|
Last night the EA5HVK Opera decoder and my opds
correlation detector ran side by side here on one machine, fed by the same
soundcard. Opera got the full passband (136-138.2 kHz) through SpecLab and VAC
without further audio processing, whereas the other SpecLab instance for opds
had a 160 Hz bandpass filter, noise blanker and audio AGC inserted before the
0.477 mHz FFT. The QRN background from the southeast was generally lower than
the night before.
Results for both are
in
df6nm.bplaced.net/opera/Opera_vs_opds_screenshot_130624.jpg
and a
screenshot
df6nm.bplaced.net/opera/Opera_vs_opds_screenshot_130624.jpg
Although SNR results shown by Opera were
consistently better than those for opds, Opera achieved two only decodes while
opds produced ten hits (including three from VO1NA).
I am wondering whether this difference may have
been exacerbated by the regular splatter from DCF39 which is quite strong in
central Europe. This may also have affected the SNR estimates of either program.
Maybe in this situation dedicated preprocessing might be helpful for Opera as
well?
We may perhaps speculate that possible DCF
suppression may be part of the loop advantage observed by DF2JP - Joe could you
comment on your loop's orientation?
Best 73,
Markus (DF6NM)
Sent: Sunday, June 23, 2013 1:25 PM
Subject: Re: LF: VO1NA Opera-32
I have downloaded Opera v1.4.7 from the rosmodem
website. It was easy to configure and is running smoothly in conjunction with
SpecLab and VAC-3, which is needed here to convert from 135.5 to 136.0 kHz
LO. With the waterfall disabled, average CPU usage on the slow Atom Netbook
is an easy 20% from one of two virtual CPU's (10% shown in task
manager).
Opds is running in parallel from a separate SpecLab
instance. A locally generated test signal
showed up both in Opera (-28 dB) and in opds (-36.4 dB), which means that
the first round goes to Jose ;-) I will let both
run for a while, hoping for some weak signals to compare.
Best 73,
Markus (DF6NM)
| 