Hi all,

My last message seemed to be modestly well-received, and since then I've
received / seen a number of questions regarding decoding JT65 signals, so
here's my stab at addressing those.  Remember that I'm a "newbie" myself: I
haven't even been playing around for WSJT for a whole week!  Certainly don't
hesitate to correct me if I say anything wrong, or to ask further questions.

Anatomy of a JT65A signal
Although JT65A is always a MFSK mode, there are two fundamental types of
JT65A signals: the "full text" ones used to send CQs, exchange callsigns and
signal reports, or any arbitrary text, and the "shorthand" signals used to
send RO, RRR, and 73.  Both kinds of messages have a "sync" tone and other
tones -- in the case of full text messages, many other tones, and in the
case of shorthand messages, just one other tone.  However, the sync tone is
always the tone lowest in frequency, and in my experience with HF operation,
almost always between 1200 and 1300Hz (assuming you are tuned to the same
frequency as the transmitting station).  (For the sake of completeness, I
should also note here that only one tone is transmitted at any given time,
and that transitions between tones are done in a phase-continuous manner.)

Note on full text messages: arbitrary text can be, at most, 13 characters.
For those of you who have noticed that e.g. "K3UK N2YYZ FN12   OOO" is quite
a bit longer than 13 characters, JT65 has a special compression scheme for
common message components, like "CQ", "QRZ", "OOO", and callsigns, which it
uses in the standard exchanges.

Decoding basics
Start up WSJT.  Make sure "Freeze" is unchecked.  Examine the SpecJT
window.  You should see a long green bar at the top of your passband.  This
green bar represents the frequency range that WSJT will search for signals
when it attempts a decode.  That's right -- it attempts to decode anything
it can find in that range (as long as the sync tone falls within the green
bar, it should "just work").  This is a feature designed to make
compensating for EME Doppler easy.  If there's a signal that you can hear,
or even that you cannot hear, WSJT should decode it.  However, you will
probably need to increase your "Recording Settings" "line in"/"microphone"
slider to more than you use for other digital modes.  The WSJT documentation
states that the "RX noise" figure should be around 0dB while you are
receiving just noise.  I usually don't pump things up this high, but I do
make sure it's above -10dB (the associated status indication turns red if
this condition isn't met).  If you can't use your soundcard's mixer to get
things this high, the rightmost slider at the bottom of the SpecJT window is
"Digital gain" and can increase audio levels more.

(Aside: here's a point where I would appreciate feedback from others who
have been successful.  How high do you usually turn things up?  Am I wrong
here?)

Also remember that you will not decode what you just heard until the full
minute has passed (I said 52 seconds in my last message; a more accurate
figure is something like 55 seconds).  This is just due to the nature of the
protocol: each JT65A transmission is sent as a single FEC'd block, which is
one of the things that gives it the weak-signal abilities for which it is
known.

More "advanced" decoding
There are a lot of knobs and buttons to play with on the WSJT decoder.  I'll
start with the more important ones...

Clicking in the SpecJT window (or on the red line plot in the main window)
will set the clicked frequency as the frequency of the sync tone.  I'll
reiterate that -- when you click on something in the waterfall, it should
always be the leftmost sync tone!  Once you've given WSJT a frequency value
this way, you can check "Freeze", which locks the search for the sync tone
down to within "Tol" Hz of the frequency you selected.  You can vary Tol by
left or right-clicking on it.  Watch the length of the green bar in the
SpecJT window change as you do this.

(Cool tip: in the cases where multiple stations have both transmitted in the
same timeslot (with a slight frequency difference), I have often been able
to decode both transmissions by setting Tol to 10, clicking one sync tone,
decoding, and then clicking the other and decoding.)

Once you've tuned a station, you should leave "Freeze" on.  It gives you a
little more sensitivity, (I think) decreases decode times slightly, and also
makes the shorthand signals (RO/RRR/73) decode without question marks.  Once
you've clicked on the sync tone of a station, SpecJT also shows you "helper"
red ticks on the frequency scale, that indicate where tones for RO, RRR, and
73 will be seen (from left to right).

"NB" is what you'd expect -- some sort of noise blanking.  I've left it on,
but I'm not sure if it's actually necessary.  "Zap" somehow attempts to
"zap" away receiver birdies.  I usually keep this off.  Enabling either
probably reduces sensitivity slightly.

There's another feature of WSJT that is useful for very weak signals -- so
weak that they can't be decoded in one RX period.  It has two slots (one for
the even slot, one for the odd -- I think; not sure on this) where it stores
averages of past signals.  These are displayed in the two-line textbox below
the main textbox.  Thus it is possible to monitor a "dead" channel for a few
RX periods, and suddenly see a signal "pop out of the noise" once you've
done enough averaging.  To use this effectively, though, you must
occasionally clear the averaging buckets and allow new data to accumulate.
That's what the "Clear Avg" button is for.  (I haven't actually decoded an
otherwise-unreadable signal with this functionality myself, but maybe one
day.)


I hope this helps, and that it explains more than it confuses.

best & 73,
-chris N2YYZ

Reply via email to