Geoff Cottrell, G3XGC, describes his coaxial receiving loop in the March issue of Practical Wireless  (download here). This article is widely available over the internet, which is a good point. The less pleasing fact is that this article contains a ‘bug’ – the primary of the transformer T1 should not be connected to C1 but to the other end of the loop (ie. to the ground).
Here is the corrected drawing:
The corrected drawing of the feeding point.
The antenna is electrically balanced and symmetric about its vertical axis. There is a braid-break in the coaxial outer conductor at the top and a simple matching circuit at the bottom feed-point. It is important to realise that this loop is not a purely ‘magnetic’ antenna – the coaxial cable outer conductor is not an electrostatic shield but forms an integral part of the antenna.
The loop works best when sited far away from local houses. This means putting it at the end of my garden where it feeds some 40m of RG58 coaxial cable running to the shack in the attic. Despite having only passive components at the feed-point, losses are minimal and the signal levels I get at the receiver are in general high enough for me not to need any additional pre-ampliﬁcation. The loop is a resonator. A parallel tuned circuit is formed by connecting the inductance of the loop (i.e. the two ends of the inner coaxial conductor) to the tuning capacitor C1, which can be varied to resonate the loop to any desired part of the band. I chose the c.w. end of the band. The bandwidth of the loop is about 50 kHz, so the loop also covers the s.s.b. DX section (1.840-1.850 MHz). In fact the loop also works well from 1.9-2 MHz, although with reduced output.”
“Transformer T1 uses a type 43 binocular shaped ferrite core with: 6 turns of 22 SWG enameled copper wire (primary) and 6 turns overwound (secondary). One complete pass of a wire through the core counts as one turn.”
The complete drawing of the ‘corrected’ G3XGC Loop is here:
The G3XGC Receiving Loop revisited
My comments to this design:
Although this antenna does not come close to the overall bad design published years ago in the ARRL Handbook:
The ARRL Handbook Loop
The G3XGC Loop is a direct cousin to the K9FD loop design which occupied my ‘test roof’ for more than a year.
The K9FD Loop
Compared to the ARRL Handbook Loop, it provides much better symmetry but still lacks on the impedance matching.
When comparing the G3XGC design to the K9FD, there is improved symmetry due to the used transformer. Anyway, the antenna still has some disadvantages:
although the system symmetry is improved, it is not perfect. The antenna is still CM (Common Mode) vulnerable. Better symmetry = better performance.
this antenna is a typical low-output antenna, the overall performance can be dramatically affected by CM. Better CM rejection = better performance.
for this antenna is the coax feedline NOT the best solution. Usually a balanced line (twisted pair) provides better results.
the 43 ferrite mix used for the transformer is not the best solution for 160 or 80 m. 73 mix (ie. BN73-202 core) with two turns primary/two turns secondary would provide better results (less turns => less capacitive coupling => better CM rejection), the 73 has also lower losses in the LF end of HF spectrum.
the impedance matching is still provided by a simple capacitive divider.
If the incorrect interpretations of the antenna principles are avoided, there is still possible to improve the performance.
this loop is NOT a magnetic antenna.
this loop is NOT a shielded loop, actually the outer conductor forms two short, bent (C shaped) phased verticals. The inner conductor forms a coupling loop.
the visual similarity to the small, high-Q loop used for transmitting purposes by many QRP freaks is misleading. This can lead to misinterpretations.
Often disputed is the tuned vs. untuned design. A tuned version provides higher output, the antenna can be built without any preamplifier. The resulting BW is some ~30 kHz on 160 m or ~ 60 kHz on 80 m (tested here). I operate exclusively CW so the antenna tuned to 3525 kHz meets my requirements. A CW/SSB coverage requires two relay switched sets of tuning/matching capacitors (a motor driven tuning capacitor would be probably an overkill). The additional selectivity is worth in the areas densely populated with AM MW broadcast stations, the overall IM performance of the receiver is improved.
An untuned (aperiodic) version needs an amplifier. A nice version providing an outstanding CM rejection, employing a twisted pair from an ethernet cable as the feedline described LZ1AQ  (the PN2222A or 2N2222A transistors should be replaced with BFQ18, NE46134 or NE85634 for better noise figure and IM performance). Other conductors of the ethernet cable can be used for the powering and/or the control of additional relay protection circuits which is a must if the antenna is used at the same location with transmitting antennas. The antenna can be used not on 160 and 80 only but on any band and often beats the transmitting antennas (verticals) also on 40, 30 and 20 metres.
My design of coax receiving loop antenna will be described later.
 Geoff Cottrell, G3XGC: A Noise-Reduction Receiving Loop Antenna, Practical Wireless, March 2007, p. 44-46
 Chavdar Levkov, LZ1AQ: Wideband Active Small Magnetic Loop Antenna, http://www.lz1aq.signacor.com/docs/wsml/wideband-active-sm-loop-antenna.htm