[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [linrad] users_extra

Hi Roger and all,

> I was going to send you an email with this question, but figured I'd ask  
> it here.  With my new install of 01.22 [I know I am 2 versions behind; I  
> will fix that this week if I have time] on my PIII [the one with 
> the noisy  
> power supply, which will be replaced as soon as the new one arrives] I  
> find that if I click on the W3CCX beacon, which is at 60 dB signal  
> strength [well above the 'weak signal realm'] the signal 'blows up' and  
> widens markedly in frequency on the main waterfall.  It is still 
> copyable,  
> but very wide on the waterfall.  If I click away from it, it goes 
> back to  
> its previous narrower width.  I tried playing with the blue and red bars  
> at the left of the baseband graph but this didn't fix things.
This is a bug - or rather it is something I did not care about before;-)

This is a consequence of using 16 bits in the processing. Linrad has a
selective AGC which is controlled by the blue and red bars but when you
click on a signal you disable the operation of the bars over the frequency
range you have specified with the baseband filter. Some frequencies may 
still be forced to become red because they can not be fitted in 16 bits 
otherwise and this will destroy the spectrum of the selected frequency range.

In case you want to play with strong signals, deselect the second fft
to avoid the 16 bit complications completely. You can also reduce the digital
signal levels with the "First FFT amplitude" parameter. Check the digital
signal levels so you do not reduce it too far because you could introduce
digital noise.

You might make the fft1 and fft2 sizes equal. You loose one bit out of the 16
ones available for each factor of two that fft2 is bigger than fft1.

You can also increase "Sellim maxlevel" to leave less margin for saturation 
of the 16 bits when computing the second FFT.

Use "A" to see what amplitude margins you have and read about "setting the 
digital signal levels"

> I haven't used Linrad much since the ARRL EME contest last year, and my  
> memory is a bit dim on its nuances.  I recall that this may be expected  
> behaviour, and a result of using the tool in a way that is not intended,  
> and which doesn't make much sense [using it for very strong 
> signals in the  weak cw mode].  Nevertheless, because I will be using it 
> on HF where there are much stronger signals mixed in with weaker ones 
> once the HF module is  
> available, and because I'd like to be able to switch back and forth  
> between the weak signal portion of the band and the beacon without  
> changing modes, I wonder what I can do to change this behaviour.  

Besides disabling the second FFT (and loosing the noise blanker) you can
not eliminate this problem. The things listed above may help to some
extent but there is a need for some changes in sellim.c to handle this
problem. Another solution would be to rewrite the noise blanker routines
so they could use floating point arithmetics on fast computers.
I will look into it some day, but in the meantime all you can do is 
the above (I think).

> I assume  
> it is due to the way I have some parameters set, but which one[s] to  
> change isn't immediately obvious to me.  So I figured I'd ask the 
> question  
> which might be apropos to others, as well, rather than just fiddling for  
> an afternoon at the computer. ;)
Mainly it is because nobody complained before;-)

> I am using the Delta 44 for input and a SoundBlaster for output 
> and I have  
> all the FFT's enabled.  It doesn't matter whether or not I have 
> the AGC on as regards this behaviour.

No. The problem is that the function that prevents saturation causes
rapid AM modulation on a very strong signal in case you select it.
It has not been like this always, but it seems I introduced it while 
fixing something else long ago;-)


Leif / SM5BSZ