Dan sent this very interesting reply to me and he agreed to having me 
send it to the group as I think it is of general interest.  I copied it 
to him directly so that you can see his email in the 'to' field in case 
you want to reply to him directly.
Thanks Dan!
-- 
73,

Roger Rehr
W3SZ
http://www.nitehawk.com/w3sz
Dan's email to me:
On Hybrid Hardware Set Ups:
   There is another approach here that I was looking at earlier on, and 
that is to try using the wide crystal filters made by ICM or 
International Crystal Manufacturer which is rated +/- 50 kHz with in I 
think 12 dB.   This would cut off any energy that is way out of the 
spectrum from getting into the passband and mixing with the spectrum.  
And still be plenty wide enough for SDR use.  Being that the sampled 
spectrum is currently 48 kHz wide and attenuation at the far ends of 
the bandpass would be minimal.  Example:  I have heard 160 meter lower 
sideband QSO's inverted to upper side band on the lower medium wave 
band.
   Actually the heterodyne mixing in the passband is not very bad 
unless the signal is very strong locally.  Yet if you were to have a 
problem then with a wide filter the idea of selecting a narrower filter 
to cut off the bandwidth and narrow down onto a desired signal may be a 
good idea.  Hence have selectable crystal filters.  I live about 300 
miles from WWCR and WWRB and sometimes one will mix with the other 
being they are close to each in channel spacing.  I get strong extended 
ground waves on them here with some wild back scatter sometimes.  I do 
not think the effect is as bad with SDR as it is with my conventional 
receivers.  So a crystal filter +/- 6 Khz @ 3 dB seems the thing to aid 
this problem.
   I have one half of an analog QSD circuit built and tested here that 
operates so far from 0 to 30 MHz which is how far I tested it with sine 
waves and needs to now be tested with square waves.  It is made up of 
dual gate mosfets.  My testing of it led me to design it for Class A 
amplification to keep it linear and to prevent distortions which should 
be good on minimizing IMD when run Class A.  But this remains to be 
seen with building the other half and then getting a split signal to it 
and trying it on the airwaves.
   I can now try this with a DDS LO IN and an Analog source LO IN to 
compare the two.  And run software to test the phase noise and IMD and 
THD with graphic outputs.  And I have a digitally controlled VFO over 
here to use also in testing.  I am working with digital VFO ideas 
mostly.  It is an analog VFO with digital sampling/comparison and 
stabilization to a Varicap diode.
   A text I compiled way back now, borrowing from an article in the 
ARRL Handbook was for a stereo i.f. receiver that is all analog in its 
approach.  But it was only to explain the idea of phase shifting the 
i.f.  That receiver however is a QSD I/Q source where you can 
experiment with tailoring the audio circuits passband performance up 
into the 100 kHz region such as you can do in the manner of analyzing a 
typical high performance audio circuit with a square wave source and a 
scope.  Using typical feedback compensation around the transistor 
circuits.  I am thinking all Class A audio here also.  And it is HF 
receiver by original design in the 2005 Handbook.
   I am not certain about the following but in this receiver I have 
here from CIAO of Italy I believe there is some control of the agc both 
in the software and hardware, via a signal from the software.  I must 
look at its diagram again to be certain.  But if the agc in the 
software is sampling the tuned to signal and then provides agc in the 
software and a signal back to the hardwares front end or i.f. section 
then the effects of strong signals nearby dominating the agc in the 
hardware will be of no effect since the agc can only be derived from 
the tuned to signal.  I must look this up again.  But this idea of 
alternative agc for the hardware will work only if it comes from 
sampling the desired signal and then routing a control back to the 
hardware with agc also operating in the software.  And I think that 
this scheme is this way for the Synchronous AM detection mode.
   Claudio Re can confirm or disprove how the hardware really works.  
Yet the idea of such a agc scheme will work and overcome the undesired 
effect of a strong undesired signal dominating the agc.  So long as it 
does not introduce phased components into the hardware.
   Finally, there is a whole spectrum of useful software by radio 
amateurs for testing audio performance, in terms of noise and 
distortions.  Many of which run off of a QSD I/Q output.  You can 
capture screen shots of the graphic display of the noise spectrum, so 
remember to look around for spectrum analyzers and signal generator 
softwares and you will turn up allot of useful lab software for 
analysis of the QSD.  You can compile a whole lab of SDR testing 
software at no cost.
   The latest software I have been looking at is the G8JCFSDR_168.exe 
(a full featured SDR radio software) which currently is for use with 
serial and com port hardwares but has a future provision for USB 
hardwares and dll files for the AD9850 and the AD9835 DDS and so I am 
waiting for the full featured USB and DDS support to come out for this 
so I can use it with my hardware here.  It seems that G8JCF is making 
this to work with every sort of hardware to come.  More and more use of 
the USB in the future will come about for receivers and so dll files 
for these receivers in the DRM software and other softwares will be the 
thing to do by some who like to write such files for others use.  I 
think that in the long term manufacturing that will come about; the USB 
port will dominate since it can do all things needed for SDR in a no 
fuss manner on the end user side.

   "the QSD can get away with no sections before it"
   As for analog, whenever possible Class A will be more desirable than 
Class B in the front end amplification sections if this is what people 
are exploring.  Someone might even try a vacuum tube in the front end.  
A good QSD can be made from two Pentagrid converter tubes if one wanted 
to try it.  So for analog front ends the whole thing should be made up 
of low noise mosfet amplifiers and run Class A for least phase and IMD 
distortions.  One or two transistors with perhaps agc added via 
software is all you need before the QSD section.  And no AGC for DRM.
   Remember though, the QSD can get away with no sections before it.  
It is actually very amazing just by itself directly connected to the 
antenna via a low pass filter.  If the sections before it add undesired 
effects then it must operate stand alone to the antenna.



Roger Rehr
W3SZ
http://www.nitehawk.com/w3sz



[Non-text portions of this message have been removed]

Reply via email to