[Elecraft] K3 to IC-7800 Comparison?

[Adam Farson]

For Joe, W4TV:

To quote: "On the other hand Adam limits noise power for direct sampling SDR
designs to a lower level than used with traditional up/down conversion
transceivers.  The lower noise power input gives the direct sampling
designs an unfair advantage be ignoring strong signal environments."

As explained in my web article (and also in my QEX article), the optimum
noise loading points for an ADC and a conventional receiver are different.
In the conventional receiver, optimum noise loading is reached when the
noise power induced in the IF passband within the notch (idle-channel noise)
is equal to the DUT's intrinsic thermal noise power in the same bandwidth.
At this point, the DUT's audio output rises by 3 dB. 

Walt Kester of ADI states in ADI Tutorial MT-005 that the optimum noise
loading point for an ADC is where the device's quantisation noise equals the
noise generated by clipping, i.e. the noise loading is run right up to clip
level or 0 dBFS. I use -1 dBFS as my optimum noise loading point, to ensure
that no clipping takes place during the test. In fact, when an ADC is driven
to clip level, it crashes, thus invalidating any tests attempted above clip
level. 

I do not compare NPR test data for direct-sampling receivers directly with
data for conventional receivers. The benchmark I use for direct-sampling
receivers is the theoretical value of NPR for the ADC in use; this can be
calculated using the procedure presented in MT-005 and described in my
articles. The closer the measured NPR value is to the theoretical one, the
better the front end will perform under heavy loading. A large drop in NPR
(10 dB or more) as compared to the theoretical value indicates an anomaly
such as passive IMD in the preselector or IMD in an active stage ahead of
the ADC.

For a conventional receiver, the closer the NPR figure is to the bandstop
filter's stopband attenuation, the better the receiver (at least from the
NPR standpoint). I do not use NPR as the sole criterion for receiver
selection; my intent in adapting this test method to HF receivers is to
provide the test engineer with an additional test tool for evaluating a
receiver's behaviour on a band packed with extremely strong signals.

Along these same lines, it is virtually impossible to correlate certain
narrow-band test results for a direct-sampling receiver with those for a
conventional receiver, as the familiar traditional test metrics (DR3, IP3,
blocking gain compression) are completely meaningless in the context of an
ADC. Phase noise (RMDR) is still very much a valid parameter, but RMDR in a
direct-sampling receiver is usually very high as the ADC clock is the only
major source of phase noise. (ADC aperture jitter is a minor phase noise
source.) Of course MDS is valid for both receiver types. I have proposed,
and myself use a front-end IMD test method in which I measure the absolute
power of the IMD3 products at 2 kHz spacing over a range of input power
levels, and draw a chart. I then draw lines across the chart at the typical
ITU-R urban and rural band noise levels. If the IMD product is below the
site band noise level, it is inaudible and can thus be disregarded. I term
this test IFSS (interference-free signal strength) and use it exclusively in
my direct-sampling SDR test suite.

Ultimately, the decision as to whether to acquire a direct-sampling SDR or a
conventional transceiver comes down to the operator's personal operating
preferences.

73, Adam VA7OJ/AB4OJ