[hpsdr] Cyclops ideas

[David McQuate]

(1) FR4 loss is about 0.25 dB per inch at 4 GHz for 100 ohm 6.2 mil wide 
trace on 0.5 oz. copper.  The dielectric loss is about twice that of the 
copper loss at that frequency, and dominates for higher frequencies.  
Thus wider traces, as for 50 ohm, won't have much lower loss.

(2) For the tracking generator are you thinking of up-converting the 
output of Penny?  With suitable filter this would permit use of Cyclops 
as a transverter.  The phase noise of the ADF4351 is pretty good.

(3) The ADL5801 looks reasonable. All 3 ports have differential ports.  
The ADF4351 has differential outputs, so a balun on the LO port won't be 
needed.  The IF port, being connected to Mercury, is fairly low 
frequency, so an IF balun shouldn't be too hard to find.  That leaves 
the RF input.  The balun used there may be the largest factor in 
determining the overall Rx bandwidth and frequency response flatness.

Since the ADL5801 noise figure is about 10 dB, an input LNA might not be 
necessary for spectrum analyzer use, but for Rx use, an LNA would 
certainly be wanted.  If the LNA had balanced outputs that would 
eliminate the need for the input balun.  Passive mixers have noise 
figures in the 6 to 10 dB range, but usually require much higher LO 
power.  To me, the wide bandwidth, high IP3, and low LO drive 
requirement of the ADL5801 speak strongly in its favor.  It's 130mA 
power supply current might be not that much more than that required for 
a high-power broadband LO amplifier.  (And not having a high power 
signal on the board would help reduce the generation of unwanted signals.)

(4)  Most spectrum analyzers operate by default with a 10 dB input 
attenuator, improving the input match and providing a little protection 
for the mixer.  If one wanted to switch attenuators or filters, there 
are small, inexpensive relays rated to at least 2.4 GHz.

(5) For spectrum analyzer use, since the Rx image is only 100 MHz away, 
an input filter doesn't seem practical, except perhaps for use in a 
specific (small) frequency range.  Harmonics of the LO will not be a 
problem except perhaps at low frequencies.  Experiments with a prototype 
will be needed to see if LO harmonics are a problem.

73,
Dave
wa8ywq

 1332136888.0