Message-ID: <4AA22EDB.2060704@sighthound.demon.co.uk>
Date: Sat, 05 Sep 2009 09:26:51 +0000
From: John P-G <gm4slv@sighthound.demon.co.uk>
User-Agent: Thunderbird 2.0.0.19 (Windows/20081209)
MIME-Version: 1.0
To: rsgb_lf_group@blacksheep.org
References: <8E070AD22619484C9E2D22046F45C521@JimPC> <4AA224DE.17607.C34F15@dave.davesergeant.com>
In-Reply-To: <4AA224DE.17607.C34F15@dave.davesergeant.com>
Subject: Re: LF: Re: WSPR beacon
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org


Dave, LF,


Dave Sergeant wrote:

> 
> Well prompted by all the mails on here I have just installed the WSPR 
> software and had a listen to your signals this morning.
> 
> I noticed a couple of things which might explain why I was not getting 
> any decodes:
> 
> 1. On switch on of my K2 and Datong UC1 which I use for 500kHz receive 
> there is a noticeable drift, initially as high as 10Hz in the 2 minute 
> sampling period but eventually dropping to within 1Hz. I could not find 
> a spec as to how stable the receiver needs to be for WSPR. I believe 
> most of this drift is the 116MHz downconverter crystal in the UC1, 
> which is a big limitation to using it as a serious 500kHz receiver.

The frequency needs to be reasonably stable - the WSPR software will 
report the frequency shift seen during a reception period as "Drift" and 
I've seen a few reports of 1 or 2 Hz. This implies that it will work 
with a 1-2Hz drift across the 2 minute period. I'm not sure what the 
absolute limit is though.

> 2. Although I synchronised my computer clock to the time standards, and 
> it seems just a fraction of a second different from my 60kHz clock, 
> your transmissions seemed to be starting a second or two early, before 
> my WSPR started receiving, and finishing at around the 1.52 minutes 
> point. Is your clock out, or is this how it is supposed to work?

My PC is running NTP and is (hopefully fairly well locked to NTP time) 
and WSPR reports  "DT" (Delta Time) of +/- 0.1 Seconds on Jim's signals, 
so I guess that he and I are in agreement of what the correct time is, 
and he's getting lots of spots reported from other stations, so I don't 
think his timing is an issue. His trace on the waterfall in WSPR sits 
nicely in the 2-minute window.

I've seen successful decodes with DTs of up to 2 seconds.

The SNR/clock rate issue (see below) might muddy the waters  on absolute 
timing accuracy though?

> 3. You are of course s9+ at this range, and show as bright red on 
> WSPR's spectrogram. Hardly 'weak signal'. Does the software cope with 
> strong signals or does it overload?

This might be significant though.

I've seen may times that very strong signals fail to decode whereas much 
weaker ones from the same station (eg during QSB or after reduction in 
TX power) decode fine. There is one theory that it's to do with the 
accuracy of the soundcard clocks. With weakish signals some of the 
transmission is lost in the noise and the FEC will kick in and allow a 
decode by "filling in the gaps". With very strong signals no data is 
missed, and due to clock rate differnces the decode fails because the 
received bits are outide the tolerance of the decoder, timewise (A very 
poor description, sorry - Andy JNT will explain better).


This could imply that either yours or Jim's soundcard is not up to the 
job. Since Jim has had spots reported from close stations with SNRs of 
+9dB I guess this implies Jim's timing must be okay. I've seen this 
effect kick in when signals start to go into the positive SNR region.


> The only decode I got was for 'C1N/NU0RWS' in one of your non-transmit 
> periods.

That's a false decode and might indicate that you have a noise problem. 
I've seen similar gobbledegook decodes from people with high noise levels.

HTH

I'm receiving Jim consistently, in daylight, at -21dB SNR at 953km 
distance - on a Wellbrook ALA1530 loop.


Cheers,

John