--- Martin Ewing <martin@aa6e.net> wrote:
> I think there is a useful distinction between the
> various CPUs, and it's important for the Orion.
At least from a high level, based on various comments
here I do get the sense that quite a few don't really
grasp what the basic CPU/DSP architectural design of
the Orion is (or any other modern DSP IF radio for
that matter).
> TT doesn't tell us quite enough about the logical
> architecture of the Orion, but here's what I've
> learned so far. The "Sharc" DSP chips do
> the filtering, detection, and modulation.
>
> But the "Dragonball" processor (Mot 68000 family)
> handles the LCD, VFO control, the keyer (with
> jitter), serial I/O, and front panel
> functions. Most of the firmware problems seem to be
> in the Dragonball code.
I have noticed this too. And it sure does seem to be
the area that gets most of the discussion space here.
Whether they are genuine problems, or just
disagreements about how things should actually
work/look or not is somewhat moot, - or should be. In
reality the control processor hardware should be able
to accommodate all of them, for it to do so it needs
to be powerful, not a likely proposition as long as
the radio design uses an embedded control CPU of some
kind.
> The interesting feature (for me) is that, while the
> control processor directs the DSP operation, the
> received audio/IF data are mostly not
> available to the control processor. (Excepting the
> spectrum display) This means, among other things,
> that the LCD can't show a tuning
> indicator (for FSK or PSK), can't show demodulated
> digital modes, etc. Also we can't show signal
> statistics that might help set up the digital
> AGC.
I've become completely convinced that these kinds of
user interface and display functions do not belong in
the "radio box" itself. They are far better handled in
the world of generic, commodity PC hardware that is so
much more powerful and can be had for just few hundred
$. The Flex-Radio SDR1000 is much closer to this
model; though I still believe that that IF A to D and
the heavy DSP processing are better handled in
dedicated HW (both of which could fit on a single PCI
bus plug-in card). That said I have seen some pretty
impressive preliminary RX performance numbers for the
SDR1000 running on fairly modest PC's. It's the highly
variable nature of the sound cards needed for that
radio to function that scares me the most. Its
software is however progressing, which is very
encouraging.
Perhaps Yuri's Dream Radio will also embrace some
those same design ideas as the SDR1000, we'll just
have to wait and see I guess. Regardless I'm convinced
that the amateur radio market place is really ready
for a major shake up of the status quo for the basic
design approach of our radios, - like Ten Tec did when
they came onto the scene 35 or so years ago. Ten Tec
was early innovator with their mostly solid state gear
when most everyone else was still predominately tube
based. Those manufacturers, now gone, came up with too
little, too late to survive.
> Some of these functions are available in the
> IC-7800, and hopefully we will see them "trickle
> down" to us proletarian Orion users.
> (no flames, please!)
For what its worth I don't think that the 7800 is that
much more capable in this area either. Yes it does 'in
the box' PSK31 and RTTY, but do users really have that
much control over how it looks or works? Or access to
the raw data stream? From what I can tell, no. That
radio too was designed from the traditional radio
design (look and feel) perspective first, and then
generic PC (or computing) second, - it's just more
legacy thinking. Until that fundamental radio design
thought process is reversed we will see very little
really cool user interface concepts from anybody.
> I'd say that control processors, high-level signal
> processing (codecs), and the user interface is
> where the real innovation will be in the
> future, because RF and basic DSP are becoming
> commodities. Anyone agree with that?
Absolutely. Solid state RF has been largely a
commodity for years, it's just that companies have
been far too willing to trade away performance for a
few dollars in manufacturing component cost savings.
So we have had years of mediocre to poor performing
radios, even though the technology and components have
been around (and relatively cheap) for years. Just
look at what Elecraft did with a pretty generic
collection of through-hole semiconductors. There is
nothing uncommon or exotic in that radio at all.
Clearly DSP hardware is rapidly becoming commodity
too, it's the somewhat specialized code that we need
as radio amateurs that will differentiate "our" DSP
implementations from others. And it will the amateur
gear manufacturer that writes the best and widest
ranging code that will become the leading ham gear
manufacturing company in the future. To that end I
have been very impressed with what Ten Tec does
achieve with the level of DSP hardware power that they
do choose for their designs. They really do get every
last bit of performance out of those DSP processors,
it's just too bad they don't start with more raw DSP
horsepower to begin with.
Duane
N9DG
__________________________________
Do you Yahoo!?
Yahoo! Mail is new and improved - Check it out!
http://promotions.yahoo.com/new_mail
_______________________________________________
TenTec mailing list
TenTec@contesting.com
http://lists.contesting.com/mailman/listinfo/tentec
|