Another Charleston SDR fix

[Terry Fox]

Warning: the following is about the Charleston SDR Receiver.  If you get 
easily bored, hit the delete button now.

I've been a little concerned about how easily the Charleston Receiver 
gets overloaded whenever I hook it to my 40M dipoles late at night.  I 
cannot seem to duplicate this problem using a signal generator.  When I 
hook either dipole to a spectrum analyzer, I see a concentration of very 
strong signals in the AM broadcast band, and a second group of very 
strong signals from roughly 5.5 to 7.5MHz.  Some of these signals peak 
to about -25dBm.  When the Charleston board gets overloaded, I start 
seeing a lot of spikes show up all through 40M, about every 10kHz.

OK, a BCB high-pass filter gets rid of the first group, but what to do 
with the second, since they are right around 40M itself?  My first clue 
was that the spikes were there whenever I enabled the preamp (exited 
preamp bypass mode), and were there whether the preamp/VGA was in high 
or low gain modes.  They were there regardless of the preamp/VGA 
variable gain setting as well.  I assumed it was NOT the A/D or 8201 
itself, but rather one of the switches or the AD8331 preamp/VGA.  
Scoping the RF chain with the spectrum analyzer confirmed that the first 
preamp stage of the AD8331 VGA chip was causing the problem.  But, I 
wasn't hitting the preamp's input with anywhere near an overload signal 
level of -7dBm or so (275mV), but rather signals of about -25 to -30dBm 
or stronger caused the problem.  Using an oscilloscope to look at the 
preamp output/VGA input pins showed that an input sine wave from a 
signal generator was starting to distort on the preamp output pin way 
below the -7dBm level.  So, the preamp was not meeting spec (a term I 
have always hated!).

Further research with the AD8331 data sheet showed a couple of 120nh 
inductors around the preamp section of the AD8331 that are missing on 
the Charleston board.  One is in series with the preamp power (pin 3), 
and the other is in series with the preamp input pin.  Looking at a bare 
board showed that cutting traces to add the missing inductor in the 
power line wasn't going to happen.  The trace in question is under the 
SMT chip itself, and would require significant rewiring if cut.  So,  I 
looked at adding the other inductor, in series on the signal input.  
Aha! There is a SMT chip capacitor (C48) that provides DC isolation 
between the LPF and the preamp.  Taking two soldering irons, I carefully 
removed C48, and placed it vertically on one of the SMT pads.  I then 
put a 120nh SMT inductor on the other SMT pad originally for C48.  Then, 
I soldered across the tops of the keystoned cap and inductor.

Initial testing proved that this simple mod has made a tremendous 
improvement!  I can now hit the preamp with signals up to -6dBm without 
preamp output distortion.  And, I have not seen the overload artifacts 
using on-air signals since I put this mod in place.  I then checked a 
second Charleston Receiver board, and clearly saw the spikes.  After 
five minutes of making the same mod on that board, the spikes were gone 
from it as well.

I will report back further, as there seems to be fewer signals on the 
air tonight.  One of my strongest/worst offenders was "Family Radio?" at 
6985.  I guess they did not get the memo that their prediction of the 
end of the world did NOT happen yesterday, and don't seem to be 
transmitting tonight.  Maybe they take Sundays off, even when the world 
does not end.

Anyway, that's a pretty easy fix to solve a significant dynamic range 
issue with the Charleston Receiver.  Between that change and the 
oscillator change to an Si570, my Charleston boards are beginning to 
operate like real SDR contenders!

Anyway, that's my story, and I'm sticking to it.
Terry, WB4JFI

PS:  My pre-alpha Windows version of Quisk is still working fine.  I'm 
trying to figure out how to package it for distribution.
Terry