[Fwd: Re: de W5UN MAP65 Question]
Joe Taylor
the message I sent to him yesterday. Others may have
wondered about some of the things about MAP65 that I
explained to Dave.
-- 73, Joe, K1JT
-------- Original Message --------
Subject: Re: de W5UN MAP65 Question
Date: Wed, 29 Apr 2009 10:52:09 -0400
From: Joe Taylor <j...@princeton.edu>
To: W5UN <w5...@wt.net>
Hi Dave,
I'm happy to hear that you have MAP65 working well.
> ... A couple more
> questions: What does the number in the right most column mean?
Hmmm, I guess you must mean the rightmost column of lines in
the main decoded text window. You are an experienced WSJT
user, so I guess you know what it means there, right? It's
the same in MAP65 -- a "goodness of fit" or quality
indication from the Deep Search decoder -- except that in
MAP65 the number is not capped at 10. For example:
139 418 144 0745 3.3 -15 CQ EA6VQ JM19 1 447
The "447" is a number (on a linear power scale) indicating
an extremely good match to the message "CQ EA6VQ JM19".
The DS Quality number is used by the "Find Delta Phi"
procedure that you can trigger from the Setup->Options menu.
This option causes the decoder to be run multiple times at
the selected QSO Frequency, with phase shifts of 0, 30, 60,
..., 360 degrees introduced between the X and Y polarization
channels.
For the same data that produced the decoded line above, the
following will appear in the decoded text window when "Find
Delta Phi" is invoked:
139 418 144 0745 3.3 -15 CQ EA6VQ JM19 1 447 0
139 418 165 0745 3.3 -16 CQ EA6VQ JM19 1 737 30
139 418 32 0745 3.3 -16 CQ EA6VQ JM19 1 368 60
139 418 38 0745 3.3 -15 CQ EA6VQ JM19 1 260 90
139 418 39 0745 3.3 -15 CQ EA6VQ JM19 1 242 120
139 418 39 0745 3.3 -15 CQ EA6VQ JM19 1 249 150
139 418 36 0745 3.3 -15 CQ EA6VQ JM19 1 297 180
139 418 15 0745 3.3 -16 CQ EA6VQ JM19 1 704 210
139 418 148 0745 3.3 -16 CQ EA6VQ JM19 1 498 240
139 418 142 0745 3.3 -15 CQ EA6VQ JM19 1 421 270
139 418 141 0745 3.3 -15 CQ EA6VQ JM19 1 407 300
139 418 141 0745 3.3 -15 CQ EA6VQ JM19 1 411 330
139 418 144 0745 3.3 -15 CQ EA6VQ JM19 1 447 360
Best-fit Dphi = -55 deg
The third number at the end of these lines is the phase
offset in degrees. The top figure in the attached graph is
a plot of the final two columns: DS Quality vs Phase Offset.
The data for this test were recorded when I had
intentionally introduced a 45 degree offset (1/8 wavelength
of extra coax) in one of the Rx feedlines, so the peak
response should appear at about 45 degrees (or 225 degrees,
since polarization is "sign independent"). As you can see,
the program determines where the peak occurs and recommends
that for optimum performance a phase adjustment of about -55
degrees should be made to "phase up" the two polarization
channels.
> I seem to be decoding most of the time, but I rarely get a decode at
> less than -23 db, and never a decode less than -26 db. Any ideas about
> this?
Sounds pretty much OK to me. The attached file
all65_nov.txt is a copy of MAP65's output file ALL65.TXT
from the November weekend of last year's ARRL EME contest.
It includes nearly 10,000 decoded messages. The lower
figure in the attached graph shows how these decodes are
distributed by S/N. As you note, most of them have S/N -24
dB or above.
At the selected QSO frequency -- those decodes appearing in
the main screen's decoded text window -- performance of the
MAP65 decoder is essentially equivalent to the WSJT decoder.
There you will see decodes down to -26, -27, -28, -29 ...
dB, the smallest ones almost always using the Deep Search
decoder.
The difference in performance between the wide-band decoder
(the "Messages" and "Band Map" windows) and the narrow-band
one at the QSO frequency is intentional. My aim was to
produce a good balance between the best possible sensitivity
and freedom from excessive garbage decodes such as "/1NVHHF
AFTD4" (one that I found in the attached file
all65_nov.txt). You will probably see a few garbage
decodes, but there should not be many.
> I have WSJT running on a seperate computer fed by SDR-5000 and a
> transverter. It sees horizontal only at this time. Today, with Europe at
> vertical all day long, I rarely got a decode on WSJT, although I could
> see the signal on the waterfall, and it looked adequate for decode. Some
> decodes on MAP65 were possible where I saw no hint of a signal on WSJT's
> waterfall. A real time waterfall on MAP65 would be a welcome addition
> because often, one can "see" the report before decode.
I do real-time "decoding by eye" all the time, using the
Linrad baseband waterfall. For this reason I haven't
considered it a high priority to implement a real-time
waterfall in MAP65. Of course this could be done, and maybe
I'll get to it one of these days... or maybe not.
You probably still need to do your Dphi adjustment, right?
Or have you done that already? You may also need to make a
sign correction; without that you will not be able to
accurately predict which Tx polarization to use when calling
a station, after the program has measured the polarization
of his received signal.
With your approval, I will post a copy of this message to
the Linrad reflector. It's best to have such Q&A as this on
a reflector; otherwise I spend far too much time with
redundant answers to repeated questions.
-- 73, Joe, K1JT