Why is my best placement of a directional antenna so weird?

Question

I moved to an offgrid house with a 3G/4G repeater, with a short, enclosed directional antenna. It was wobbling and poorly mounted on an unstable wood plank at the base of a window, so I decided to fine-tune its placement.

I checked maps and azimuth for nearby public GSM antennas. The only realistic, nearby one (at about 2km) is indeed right in front of the window, however there is a small hill in between. Others are much farther and way under the horizon.

To my surprise the best result I got is when my antenna is the closest to the window glass, and more surprisingly, my SNR even improved when I placed it parallel to it, pointed towards the concrete side wall of the window, not towards the GSM pole which should be pretty much at right angle.

The concrete probably has rebar so it might create a loop, but I imagine it would certainly be grounded.

This really makes no sense to me and I am looking for an explanation (beyond black magic.)

My only guess is that the triple-glazing wooden framed window might be metal-coated, and it would "amplify" the signal, that the directional antenna catches laterally? Does this even make sense?

I know just enough about radio engineering so as not to trust myself too much and keep a low profile, so I also cowardly checked every angle with the antenna mounted on a tripod on my terrace (same height as the window,) with 5° increments.

No way, the best placement is the one shown on the picture.

• AFAIK it is an LDPA antenna since it catches may wavelengths, not a Yagi.
• The coax cable between the repeater and the antenna is too long. It was stacked vertically along the frame. Worse, it was originally coiled. I use folds instead to avoid loops, but I am not sure it has any impact on the direction (only noise/loss, which I admit was barely noticeable.)

Answer 1

My guess it that your thick concrete wall opening backed by window glass, is acting as a cavity-backed aperture antenna. This type of antenna is more conventionally designed to be sharply resonant, with highly conductive metal sides and back. However, in your case, the reinforced concrete and metallized glass are lossy materials, and the dimensions are several wavelengths long, so it's not so narrowband, but the large aperture helps to make up for the inefficiencies.

As others have pointed out, your antenna is a short log-periodic and has very modest directivity. It's really just functioning as a broadband dipole, so pointing it sideways rather than towards the GSM tower doesn't cause much loss. I think you stumbled on (or, by subconscious genius, selected) a good location for this to serve as an exciting element for the large cavity antenna.

If you're up for more experimentation, you could try moving the antenna up and down along the side of the window aperture.

Answer 2

Given the complexity of your real life problem, I don't think a simple answer will be possible.

But indeed close to metal object, reflections due to the metallic environment will modify the far field pattern of your antenna. In fact, now your antenna isn't just the antenna itself but all the metallic object around it.

If the subject is really ticking your curiosity, you could replace the actual antenna by a dipole and try checking multiple positions to see if one seems better if any.

I'have already seems antenna coupling to rebar, or metallic frames, which seems to act as director, and reflector of a yagi antenna.

Answer 3

If you're using the antenna with a repeater, consider a different problem: what if the repeater is picking up feedback? Perhaps it's clever enough to detect this condition and decrease its gain... The more separation you provide between the inside and outside antennas, the better "performance" you percieve, because the more gain the repeater can reach without oscillation.

Repeaters are double edged :-) Some may cause periodic "outages" in your mobile phone signal, depending on internal construction and on the level of separation between inside and outside antennas. The repeater keeps slowly increasing its gain, and when it detects feedback (instant overdrive) it drops the gain stepwise significantly - starting a new cycle of gradual adjustment until the echo kicks in again, ad infinitum.

Any chance for you to try e.g. RTL-SDR on the bare antenna, without the repeater active? It is still possible that your antenna placement in front of a reflective window is indeed optimal in terms of signal gain towards the tower - but until you try to actually measure something, you will be no wiser. BTW, RTL-SDR is only any good up to maybe 1800 MHz, but the 2100 MHz and higher bands are definitely out of reach. Are you perhaps able to find out, what particular frequency band your phone is picking up? Do you have access to a "stand-alone" GSM/LTE modem by any chance?

GPS or GSM/LTE repeaters are a plausible choice on buildings that have a sheet metal roof (or are coated all around by sheet metal). Like a mountain house with a rugged metal roof or a warehouse/hangar. Even then, any analog repeater principally adds some noise of its own (which can be alleviated by optimal antenna placement) and making a repeater work for multiple bands adds complexity and cost (starting from the design stage). You may find that your off-the-shelf repeater does not support an interesting frequency band that your devices would otherwise choose etc.

In a particular situation = a mountain top with a number of towers visible in a distance, the phones would pick up a great number of "neighbour cells" at extremely weak levels (-120 dBmW) and relatively plausible SNR. And the network would hint the phones to use the higher, urban bands (2100+ MHz) from the towns in the valley, as these have more data capacity and are less occupied by devices. One of the people in the mountain village got a repeater, which reportedly helped him - the phones stopped seeing the weak remote signals (obscured by the repeater's own noise?) and started clinging to the stronger signals in the lower bands (900 and 800 MHz) which was desired from "call stability" perspective. The undesired effect was "pulses of feedbacked interference" due to repeater AGC, as debated in the comments. The wooden house just did not provide enough separation between antennas, in spite of having a metal roof (and yes the outside antenna was placed inappropriately, on a wall).