[TenTec] Pegasus DSP Project

[Carl Moreschi]

Mark,

Your 1.05 filter shape still won't be in the pegasus AGC loop.  This will
make the filter the same as can be found on any good add on audio DSP unit.
In order for the steep skirts to really help, the AGC loop has to go through
the filters.  Otherwise, AGC pumping on signals outside your external
passband will ruin the effect of the steep skirts.

Also, 2000 taps will cause a significant time delay and ruin any hope for
QSK.

Carl Moreschi N4PY
Franklinton, NC
----- Original Message -----
From: "Mark Erbaugh" <mark at microenh.com>
To: <tentec at contesting.com>
Sent: Monday, April 19, 2004 6:22 PM
Subject: [TenTec] Pegasus DSP Project


> Is there anyone out there who is a DSP guru who could help me learn DSP?
>
> I'm working on a project where I take the same Recv IF that is fed to the
> Pegasus ADC and feed that to a computer soundcard line input. From there,
I
> demodulate the IF signal using custom built DSP software running on a
> Windows XP computer.  I have been able to successfully demodulate AM and
SSB
> signals using some code I adapted from something I found on the web from
> JI3GAB.  However, I need to learn how to design FIR filters with and
without
> Hilbert phase shifting.  Most of the resources that I've found seem to
> assume that I will be using a commercial filter design program and that
> those features will be in that program. What I'd really like to do is
> include the filter design code in the DSP program so that the user can
> custom design a filter on-the-fly rather than having to choose from a
preset
> list.
>
> I'm currently using the Intel signal processing library (SPL). That
library
> has a routine that does a nice job of designing FIR filters with a zero
> phase shift, but there doesn't seem to be a way to make it design a filter
> with the 90 degree phase shift I need.  It there a simple technique for
> changing the phase shift of an FIR filter?
>
> I found some code on the web that will design FIR filters using the
> Parks-McClellan method, but the code seems unstable when I try to design
> more than 350 taps or so. I'd really like to try something like 2000 taps
> (that should give me some 1.05 filter skirts).
>
> Mark
>
> _______________________________________________
> TenTec mailing list
> TenTec at contesting.com
> http://lists.contesting.com/mailman/listinfo/tentec