--- Bill Tippett <btippett@alum.mit.edu> wrote:
> KC2G wrote:
> >On the Orion: do the two processors parallel or
> co-process, or do they
> have discrete functions?
>
> It's my understanding that they are
> parallel, but I
> don't know more about how resources are allocated.
> Much of
> this 32-bit DSP processor fanfare is academic BS
> anyway. The
> "DSP system" is limited by the 24-bit resolution of
> the ADC
> and DAC. It makes no difference if the DSP itself
> processes
> 32 bits or 64 bits, since the "DSP system" dynamic
> range is
> limited by the ADC/DAC, not the DSP processor
> itself.
I have wondered about this myself, since the Analog
Devices SHARC processors can be used in parallel. The
advantage wouldn't be for better dynamic range as
already pointed out but would instead be for
completing the computations of two 24-bit numbers in
fewer clock cycles, important for QSK operations and
to keep the latency of the signals journey through the
radio to a minimum. So if they are parallel then the
signal processing of the two RX's must be done in an
interleave fashion. What the pro's and con's of
separate 32 bit processors running each RX is vs.
paralleled and then interleaving (multitasking) the
computations is beyond my expertise, but is fun to
ponder.
> The unique beauty of Orion is its strong
> analog signal
> path (ham band-limited with narrow roofing filters
> at 9 MHz)
> and excellent phase noise. Without these, it would
> be just
> another DSP radio, most of which have the same
> 24-bit ADC/DAC
> and 32-bit DSP processors (IC-PRO2, IC-7800,
> Jupiter, etc.)
> Actually its sub-RX is one of those too (i.e. a
> Jupiter)!
The Orion sub receiver is NOT just a Pegasus/Jupiter,
RX it only shares the choice of IF frequencies in
common with the Peg/Jup. I believe that it uses a 24
bit A to D and the 32 bit Sharc DSP, these two items
alone should allow for better performance. The
Pegasus/Jupiter are built around a 16 bit A to D and
16 bit DSP. In fact the TT literature describes the
sub RX DSP as being identical to the main RX in the
Orion. But yet the claimed specs are quite different,
this just underscores the importance of having good
analog circuitry ahead of all the digital stuff.
And it further validates the reality that a radio can
only hope to preserve the S/N ratio and quality of the
incoming signal, once it is degraded by the various
RF/IF stages in the radio no amount of DSP magic can
truly bring it back. So it should be no mystery of why
Icom chose dual conversion instead of quad like they
have in the past. It is equally true that TT's
decision to not up convert is a choice to battle
similar underlying design problem but from a slightly
different angle, they seek to limit phase noise by not
having any VHF frequencies in the IF chain. Icom on
the other hand seeks to minimize the distortions
introduced by multiple mixer stages.
Duane
N9DG
__________________________________
Do you Yahoo!?
The New Yahoo! Search - Faster. Easier. Bingo.
http://search.yahoo.com
|