[hpsdr] Blackfin & SDR

[Chris Albertson]

--- Philip Covington <p.covington at gmail.com> wrote:

> ***** High Performance Software Defined Radio Discussion List *****
> 
> On 8/18/07, Christophe DONZELOT <hb9tln at hb9ww.ampr.org> wrote:
> > ***** High Performance Software Defined Radio Discussion List *****
> >
> > >
> > > Someone complained that 16-bits is not enough dynamic range but
> we
> > > > should all remember that a 16-bit machine can handle numbers of
> any
> > > > size just like we can multiply large numbers using pencil and
> paper one
> > > > digit at a time 16-bit computers can process 64 or 32 bit
> numbers 16
> > > > bits at a time.
> > > >
> > >
> > >
> > > Yeah, yeah, yeah... I can do 32 bit DSP with a 4 bit processor...
> > > that's not the point.  I am not convinced that the Blackfin is
> going
> > > to be fast enough for use at a reasonable IF because of its 16
> bit DSP
> > > engine even running it at 400 or 600 MHz.
> > >
> > > What is the IF into the ADC in the FT-450?

If looks like the last (3rd) IF is about 15Khz of bandwidth.   

Lets go back a few weeks.  I pointed out the Blackfin only to serve
as a lower bound on what would work as a processor.

If I were to choose a processor I'd look at my expected production
run in numbers of units shipped.  It I were Yaesu and expected to
ship thousands of units I worry about the cost of each little part
because I can't afford to pay $100 for a processor in radio that will
retail for $850  For that retail price the cost to build is going to
have to be about $425

But if I'm building just one radio the effort and time required to
write the software completely over shadows the cost of even a
high end processor and typically the bigger the processor the lazier I
can be as a programmer.  I'll gladly buy a $200 processor if
it cuts weeks off my development time.

If I were designing a kit that wold sell 100 copies I might compromise.
I'd still do the design using the biggest processor I cold find but
then when I'm done I'd down size it to the least expensive one
that does the job while leaving some room for growth   For this kind
of work you want to select
a processor family that has a wide range:  Cheap low end parts
and high end parts that are all software compatible.  

I think Yaesu did the right thing.  They had a parts budget likely
around $300.   But they could afford to send a lot of up front
on enginerring to get the unit cost down.  I think this is what's
different about professional designs vs. amatuer designs.  We
really do not want to invest tens of man years of engineering to
make one radio.


Chris Albertson
  Home:   310-376-1029  chrisalbertson90278 at yahoo.com
  Office: 310-336-5189  Christopher.J.Albertson at aero.org
  KG6OMK


      ____________________________________________________________________________________
Shape Yahoo! in your own image.  Join our Network Research Panel today!   http://surveylink.yahoo.com/gmrs/yahoo_panel_invite.asp?a=7 



 1187452840.0