After I presented my position based on facts and experiments ("Feedback," March 2012), Mr.Monzello attacks again, without any facts to support his erroneous position. GPS and satellite systems all utilize superheterodyne receivers with noise figures significantly lower than 3 dB, which as Mr. Monzello repeats is "not possible." His own experiment results confirmed my position. How he knows what "is not disputed by the engineering community" I do not know, but his conclusion is plainly wrong.
In a final response to Mr. Polivka, I would first like to clear up some fallacies that he continues to espouse. First, I do not, and have never been, in the business of selling analysis software. Like many, I have always enjoyed writing my own code to aid in my hardware design efforts. Second, I have never stated nor implied that you cannot achieve receiver noise figures less than 3 dB. By eliminating the image noise, or employing the proper circuit architecture, receiver noise figures of less than 1 dB are achievable. This can be accomplished by employing architectures such as DSB receivers, zero-IF receivers, and image reject filters and mixers.
Third, the Friis equation for cascaded two ports cannot be directly applied to receiver chains that do not eliminate the image band noise. The Friis equation may still be used if the proper effective noise figure is used to account for the noise contribution from the image band (see "Practical RF System Design," by William F. Egan). And lastly, the lab measurements I supplied to Mr. Polivka are clearly in agreement with the established theory as put forth in the article, and have been incorrectly interpreted by Mr. Polivka.
The theory of receiver noise figure degradation, due to image noise folding over into the IF band, is well established and thoroughly discussed in the literature. A list of some of the technical publications available that support this view are available upon request.