Avery Fineman ) writes:

Stirling's description isn't quite good enough in the '74 article covering
the included "reciprocating detector." The principle is that of simply
filtering out the carrier, amplifying it, and mixing it back with the
incoming carrier-plus-sidebands. At the output the carrier, mixed with
itself, becomes a DC level. The sidebands mix with the amplified-and-
limited/filtered carrier to result in the original audio.

Motorola used the same principle in the MC1330P video detector chip
introduced in the early 1970s (1972?). The 1330 had what amounted to
a limiter ahead of the mixing stage. The so-called "capture effect" of
the limiter will output the stronger signal which, in this case, is the AM
carrier. For the "filtering" (as Olberg called it), the 1330 used a simple
resonant circuit tuned to the carrier frequency.

While the MC1330 internal schematic is not well layed out for clear
understanding, it is different from the reciprocal detector.

Unique to the RD design is the fact that the filter is at the output of
the differential amplifier, what it sees at it's input is not just
the incoming signal, but the incoming signal modified by the output
of the filter. That seems to be a key to the design.

The MC1330 splits the incoming signal, with one path going to a mixer,
and the other path going throught a limiter and filter which then feeds
the other input of the mixer. It's behaviour is obvious, ie limit the
incoming signal so it's you get a constant amplitude carrier and mix
it with the incoming signal to beat it down to baseband, and shows up
in plenty of designs, both before and after the Badessa patent. There
was an article in Ham Radio for September 1970 about the MC1496 double
balanced modulator, and Roy Hejhall specifically mentions it's use as
an AM detector in a similar scheme. He said that a limiter wasn't
even needed since the 1496 will limit with enough signal. The same
thing is stated in the MC1496 datasheet though you have to dig a bit
to find it since it's under the "product detector" heading. The filter
in this scheme is optional, because not only is there the 1496 example
but I've seen similar schemes with no filter. The same scheme shows
up in that fairly recent QST article about a synchronous detector (it's
been in the Handbook too), though there it's labelled as
"quasi-synchronous" and it's merely a side circuit to the main part using
a PLL. But if you look in old literature it gets the "synchronous
detector" label.

What is a puzzle is why Olberg did not reference such articles, because
they were halfway there to explaining the reciprocal detector, and what
is vague in his articles is what makes the RD different.

One thing is certain. That "amplify the carrier, limit it and use
it as a locally generated carrier" scheme will not work with SSB,
unless the original carrier is not well suppressed. And it's not going
to work with CW either, since beating a carrier, as you say, against
itself will result in DC and double the carrier. There'll be no beatnote.
Since the RD is claimed to work on these modes, something else has
to be going on.

Michael VE2BVW