Chris Bartram G4DGU
Having, courtesy of Dom, seen information on the Norton feedback amplifier circuit, can I just make a couple of comments from rather a lot of direct experience of 'noiseless' feedback linearised amplifiers, and related circuitry, make a few observations and comments?
I'm not suggesting that the basic negative feedback topology doesn't work. It does, but really only in its original format at VHF. The problem is that both the active device and the feedback transformer, like most circuit elements at high frequencies, have phase shifts which can easily combine at frequencies above the operating frequency to produce positive feedback. If the gain of the amplifier is greater than unity at that frequency, a spurious oscillation WILL result! I have a tee-shirt collection ...
Those of us who were using the technique 40 years ago were able to make stable amplifiers using now historical devices like the BFT66, BFR91, BFR96 etc. They worked - with a bit of fiddling, such as a ferrite bead in series with the collector lead simply because the fT of the devices was quite small. I even managed to design a reliable NFB LNA using a C-band GaASFET at VHF, albeit using a different circuit. A lot of these designs found the way into production - indeed the FET amplifier circuit was ripped-off by another designer for an IF preamp design for a military radar!
The Ft of the next generation of bipolar devices was rather greater than that of the previous parts, and the simple Norton approach became untameable. Simply replacing the transistor in DJ7VY's design with something more recent is very likely indeed to result in an 'informal oscillator'!
There are other -ve feedback circuit topologies which can be used to make feedback linearised low-noise amplifiers (and interestingly transmitter power amplifiers) but they suffer from the need to have a hybrid in the input signal path. I have looked at linearising modern broad-band MMICs, but there isn't a simple reproducible topology which I've yet found despite a lot of playing both on the bench and by using modelling. I have achieved good performance from a couple of prototypes at 144, though, with IPI3 figures in the +30dBm region and NFs ~1.2dB, but they haven't proved reproducible.
My approach, FWIW, to the design of a second stage of a 'bomb-proof' receiver at 432 or 1296MHz would be to use a group of more modern MMIC devices such as the PGA-105 combined together with hybrids. Two groups of two combined using 0degree hybrids and then combined into a push-pull amplifier would give very good performance, but beware of mixer damage if you have strong local signals!!