Re: LF: WSPR or QRSS: which is better?

[Andy Talbot <andy.g4jnt@gmail.com>]

I know what you're referring to.  In cases of impulsive noise, there is a finite probability of something eventually  getting through the decoder and being flagged as valid.  The very nature of heavy source coding, means that the resulting random output will look like a valid callsign.   However, WE then apply the next level of error detection, by knowing the combination must be rubbish.

 

'jnt   [and there is another example of source coding]

 

On 24 August 2011 17:47, Graham <g8fzk@g8fzk.fsnet.co.uk> wrote:

>>  and gives absolutely guaranteed error free decoding, or nothing at all.  <<

 

Are you sure ? 

 

G..

From: Andy Talbot

Sent: Wednesday, August 24, 2011 5:12 PM

To: rsgb_lf_group@blacksheep.org

Subject: Re: LF: WSPR or QRSS: which is better?

WSPR works in a 1.46Hz signal bandwidth and because of its very high level of error correction and soft-decision decoding, means that it will work at a S/N of about 3dB in this bandwidth, and sometimes a bit lower still  (Normally, FSK with no correction at all needs about 10 - 12dB S/N for near error-free performance)

 

QRSS has to show about 6 - 10dB in its signal bandwidth to be able to discern fully what is sent, although a slightly lower S/N may be useable when you 'know' what you should be receiving.  (A form of forward error correction is now in use here as well perhaps :-)  So lets say 5dB S/N is a working value..

 

So take 3dB in 1.46Hz as a starting point and derive the bandwidth for QRSS needed to get 5dB S/N with the same signal.  This will have to be narrower to get a 2dB higher S/N and works out as 1.46 / 10^(2/10) = 0.92Hz

 

So QRSS used with a 0.9Hz bandwidth - which I think means about a 2 - 3s dot period ought to be decoded at the same S/N as a WSPR signal.   Which is probably the info you wanted.

 

But now compare source coding efficiencies.   WSPR fits a callsign, locator and power level into a 110 second transmission - and gives absolutely guaranteed error free decoding, or nothing at all.  About 12 characters in actuality, but that is being a bit unfair as the coding forces certain callsign and locator formatting.   So in all probablility, more like 7 or 8 effective characters (I'm being a bit empirical here)

 

Assuming standard QRSS - not DFCW - , which if like standard Morse, then 5 characters takes about 50 dot symbols to send (12WPM = 60 chars in 1 minute, = 1 char / second, or about 10 dot periods / second.    Dot speed = WPM / 1.2)   If we have 2s dots, that is 5 characters can be sent in the time for a WSPR transmission.

 

So as a quick estimate, WSPR wins by roughly 2dB in S/N terms for a given dot period / noise bandwidth.  And at similar S/N values, WSPR is about 1.5 times faster

Andy

www.g4jnt.com

 

 

 

 

On 24 August 2011 16:42, Roger Lapthorn <rogerlapthorn@gmail.com> wrote:

A question for the coding experts here: WSPR is an excellent weak signal beaconing mode, but at what QRSS speed is QRSS "better" ?

73s
Roger G3XBM

--
http://g3xbm-qrp.blogspot.com/
http://www.g3xbm.co.uk
http://www.youtube.com/user/g3xbm
https://sites.google.com/site/sub9khz/