[hpsdr] Odyessey-Siren Rev B.

[Graham Haddock]

 > Hi Graham,
 >
 > When deciding to use 180 degree switches, the circuit
 > simplicity and power saving were the primary concerns in my
 > mind. Any possible 2 dB noise figure increase on the fixed IF
 > frequency, or even across the whole HF spectrum, doesn't mean
 > very much.
 >
 > On the other hand, due to the non-ideal switches
 > (unsymmetrical switching delays, rise times) there are double
 > number of lossy switching transitions per time with the 90
 > degree solution than the suggested 180 degree configuration.
 > When optimizing all other parameters (especially the
 > impedance match) for the corresponding case, the losses in my
 > limited experiments indicated slightly lower losses with the
 > 180 degree switching.
 >
 > Using ideal components, I think, the both configurations
 > should be equal. This is only my layman's understanding and
 > findings so I may have made some mistakes in my measurements.
 > Also the comparisons naturally were made with different
 > switches that possibly had different on-resistances.
 >
 > Are there any mathematicians around to tell the theoretical
 > truth?
 >
 > 73, Ahti OH2RZ


Hi Ahti:

The 2 dB difference in gain and Noise Figure between 90 degree
sampling and 180 degree sampling is mathematical, and includes no
implementation losses.  Implementation losses could increase
or decrease the difference. 

The number of switching transitions are the same for 180 degrees
and 90 degrees sampling, that is, one ON and one OFF per cycle. 
The only difference being how long the time between ON and OFF.

I include part of an email discussion (below) with Dan Tayloe, N7VE,
as a part answer for your "theoretical truth."  It should be very easy to
verify in a SPICE simulator with ideal switches, and compare the ratio
of input signal level to the output level.

====

Dan Tayloe, N7VE wrote:      [softrock forum Dec 14/2006]
 > ... .
 > The primary difference in detectors is the type that samples
 > 180 degrees, half an RF cycle onto detector cap, and the type
 > of detectors that sample a quarter an RF cycle onto a
 > detector cap.
 >
 > ... .  In theory, a half cycle sampling gives a 3 db
 > conversion loss as it averages half a sine wave onto the
 > detector caps. The quarter cycle RF cycling detector in
 > theory produces a 1 db conversion loss. Think of the average
 > of the 1/4 cycle sine wave peak producing 0.9x of the peak 
 > input voltage. This voltage loss is not readily viewable,
 > since the loss is across the source impedance.
 >
 > In theory, sampling 1/8th of a cycle would improve
 > sensitivity even further, but after you get to 90%, there is
 > not much reason to go further. In addition, you would need to
 > figure out how to put back together 0, 45, 90, 135, 180, 225,
 > 270 and 315 degree outputs.
 >
 > - Dan, N7VE

====

Your particular circuit might not favor 90 degree
sampling, because in that case, each half of the differential
input transformer would be unterminated for half of the time,
Simply switching from 180 degree switching to 90 degree
switching may not work well in your circuit.

I agree that 2 dB reduction in gain and NF would
not be noticeable on HF bands.  I thought that SIREN was
being designed for IF for VHF/UHF receiver, and the extra
2 dB gain and NF might be welcomed, or perhaps not.

Your circuit may have an advantage on the transmit side,
because of the differential transformer.  Have you made
any measurements of the (switching) carrier leak through
on this design?  As compared to the original QSE design?

My best regards, --- Graham / KE9H

==

>   


 1166563811.0