Hi All,
Just a quick reminder to submit your NAQP SSB logs directly to Steve,
N4TQO. The address is: Steve Merchant N4TQO, 1795 Cravens Lane, Carpenteria,
CA. 93013. I'm sure Steve would be happy to take your logs via E-Mail as
well. Steve's E-Mail address is merchant@crl.com. Contact Steve if you need
further details on electronic log submissions to him. Thanks!
73, Bob W9NQ
PS, I also take NAQP CW logs via E-Mail. If you're interested in doing so,
let me know.
>From H. Ward Silver" <hwardsil@seattleu.edu Sat Aug 19 16:53:11 1995
From: H. Ward Silver" <hwardsil@seattleu.edu (H. Ward Silver)
Subject: DSP power
Message-ID: <Pine.3.07.9508190809.D16520-c100000@bach.seattleu.edu>
On Fri, 18 Aug 1995, Joe Subich wrote:
> > Good question! Next question? A quick look in my references don't give a
> > clue about this one. Analog blow-by may actually be a limiting factor.
> > 100dB of isolation is tough to obtain without special techniques.
>
> Why should analog blow-by be a problem? It should be no more difficult to
> design a DSP which does it's sampling at 455 KHz or 50 KHz and provides its
> output at baseband (audio). The ideal decoder functions are relatively easy
> to implement in DSP. Since the output would be at a frequency well separated
> from the input (IF), blow-by would be non-existent.
>
Speaking from an analog viewpoint, 100dB of isolation between input and
output of a filter stage is mighty tough. Let's take an example:
Assume 500 ohms input and output impedance through a filter stage. For a
signal to leak around the filter requires a parallel path of 100,000 x 50
or 5Mohms. At 455kHz, this is equivalent to a 0.07pF capacitance. It
gets worse as frequency increases. This is the leakage capacitance budget
which must account for the PC-board, wiring, supply impedance, ground
contaimination, etc.
As Joe points out, though, by performing the demodulation in the same
algorithm, there is no "output" to contaminate by an analog path. So what
does it take to get 100dB of isolation? Well, after some reflection, the
simplest consideration is that you need 100dB of dynamic range in the
converter. That's a voltage ratio of 100,000:1 according to 20 log (100000).
You need 17 bits of resolution (131072:1) to get there. Then add two or
three bits to get some margin and handle math wierdness, and you've
specified a 20-bit converter running at 1Msample/sec. That's just the
converter and says nothing about the processor that's going to do the
dirty work of performing all those multiplications at a heckuva lot faster
than 1Moperations/sec.
This is not to say it's impossible...it's not! It's being done right now,
but not in a consumer device. I would not bet against it being in a
competitive radio by the year 2000...they would have to be sketching out
the design starting next year in order to put it on the shelves by then.
At today's technology level, there's a whole market basket of stretches to
get there, though.
What about math tricks to "synthesize" 100dB of dynamic range out of
something more attainable, like maybe 80dB? Could be, could be. You can
save sampling rate by undersampling. (sacrifice s/n and aliasing
performance) You can save on sample resolution by using parallel
converters with different resolutions, combining the results for a
composite representation. (hardware solution) Clever AGC can maintain
input level within a restricted range. (software/hardware solution) These
are all tradeoffs that require certain assumptions to be made...most of
which will be violated during SS on 20 meters ;-)
I am sure that incremental improvements over the next several years will
be enough to replace *most* of the problematic analog hardware with digital
solutions at the consumer level. The radio-on-a-disk is probably a ways
off, though ;-)
Come to think of it...isn't PED the radio-on-a-disk?
A cool and substantive thread, this...
73, Ward N0AX
>From H. Ward Silver" <hwardsil@seattleu.edu Sat Aug 19 17:27:39 1995
From: H. Ward Silver" <hwardsil@seattleu.edu (H. Ward Silver)
Subject: New Radios
Message-ID: <Pine.3.07.9508190939.H16520-a100000@bach.seattleu.edu>
On Fri, 18 Aug 1995, Jeff Singer wrote:
> As long as they're bringing out this new generation of radios, why
> don't they do something really elegant: Eliminate half the buttons,
> switches, knobs, etc.
Right on the money, Jeff! The basic radio hasn't changed since the 20's.
One each box, several each knobs comma pushbuttons, one each dial...etc.
While we need all of the various adjustments at one point or another, how
many do we use in real time? Five? What we need is one exquisitely
responsive input device that simultaneously allows control of center
frequency and channel characteristics (width, gain) like a big joystick
with a couple of thumb buttons - l/r to tune, back and forth to widen and
narrow, up/down to control gain.
The radio? Oh, it's a box under the table next to that kW brick.
73, Ward N0AX
|