[hpsdr] Mercury

[John B. Stephensen]

It would be nice to have at least as many resources as a Xilinx XC3S400 (16
18k dual-port RAMs, 16 18x18 multipliers and 7200 logic elements). I'm
finishing off the second implementation of an OFDM modem on the XC3S400 and
it would be nice to have more hardware available for it to run on. The modem
is being done as an ARRL HSMM WG project and I'm planning on publishng the
design in QEX. Fast VHF and UHF modems should be of interest to TAPR
members.

Also, consider the Xilinx XC3S500E. Its very inexpensive, available in a
QFP, and has 20 18k RAMs, 20 multpliers and  9,300 logic elements. The free
Xilinx development tools now include the Core Generator which contains free
CORDIC and FFT implementations in addition to simpler functions.

The OFDM modem is much more resource-efficient than my first implementation
and is designed for run-time configurable carrier spacing and bandwidths so
that 64 to 1,8432 kbps data rates can be supported. The implementation is in
Verilog and includes a DDC, interpolating/decimating FIR low-pass filters,
FFT (Xilinx), phase modulator, phase demodulator (Xilinx CORDIC),
convolutional encoder and Viterbi decoder. It is meant to be used with a
12.288 Msps ADC and a 6.144 Msps dual DAC. The low sampling rates allow the
DDC, filters and Viterbi decoder to be much smaller than the Xilinx IP. With
an FPGA like the XC3S500E, a much higher sampling rate could be supported.

73,

John
KD6OZH

----- Original Message ----- 
From: "Philip Covington" <p.covington at gmail.com>
To: "John B. Stephensen" <kd6ozh at comcast.net>
Cc: <hpsdr at hpsdr.org>
Sent: Saturday, June 10, 2006 01:04 UTC
Subject: Re: [hpsdr] Mercury


> On 6/9/06, John B. Stephensen <kd6ozh at comcast.net> wrote:
> > ***** High Performance Software Defined Radio Discussion List *****
> >
> > What size FPGA is planned for the Mercury board?
> >
> > 73,
> >
> > John
> > KD6OZH
>
> Hi John,
>
> Phil H and I were just talking about this.  It is TBD...but I can say
> it won't be the little Cyclone II on the OZY.  We should have a better
> idea pretty soon.  I was doing some filter design stuff with MATLAB
> today.  I want it to be big enough to be able to do some serious DSP
> in it.
>
> 73 de Phil N8VB


 1149994811.0