Dave,
I'm not sure if the actual frequency error is critical to the wspr system,
from the problems I experienced last year with the system, as the quality
of the received transmission in terms of signal to noise was identified
as the 'only' factor that determined the ability of the system to decode.
As the data would appear to be' array processed' with decodes extracted
from a received 2 dimensional block of data, any frequency error is not
significant, however it was discussed that the possibility existed that the
exact time interval of the symbolisation 'frames' may be truncated by a
sound card clock running 'fast'
This was proposed as an explanation to the observed situation, where a
strong constant signal, resulting in the reception of a full set of sync
frames failed to decode, where as a signal from the same station,
considerably weaker either by a power reduction or coinciding with a deep
qsb fade produced a reliable decode to the systems lower operating limit.
By off air measurement of the time frame, it could be observed that the
transmission finished some 2 seconds before the end of expected period.
switching to a usb card and re running the test, resulted in the correct
time frame and 100% decode of a 'clean' stable signal.
The Initial card M-Audio delta-4 , substitute card ADS-rdx-150,
interestingly I have used the 44khz AWE-64 isa soundcard for wspr
recently and where as long range decodes looked to be reliable, Gary g4wgt
who is some 25 miles , has noted a high percentage of failed decodes , but
not a total failure as when the delt-4 was used.
May be (we hope) later issue of wspr will have a clock check facility?
G .
--------------------------------------------------
From: "Dave Sergeant" <[email protected]>
Sent: Saturday, September 12, 2009 11:59 AM
To: <[email protected]>
Subject: LF: WOLF/WSPR Sound card issues
I have been trying to work out why I have such bad copy of WOLF and
WSPR (on HF as well as 500kHz) and others report it as 'plug and go'
and came across a page at http://wsprnet.org/drupal/node/336 which
explains the problem exactly. It seems that both WOLF and WSPR are
designed for current sound cards which have 48,000 and 96,000 clock
rates. Mine is obviously from the old stable so has 44,100 etc rates.
For information it is the sound card interface built into the
motherboard, an ASR Rock K7S41GX dating from 2003 with generic AC'97
sound. Zilch more given about its spec in the manual.
I did a test using a 40m broadcast carrier on 7355kHz, with the K2
tuned to get a 1.5kHz tone.
Argo reports it correctly at 1502Hz, pretty spot on
Wolf (default 8000Hz sampling) reports 1553Hz
WSPR (2000Hz sampling) reports 1534Hz
Obviously the interpolation works differently for the different
sampling rates of WOLF and WSPR but the effect is obvious.
Being an older card there is no way it will support the 48,000 etc
sampling rates natively, and there is certainly no switch. In Wolf I
can play around with the calibration, but there seems no way to do this
in WSPR.
At least I now know why I, and presumably others, have problems.
73 Dave G3YMC
http://www.davesergeant.com
No virus found in this incoming message.
Checked by AVG - www.avg.com
Version: 8.5.409 / Virus Database: 270.13.93/2365 - Release Date: 09/12/09
06:37:00
|
