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

[linrad] Linrad-01.26



Hi All,

The usage of 16 bit integers in the noise blanker makes the
Linrad setup rather critical in order to not loose dynamic 
range. It has been necessary to use 16 bit because of CPU
load but on modern computers (PentiumIV) the CPU is fast
enough to allow all processing to be done with floating point
arithmetics.

If you select fft versions = 0 (floating point) for the 
backwards fft1 and for fft2, the noise blanker will use
floating point arithmetics.

If you have selected the parameters (gain levels) correctly
there is no practical difference in performance when you listen
to weak signals. The blanker will work equally well with
16 bit and floating point. When you listen to strong signals
there is a difference however. For details, look here:
http://antennspecialisten.se/~sm5bsz/linuxdsp/blanker/fp/fpblank.htm

Linrad-01.26 contains prototype routines in users.c to make
it easy to make user defined routines (in users_hwaredriver.c)
to open windows and control hardware with mouse and keyboard.

Some errors in the speech processor simulation are corrected.
There could still be errors in these simulations....

73

Leif / SM5BSZ
LINRADDARNIL
]