[LCC] K3 vs. R-4C MDS measurements

[Randy Prewitt]

More forwarded from an email discussion between Tom, W4YOK in Plano, TX and 
his friend Randy Parmer, W8FN in Arlington, TX concerning a shootout between 
a super Drake R4C and Elecraft K3.  We're sort of like -- eavesdropping.
randy  K4LJA -- an interested observer.
                                      ---------------------------------------------


Receiver sensitivity is probably the easiest thing to
> achieve and the least useful to have at HF. It's a non-issue when it comes 
> to predicting performance in the real world. Ever since the '50s or '60s 
> almost any decent receiver (that excludes my old piece of crap NC-109) has 
> had more than enough sensitivity to let you hear down to the band noise 
> floor, even on a quiet 10 meter band. You can get good sensitivity simply 
> by throwing lots of gain in the lineup and calling it finished; making a 
> receiver with sufficient gain to get the job done that can simultaneously 
> handle the tremendous dynamic range of the environment in a major contest 
> is something else again -- and that's what the Elecraft folks have done. 
> MDS is a pretty useless specification when it comes to characterizing 
> real-world performance. -130 dBm or so, typical of what decent modern 
> receivers can do easily, is so far below the realized noise level when an 
> antenna is attached to the receiver that there's zero benefit in extending 
> the detection capability any lower.
>
> The architecture of the receiver determines the usable dynamic range, 
> defined as the instantaneous difference between the weakest signal the 
> receiver can detect and the strongest signal it can handle without 
> producing distortion. Cheap rice boxes can do around 70 dB dynamic range 
> with a -130 dBm sensitivity, while the K3 can do around 90-100 dB range at 
> a similar sensitivity.
>
> Bad choices in gain setting can easily negate the advantages of having a 
> superior receiver. Let's say for example that the actual noise level in a 
> receiver's bandwidth is -110 dBm. This is probably a conservative estimate 
> of band noise. Even W8JI, who lives in an exceptionally low noise 
> location, has measured his noise floor at 160 meters to be about -125 dBm. 
> Those of us living in or near "civilization" can count on the noise being 
> a great deal higher.
>
> With a -110 dBm noise level if you take the 70 dB DR receiver and adjust 
> its gain to where the noise is just detectable, it can handle a pretty big 
> signal of around -40dBm (S9 + 33 dB, assuming S9 is 50 microvolts or -73 
> dBm) before it gets into intermod trouble. S9 +40 dB signals are easy to 
> find in a big contest, so this receiver is pretty marginal for contesting, 
> even with the gain set optimally. If the gain is run wide open, it can 
> handle a maximum signal of only -60 dBm (S9 + 13 dB). This makes it 
> essentially worthless.
>
> For the 100 dB DR receiver, the highest level signal tolerable with the 
> gain set properly is -10 dBm (S9 +63 dB). With the gain run wide open it 
> can survive maximum levels of -30 dBm (S9 + 43 dB). So just by running the 
> gain up we've managed to reduce the higher performance receiver's signal 
> handling capability to where it is essentially no better than the poor 
> receiver with its gain set correctly.
>
> AGC is used to increase the overall signal handling capability of 
> receivers, but AGC can't improve their instantaneous dynamic range. As a 
> matter of fact, letting strong signals control the receiver gain is a 
> prescription for increased intermodulation distortion. Cranking the gain 
> up to where noise can influence the AGC loop produces even more 
> catastrophic effects, since the receiver gain is effectively modulated by 
> noise filtered through the AGC loop time constants. This produces all 
> sorts of garbage in the receiver's audio that can cover up marginal 
> signals.
>
> Raw sensitivity can't cause a signal to magically appear out of the noise. 
> If the signal to noise ratio is less than unity no amount of amplification 
> will produce anything other than more and more noise. The only things that 
> can extract weak signals from stronger noise are either a reduction of 
> receiver bandwidth or some form of signal correlation technique. Modern 
> DSP techniques allow us to reduce bandwidth to the minimum required to 
> pass a desired signal with no particular difficulty, and there have been 
> some advances in processing techniques that seek to suppress random noise. 
> We're probably pretty much up against the physics of the situation now. 
> The only avenue for receiver performance improvement has been and 
> continues to be extending dynamic performance.
>
> 73...
> Randy, W8FN