Date: Sat, 01 Jan 2005 18:18:10 +0100
From: Alberto di Bene <dibene@usa.net>
To: lowfer@lwca.org, LF Mailing List <rsgb_lf_group@blacksheep.org>
Subject: [LW] Considerations about Jason and its new Turbo (?) mode...
Mime-Version: 1.0
Message-ID: <409JaaRsk0128S08.1104599890@uwdvg008.cms.usa.net>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Z-USANET-MsgId: XID320JaaRsm0391X28
Content-Transfer-Encoding: 8bit
Sender: owner-lowfer@lwca.org
Precedence: normal
Reply-To: lowfer@lwca.org

Greetings all,

 after slowly recovering from the excess drinking of New Year Midnight,
I have coded a new function in Jason, which was originated by a comment
of Bill de Carle, who said that the duration of tones maybe could be
shortened, at least when the SNR is good.
The duration was not chosen randomly, it obeys to a physics law, which says
that in order to nail down a signal frequency with an uncertainty of df (delta
f), you must observe it for a time not shorter that 1/df.
But what Bill said made me think that in conditions of good SNR you could
accept a reduced likelihood of a correct identification of the frequency,
without increasing the error rate.
So I have introduced the "Turbo" mode, which basically cuts in half the
duration of each tone, leaving all of the other parameters unchanged. 
Some simple audio tests done here (laptop in one room squeaking Jason tones
from the speakers, and PC with a microphone in a noisy environment in another
room) gave positive results, at least above a given SNR level.

So now the possible speeds are :

Slow, Turbo OFF    Throughput = 18.93 char/hour, 0.315 char/min
Slow, Turbo ON         "      = 37.86     "      0.631     "
Normal, Turbo OFF      "      = 151.4     "      2.523     "
Normal, Turbo ON       "      = 302.8     "      5.047     "
Fast,   Turbo OFF      "      = 1214.2    "     20.24      "
Fast,   Turbo ON       "      = 2422.5    "     40.37      "

The last row is interesting. There are many key-hunting "typists" that are
slower than that, so may be the Fast/ON mode could be tried as some sort of
RTTY on HF... what are the frequency stability requirements  from the rig ? 
Due to the nature of IFK, a slow drift is of no consequences, provided that
the frequency stays in the same FFT bin for the duration
of the tone. At the fastest speed, with Turbo ON, the tone duration is abt.
0.743 sec, and the bin size is abt. 0.673 Hz. This implies that, e.g. at 14.1
MHz, the frequency must not change more than 38.5 ppm in a 10-minute interval.
This does not look like as a difficult figure to achieve. Any takers ?

So, in the light of all the above, I have uploaded Jason V0.98, which you can
normally download by visiting the Web page  http://www.weaksignals.com

I have also implemented the function requested by Scott Tilley, i.e. the
possibility to load the beacon text from a file. At each repetition of the
beacon, the file is loaded again, in case it has changed.

Sorry for the long posting, and please send reports, TNX.

73  Alberto  I2PHD
P.S. When testing, the Rx station must set Turbo OFF or ON in accord with the
Tx station, as this changes a parameter also in the decoder.


________________________________________
File attachments, if any, have NOT been virus scanned...check before opening!
To subscribe, unsubscribe or get other help, see http://lwca.org/groups.htm