Topband: 160m Receivers - the argument for true DSP (long)

Steve Ireland sire at
Sat Oct 18 13:20:01 EDT 2003


A couple of weeks ago, Tom W8JI wrote a critique of phasing/direct
conversion front-ends with DSP 'hearts' for use on 160m in reply to a
posting I wrote about them.

As I also received a lot of replies from topbanders who were interested in
a a viewpoint that was argued from the other side (i.e. pro-direct
conversion/DSP), I arranged for my friend Phil VK6APH - like W8JI, a
professional RF engineer - to take address some of the technical issues Tom
raised.  The results are below - and provide an lively and fascinating
alternative view.  

VK6APH operated extensively on 160m in the UK as G3WXO, designed HF/MF
receivers and antenna systems (particularly active ones) for a UK
government department and is currently the R & D director of an 3-D video
company (when not designing topband receivers for VK6VZ).

I have visited Phil's shack and used a DSP radio on 160m based on his
comments below - which impressed me so much that I ordered the KK7P DSP
processing board, mentioned in my original email, as the basis for a direct
conversion/DSP 160m CW-only transceiver.

Vy 73,

Steve, VK6VZ

W8JI: > or discover a cheap analog switch makes a reasonably good mixer.

VK6APH: The FST3125 makes a fantastic mixer - with a 3IP of +44dBm from a
handful of
bits, it does not make 'a reasonably good mixer' but a 'damn fine mixer',
bordering on what the military require in terms of performance.  

W8JI: >The largest single problem we have on 160 is noise, much of which is
>unavoidable except through directivity of  receiving antenna. 

VK6APH: "If the only tool you have in your tool box is a hammer then
looks like a nail".  Multi-directional (non-active) receiving systems are a
luxury only a very
few can aspire to on 160m. We can now build DC 160m receivers for a few
dollars that will sit in the palm of your hand.  Build a system with a
number of these
all running off the same local oscillator and connect each one to a
different active antenna and the outputs to sound cards connected to a PC.
With a little software you can now form a phased array with performance
that would cost $10,000 to build in terms of real-estate and towers. 

In terms of noise if we build a simple DC receiver and use a PC sound card
to sample the quadrature output at 44kHz then we have 88kHz and tens of
milliseconds to look at that band and work out what is noise and what it
signal. Such a wide bandwidth means that the noise pulses are not slurred by
passing through crystal filters, which simply make matters worse. Once in the
digital domain we can then subtract the noise, not just blank it or cancel
it. Software that can perform this sampling is available for free on the
Internet - the effect of such
noise reduction is staggering, especially if your current experience is
limited to IF LMS noise reduction or add-on AF DSP systems on radios such
as the FT1000MP. 
W8JI: >The curable stuff is largely caused by defects in transmitter designs.

VK6APH: Agreed - especially if they one of the current generation of radios
designed in Japan that most of us use.

W8JI: >It's true there are some bad receivers that need help, but ironic
that MOST
>of the really bad radios are DSP based.

VK6APH: Most of the DSP in use in radios today are at the end of the IF
strip.  The
reason that I think they are there is that they add (unnecessary) 'bells and
whistles' to these radios and are cheaper than analogue equivalents. The
system Steve VK6VZ is
proposing effectively digitises the 160m signal directly at the antenna
(OK, actually
after a single mixer in a DC receiver). He is not suggesting that adding a
DSP far
down the IF strip is going to make much difference.

Actually, as those of us who own a FT1000MP know, a DSP at the end of the
IF strip can make
the receiver sound nicer -  the lack of in-band IM in the DSP demodulator can
be quite pleasing.


W8JI:  > My inclination is that DSP systems, unless extremely powerful (and
expensive), bring more problems to the table
>than they cure.

VK6APH: A 2.5GHz PC and a professional quality sound card (circa $100) make an
extremely powerful and inexpensive DSP system. Having spend the past eight
months using such a system, I will never go back to an analogue radio -

W8JI:  >Some radios have spurious response problems (the 775DSP and JRC
radios I
>have seen), some have design flaws causing IM issues (Yaesu NB systems for
>one), some have noise in systhesizers. Many have filter shortfalls. The
>really is just fixing the defects, it isn't a matter of changing

VK6APH: Agreed - I have often wondered if the designers of radios by Yaesu,
etc, ever take their prototypes home and actually use them on the air?
Some of these design flaws are obvious
after using the radio for a few days - some even quicker, like the
relatively poor IF gain
distribution of the FT1000MP, which any receiver designer can spot within
minutes of turning the thing on.

Curing this problem is a matter of changing technology - many of these
design flaws are
easier to fix in software. The answer is to start by directly digitising
the RF signal at
the antenna socket and solve the problems from there back to the speaker. 

W8JI: >If you don't tightly "beam" your receiver through very strong
signals to
>to the DX (and live in a location with very low noise) you probably don't
>need anything beyond a normal "cleaned up" FT1000 or any other reasonably
>good radio. A regular good properly working radio would put most systems to
>the point where the other guy's transmitters are the problem even when
>living in a congested RF area like the USA or Europe.

VK6APH:  Tom, you really need to experience first hand what 'true' DSP
signal processing can do to weak
signal reception. A DSP-based active phased receiving array can allow the
average topbander chasing DX to 'beam' his receiver through strong signals
without requiring a
100 acres of land.


W8JI: >Those of us without large receiving arrays in lower amateur population
>regions could almost get away with a Knight-Kit Star Roamer technology
>receiver with a Time-Wave DSP and do 99.9% of what the best possible system
>would do.

VK6APH: Steve VK6VZ lives in exactly the kind of area and situation you
He has found that moving from an FT1000MP to a heavily modified Drake R4C
(modifications based on the classic Sherwood ideas by me - and with
assistance from you) regularly make the difference between R3 and R5 copy
on a weak signal on 160m.  I think we can both agree that the FT1000MP is a
rather better receiver than the combination of a Knight-Kit Star Roamer and
a Time-Wave DSP.

W8JI: >By the way, none of the data or measurements I have been able to find
>indicate these software-based radios are as good overall as what we have
>now. 40dB TOI is absolutely meaningless unless we know a whole lot more. An
>absolute piece of receiving-junk could have 40dB TOI.

VK6APH: What do you consider important in terms of measurements? 

W8JI: >As for BW. Ringing is a direct function of bandwidth, the shape of the
>bandwidth slope, and varying group delay times through the system as
>frequency is changed. As W4ZV pointed out, we can't possibly use 20Hz or
>even 100Hz  BW with "brick wall" skirts for normal CW. The filter MUST pass
>all sidebands generated by the rise and fall time, or it extends the rise
>and fall!!!
>We have to pass the significant sidebands, or the signal "rings". The
>spacing of the sidebands is dictated by the rise time and fall time. This
>would also apply to any noise pulses going through the filter. Narrower
>selectivity lengthens the duration of noise pulses...even if it is a
>software filter....and makes noise "ring". The more it rings, the more
>sounds like CW.

VK6APH: In terms of system performance, the best you can do is with a
matched filter
at the receiver - which is trivial to achieve with a DSP, but virtually
impossible with conventional
analogue filters.  Please listen to a properly matched DSP filter - ringing is
not a problem and if you have subtracted the majority of the noise pulses
in the pre-filtered bandwidth, then you are in for a very pleasant new

More information about the Topband mailing list