A regenerative receiver design was published in 2001 by N1BYT called the Wheatstone Bridge Regenerative (WBR) receiver. The design was interesting because it achieves good reverse antenna isolation (little oscillator energy can escape to the antenna) while still allowing antenna energy to enter and be amplified by the regenerative detector. The WBR achieves the reverse antenna isolation passively; no additional isolating amplifier stage is necessary.
If you're not familiar with the WBR circuit, AA1TJ writes more about it here: http://aa1tj.blogspot.jp/2011/08/ancients-have-stolen-our-inventions.html. Please see that article to understand the basic idea.
I was intrigued by this idea of passive reverse isolation but couldn't completely grasp how the circuit worked. Until now. A 1928 patent clearly describes the WBR operation: https://www.google.com/patents/US1667513. I have seen no references in online discussions to this patent; I found it myself after a lengthy period of investigation, while attempting to understand the operation of and research the background of the WBR receiver.
The key point from the 1928 patent's circuit description is that, unlike a normal regenerative receiver, the antenna signal does not directly excite the oscillator's tank at all! Instead it flows through the inductor's center tap and simultaneously goes through both arms of the bridge, which leaves both ends of the capacitor, and both non-grounded ends of the inductor, at the same voltage -- thus not exciting the tank. By going through the bridge, the signal reaches the input of the amplifying active device. Only after amplification and regenerative feedback does the signal then first appear at the tank.
Another interesting point is that the antenna energy is not filtered by the oscillator tank before it reaches the active device. This means that the antenna signals (in N1BYT's design, where the antenna load Z1 is an untuned inductance) encounter the non-linear active device without the benefit of filtering and thus AM blanketing or IMD are more likely to occur than in a receiver where the antenna signals are first filtered by the oscillator's tank. The 1928 patent uses an additional tuned circuit in place of N1BYT's untuned Z1 for additional input signal filtering.
So to summarise: Forward coupling from the antenna into the detector is high (*NOTE: see addendum below), because the antenna signal flows symmetrically from the tap point through both arms of the bridge (without exciting the tank) whereupon the signal reaches the amplifier input, is amplified, and is regeneratively fed into the tank for the first time. Reverse coupling from the oscillator to the antenna is very low -- almost zero -- because the inductor voltages at opposite sides of the center tap cancel, leaving no oscillator voltage at the center tap to excite the antenna.
Addendum, 2014-12
After publishing the above explanation of the WBR circuit, I did some more research, both in the form of circuit simulations and in the form of reading 1920's radio publications.
My conclusion from my circuit simulations is that there is a mistake in my previous statement that "forward coupling from the antenna into the detector is high". The forward coupling is not high, because the antenna signal must first go through the amplifier before it reaches the tank. The problem is that this first trip through the amplifier leads to significant loss because the amplifier is configured as a regenerative amplifier, whose output is only a tiny, less-than-unity fraction of its input. That's the essence of a regenerative amplifier: only a tiny fraction of the input appears at the output, but this output is regeneratively fed back to the input, resulting in much larger final regenerative gain after several cycles. With the antenna connected as in the WBR circuit, the signal first gets attenuated by the amplifier stage before the signal is regeneratively amplified, leading to a serious loss of sensitivity. This is especially noticeable with a short whip antenna where signal levels at the antenna are very low to begin with.
My conclusion is supported by a 1924 article by Mr. Fitch, who also authored the above-linked 1928 patent on the WBR circuit. The 1924 Fitch article is here: http://www.americanradiohistory.com/Archive-Radio-News/20s/Radio-News-1924-10-R.pdf (see p. 496 of the magazine, i.e. p. 58 of the PDF). The important thing to notice is the difference in tone between the 1928 patent and the 1924 magazine article. The 1928 patent presents the WBR circuit in a positive light and fails to mention its weaknesses, which was probably necessary to convince the patent examiner that the new invention was worthy of a patent. On the other hand, the 1924 magazine article is much more honest about the circuit's shortcomings, and states that the receivers are theoretically interesting but suffer from very poor gain if perfectly balanced.
I wrote some more about this on the Yahoo Group "regenrx-simulations" which you can see here: https://groups.yahoo.com/neo/groups/regenrx-simulations/conversations/messages/111.
Addendum, 2014-12
After publishing the above explanation of the WBR circuit, I did some more research, both in the form of circuit simulations and in the form of reading 1920's radio publications.
My conclusion from my circuit simulations is that there is a mistake in my previous statement that "forward coupling from the antenna into the detector is high". The forward coupling is not high, because the antenna signal must first go through the amplifier before it reaches the tank. The problem is that this first trip through the amplifier leads to significant loss because the amplifier is configured as a regenerative amplifier, whose output is only a tiny, less-than-unity fraction of its input. That's the essence of a regenerative amplifier: only a tiny fraction of the input appears at the output, but this output is regeneratively fed back to the input, resulting in much larger final regenerative gain after several cycles. With the antenna connected as in the WBR circuit, the signal first gets attenuated by the amplifier stage before the signal is regeneratively amplified, leading to a serious loss of sensitivity. This is especially noticeable with a short whip antenna where signal levels at the antenna are very low to begin with.
My conclusion is supported by a 1924 article by Mr. Fitch, who also authored the above-linked 1928 patent on the WBR circuit. The 1924 Fitch article is here: http://www.americanradiohistory.com/Archive-Radio-News/20s/Radio-News-1924-10-R.pdf (see p. 496 of the magazine, i.e. p. 58 of the PDF). The important thing to notice is the difference in tone between the 1928 patent and the 1924 magazine article. The 1928 patent presents the WBR circuit in a positive light and fails to mention its weaknesses, which was probably necessary to convince the patent examiner that the new invention was worthy of a patent. On the other hand, the 1924 magazine article is much more honest about the circuit's shortcomings, and states that the receivers are theoretically interesting but suffer from very poor gain if perfectly balanced.
I wrote some more about this on the Yahoo Group "regenrx-simulations" which you can see here: https://groups.yahoo.com/neo/groups/regenrx-simulations/conversations/messages/111.
