Message-ID: <4BB97982.3030303@freenet.de>
Date: Mon, 05 Apr 2010 07:47:46 +0200
From: =?ISO-8859-1?Q?Wolfgang_B=FCscher?= <dl4yhf@freenet.de>
User-Agent: Thunderbird 2.0.0.24 (Windows/20100228)
MIME-Version: 1.0
To: rsgb_lf_group@blacksheep.org
References: <4BB8451B.1090207@freenet.de> <38A51B74B884D74083D7950AD0DD85E82A1BCD@File-Server-HST.hst.e-technik.tu-darmstadt.de> <E76E868251A541159202C9B06B76F1EB@JimPC> <38A51B74B884D74083D7950AD0DD85E82A1BD0@File-Server-HST.hst.e-technik.tu-darmstadt.de> <5510C66CEB9B46FDAB474C88A25031C5@JimPC>
In-Reply-To: <5510C66CEB9B46FDAB474C88A25031C5@JimPC>
Subject: Re: VLF: 4th VLF experiment by DK7FC/p
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org


Dear Jim, Stefan and the group,


Jim asked:

> I imagine Spectrum Lab must be doing the clipping in the time domain, 
> i.e. limiting the maximum amplitude of any particular sample that is 
> above the average level - perhaps Wolf could describe how this 
> function works?
>
Yes, the 'hard limiter' / clipper in each DSP box works entirely in the 
time domain (unlike the frequency-selective limiter in the FFT-based 
filter unit):
Each input sample in the time domain, X, is full-wave rectified and run 
through a lowpass. That low-pass filtered average, scaled by the dB 
value for the relative clipping level, gives the actual comparator value 
Y. In other words, Y follows the rectified input average 'very slowly'. 
Then, for each original sample X, the program just does this in pseudo-code:
if X > Y  then X = Y;
if X < -Y then X = -Y;

There is a second, absolute, threshold value for this limiter which is 
just "dB below full scale", which would do nothing more than two 
anti-parallel diodes at the input.

In contrast to a classic 'noise blanker' X is not set to zero if the 
threshold is exceeded.


Cheers,
   Wolf .