Topband: On the "bleeding edge" - KK7P DSPx module/DSP Blaster/SDR-1000

[Bill Tippett]

VK6VZ wrote:
 >One important point to make for those unfamiliar with this kind of device 
is DSP Blaster is basically software that one uses with a PC, in
conjunction with its sound card, whereas the DSP-10 can be used
independently of a PC, as the back end of a radio, - which is why I am
interested in it as the basis of a stand-alone direct conversion/single
conversion receiver.

         Steve, it's ALL software but in different forms...the DSP-10's is 
embedded firmware and DSP Blaster is a very compact (~17k) machine-language 
program loaded into RAM.  They both use 16-bit ADC and DAC's.  The main 
difference is CPU horsepower.  The DSPx's embedded processor is about 80 
MIPS whereas a 200 MHz Pentium is about 400 MIPS, which DSP Blaster will 
consume if you really want to use all the whistles and bells the program is 
capable of doing.

 >I have sat in his shack and played with it and to be able to bring
the CW filter bandwidth down to 20Hz (or smaller) with absolutely no
ringing and attenuation of signals is a fantastic experience.

         He must be violating some laws of physics to achieve this.  Check 
out SM5BSZ's EME recordings using Linrad (linux-based DSP) at 25 and 17 Hz 
bandwidths:

http://antennspecialisten.se/~sm5bsz/emedemo/xx.htm

The ringing of 20 WPM CW through a 25 Hz filter is quite noticeable to 
me.  The simple reason is that a typical 20 WPM CW signal requires higher 
bandwidth than 25 Hz, and consequently it rings when you try to force it 
into a narrower filter.  A 25 or 17 Hz filter might be OK for 5 WPM but 
even that speed is not very practical on 160.  Remember that the QRSS mode 
which was used for the first transatlantic 136 kHz QSO was using sub-Hz 
filters, but the CW speed was 0.013 WPM!!!  This keying speed / signal 
bandwidth tradeoff is is the reason some observed that K6SE's measurements 
of steady carriers is not representative for keyed 20 WPM CW signals.  A 
steady carrier has theoretically zero bandwidth (excluding phase noise 
effects) so naturally the radios with the narrowest filters (Earl's top 
choices all had 50-60 Hz BW) work best for steady carriers.  I personally 
don't like the 50 Hz filter in DSP Blaster and would say 100 Hz is usually 
my minimum choice for most CW signals (even when using filter shapes with a 
slow cutoff).

         Another problem with ultra-narrow DSP filters is the processing 
delay the computer requires to generate them.  When I was looking into this 
I asked Dave KJ9I (ex-Topbander, now EME'er) about this since he uses 
Linrad.  Here's what he said:

 >Great.. yes Linrad is highly useful for weak signal Rx and I did consider
using it for 160m also.. However I haven't had time to try it on 160.  The
one concern is delay.   Using the very weak signal settings I use for EME
(maximum S/N enhancement) results in approx. 7-8 seconds of delay.. not a
big deal on EME, but a very big problem for HF CW (high-speed).

         DSP is good stuff, but it does have limitations and is by no means
a panacea.  Rather than going for minimum bandwidths in DSP Blaster, I've
found setting BW to 150-225 Hz and enabling the Coherent CW mode
seems to work best for me.  IMHO nothing will ever replace the DSP
between our ears, but I guess you can just call me old-fashioned.  ;-)

                                         73,  Bill  W4ZV