Message-ID: <3C43FE2C.49EE902A@usa.net>
Date: Tue, 15 Jan 2002 11:02:20 +0100
From: "Alberto di Bene" <dibene@usa.net>
MIME-Version: 1.0
To: rsgb_lf_group@blacksheep.org
Subject: LF: Re: Introducing Jason - and asking for cooperation
References: <3C4301BD.878E49BF@usa.net> <006b01c19d34$ad33d500$979a17d2@steve> <001801c19d4c$78f9d9c0$1700a8c0@home>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 8bit
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

Dave Pick wrote:

> I just get occasional wrong characters like / or ,
> [...}
> Operational questions:
> How do you clear garbage from the RX screen?
> Do you need to start the TX after the RX is running or will it pick up a
> signal half-way through a long message?
> Is there a way of transmitting a long start tone (seems to be centre
> frequency) for alignment purposes?
> How long should it take to display a character on the screen?
>

Hi Dave,
   thanks for the tests and the report.
Well, I can tell you how I tested Jason. Maybe you could initially do a similar test,
to rule out any radio-related issue.
I had it generate the audio tones, running on a laptop, then I brought the laptop
upstairs, with the speaker volume adjusted for a low setting.
Then I started Jason in Rx mode on the PC in the shack, dowstairs, with input
coming from a microphone, which picked up the tones generated by the laptop.
In the shack the sound was so low that I was unable to hear it, but I had perfect
copy. I even switched on a radio tuned on an empty portion of the 144 MHz, SSB,
to have some sort of white audio noise in the room. Still perfect copy.

Please try the above, which do not require any radio (but needs two PCs), to see
what happens. Thanks.

About what you ask:
- Garbage : there is not a clear screen function. Hey, this is just an alpha version !  :-)
- There is no need to start the Tx and the Rx in a synchronized way. Of course it will
   take one or two characters to get in synch.
- If you put the program in Tx mode, without any text present on the lower input pane,
   it will transmit an idle sequence composed of the lower and the higher frequencies
   of the range. Their difference in frequency is outside the allowed range, so no
   decoding will occur, but you can use the traces to place them exactly between the
   two yellow lines.
- Each character takes about 24 seconds to show on the screen, for a throughput
   of roughly 2.5 characters per minute.

About tuning : the white lines on the spectrogram window related to the received
signal must entirely be within the two yellow lines. If the tuning steps of your Rx
are too coarse for this (many Rx have 10Hz as the minimum step, and this can not
be fine enough), you can position the receiving window with the mouse.
Just left-click with the mouse on the approximate center of the various white lines
of the signal. The receiving window will be moved so that the point you clicked on will
be positioned exactly half way between the two yellow lines.
Probably you will have to use this technique when testing with audio only, as described
above, due to the differences between sampling rate of different sound cards.

One last thing, about the CPU speed needed. I made all the tests with a 750 MHz Athlon,
and I observed that the CPU utilization was around 40 - 50 %.
To gather a good statistics of the signal, in Jason I do a 131072 points FFT 10 times per
second (roughly). Maybe this is too much for a 200 MHz CPU.  If this will prove to
be the case, it is a matter of changing some constants to slow it down.

Oh, still one final point  : I have a report that working at 800 x 600 resolution the program
displays incorrectly its screen. Probably this can be fixed.

73  Alberto  I2PHD