes about human ability to copy CW signals "10
> dB below the noise" are referring to the ability to copy a slow OOK CW 10
dB
> below the noise *in a 2000 or perhaps 500 Hz bandwidth." If the necessary
> BW of a slow OOK CW signal is 50 Hz, I don't think that that many (or
any?)
> of these operators could copy a signal that is 10 dB below the noise in
the
> necessary bandwidth of 50 Hz. Yet some of the digital systems can manage
> that, no?
>
> 73, Ed Hare, W1RFI
>
>
------------------------------
Date: Sun, 16 Jul 2000 02:37:33 GMT
From: kleing@NOSPAMqualcomm.com (Klein Gilhousen)
Subject: CW versus hi speed digital etc.
Thank you for your thoughtful responses Peter.
Your description of matched filtering is quite correct. Sometimes it
is more useful to think of it backwards from the way you have
described it. I.e., first you design a good receiver filter. Then
To be continued in digest: hd_2000_191C
Read previous mail | Read next mail
| |
