[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Fw: Re: Linrad + SDR-IQ. Please help
- Subject: Fw: Re: Linrad + SDR-IQ. Please help
- From: Leif Asbrink <sm5bsz.com; leif@xxxxxxxxxxxxxxxx>
- Date: Sun, 20 Jan 2008 18:54:17 +0100
Below the line (snip) is the same reply that I sent to moon-net.
I post it here for the archieves.
The ssb mode parameters (par_ssb) that you can change by
pressing X, then P (or by editing the file) should be
uncritical as long as you do not need a good dynamic range.
It would however be a good idea to set the baseband
storage time to its minimum in order to not use too much
memory. In Linrad02-35 about a dozen arrays are stored
and in SSB mode they take up a consideral amount of
physical memory if they all are 200 seconds.
It is not clear to me what you want to achieve. In case you
want to see all the JT65 signals on the main waterfall
you should set the waterfall MUCH slower. Now each
line is the average of only three transforms, but since
the transforms overlap you actually use information
corresponding to 1.5 transforms.
The bin bandwidth is too narrow for every bin to be displayed
and therefore Linrad displays the strongest one out of all
those in the interval corresponding to a single pixel.
When there is noise only, the displayed point will be something
like 3 dB above the RMS noise when your averaging is only 3.
By averaging a large number of times the S/N loss will be much
smaller than the 3 dB you now have.
First of all however, you should set the fft2 bin bandwidth
to about 5 Hz to maximize the S/N for a JT65 sync tone.
That will make your waterfall slower, with a new transform
10 times each second. Choose to average 25 to 50 times for
something like 20 to 30 lines per minute. You should then
see everything that possibly can be decoded. (You might
also see a surprisingly large number of spurs, but they
do not show the characteristic 1 minute on/off pattern
You will notice that the noise floor becomes much less granular
and you could benefit from more waterfall (colour) gain.
Maybe up to 10. That in turn will make calibration necessary
because with high gain you must have a flat noise floor.
Leif / SM5BSZ
------------------------------ snip -------------------------------
> The results with Linrad and the SDR-IQ were absolutely disappointing, but
> I'm sure that it must be because I have wrongly configured something, so I
> will appreciate your help.
> I have noticed three major problems:
> 1.- The signal repeatedly seems to drop for some seconds and then return to
> the normal level. I looks like if someone was switching on and off the
> preamplifier all the time. This does not happen with SpectraVue or WinRad,
> so I don't think it's a problem of the SDR-IQ. You can see a screenshot of
> this effect at http://www.vhfdx.net/temp/lines_linrad.jpg
Your noise blanker setting is extremely aggressive. The dumb blanker, yellow,
is 6mm above the noise floor on my screen. The red line which marks the 20 dB
point is at 27 mm so you are running with the blanker threshold 3.7 dB
above the noise floor. The yellow number in the upper left corner of
the high resolution graph (1 on your image) shows the percentage of time when
the signal is gated out. During the intervals where the signal has dropped,
the blanker level is presumably a little lower, and then the percentage of
signal removed would grow.
For weak signal work on VHF you should click on the yellow A to go to manual
mode. In auto mode, Linrad tries to keep the threshold at a constant height
above the noise floor. The noise floor is however the noise of the signal
output from the blanker so when the blanker operates on noise only,
the noise floor itself is reduced causing the level to stay 3.7 dB above
a level which is actually below the noise floor.
In manual mode you can set low blanker thresholds. Normally there is noo need,
but it could be helpful during static rain.
I suggest you set the yellow line 10 dB above the noise floor in manual
mode to start with. You can try to optimize it later, it will make a difference
only if you have very small pulses that occur extremely often.
> 2.- Signals are drifting up. In the baseband waterfall I can see how all
> signals I receive are always drifting up. The same signals received with the
> TS-790 were absolutely stable.
This must be due to the SSB-Electronics 144 MHz downconverter. Try
to listen on 28 MHz with the TS-790.
> 3.- Signals in the middle of the receiver bandwidth are noticeably stronger
> that near the edges. You can also see this effect in the screenshot at
> I used M_CIC2 , M_CIC5 , M_RCF , OL_RCF .
The parameters you have selected can provide the best possible
flat bandwidth in Linrad, but only if you perform the amplitude
calibration with a pulse generator. The reason for the non-flat
response is that decimating with only 2 in the RCF filter will
use a very wide portion of the output from CIC5 which is not
flat over such a wide range. The digital filter that Linrad
will add when it is calibrated will correct for this.
In case you do not want to do the calibration, then I suggest
you change to M_CIC2 , M_CIC5 , M_RCF , OL_RCF .
That will not give a quite flat noise floor, but it will be
less curved than the one you have now.
For calibration you need a square wave at about 100 Hz - but
not exactly (to avoid it to be phase stable to 50Hz hum)
You received this message because you are subscribed to the Google Groups "Linrad" group.
To post to this group, send email to linrad@xxxxxxxxxxxxxxxx
To unsubscribe from this group, send email to linrad-unsubscribe@xxxxxxxxxxxxxxxx
For more options, visit this group at http://groups.google.com/group/linrad