How to identify the source of low frequency noise from the street?

Question

Is there a way to identify the source of low frequency (166Hz) noise coming from the street (I suspect from the opposite building in the closed yard in the city)?

The noise comes hearable in the yard as well and comes, maybe from the root or vent system, I can’t be quite sure since I don’t have access there and housekeepers say that nothing there could produce this noise.

Can I without a special expensive equipment somehow localize the direction to the source?

I have good microphone, PC and some other household appliances.

Without getting straight to the roof, just from my window, can I somehow get the exact direction to the source?

I am thinking about:

1. Two moving microphones, parallax, interference and computing direction (and even, possibly the distance) to the source.
2. Cone tube with microphone to make in somewhat “directional microphone”.

Are there any better ways to do so?

I live in the city, so the environment is noisy, but in the night, I can choose some quiet time for measurement.

The problem I am trying to solve is the 166Hz noise that about 25 Db louder than other sounds which passes through the windows (3 glasses of 4 mm) like through a filter that leaves only this noise while cancelling other sounds and working around the clock this drives me crazy.

Update

Based on comment below here is the spectrogram of the range 0-200Hz:

(Made by Android Spectroid application).

(It shows 165Hz, but usually it is 166Hz, I don't think this difference is important).

On the 50Hz peak, its loudness varies up to 30Db (can be less or more than on the picture) while loudness on 166Hz usually vary not more than 5Db until there is a strong wind or loud noise (cars, planes, street cleaning, etc.) and I created both frequencies on my PC and I don't bother with 50Hz sound (not so pressing) while I immediately recognized the 166Hz as "the same one annoying sound".

Here is the range 0-1000Hz based on the questions in comments.

With open window:

and with closed window:

In dynamics all the "possible spikes" after 200Hz are moving up and down and actually there is no stable spike like this at 166Hz. This could be seen from the bottom part of each chart, not sure what to call it, let me name it a "legend".

I have recording for the sounds, I will share if it would help.

Answer 1

Several important clues here, as follows.

First we note the very strong 50Hz spike. I will assume from the spelling of your name that you live in a place with 50Hz AC mains. Many different types of electrical machinery running on 50Hz will produce 50HZ noise- in particular, power transformers in which the core laminations are loose will "hum" audibly at the line frequency, and synchronous AC motors will hum at line frequency too as well as at their rotating speed.

Next we note that the amplitude of the 166Hz signal is almost the same as that of the 50Hz spike, and just as sharp. I don't think this is a coincidence, which means the source of the 166Hz signal is coincident with the source of the 50Hz noise. This implies a large electric motor (perhaps ~several kilowatts).

To radiate that noise across a street and into your yard means the source is most likely an air-circulating fan of some sort which is using its intake or discharge duct as a loudspeaker, since for example a water pump in a building would be much harder to hear since it isn't well-coupled to the air outside of the building.

166Hz translates into a spin speed of 8300RPM for the fan- if it had only one blade, which fans never do. So if we postulate that the fan has six blades, for example, we get a fan speed of 1400RPM which would be in the typical range for a single-phase, one horsepower electric motor driving an HVAC fan. Note also that it is common to drive a big fan with a V-belt and pulleys; the fan turns more slowly than the motor shaft in this case which alters the relationship between motor speed, fan speed, and blade number.

Now, how to trace the sound back to its source? This is hard for low-frequency sounds because they radiate in all directions and bounce off of large objects like apartment walls. This creates something called multipath in which a reflection can be mistaken for the source, and phase cancellation in which a reflection will destructively interfere with another reflection reaching your microphone from a slightly different direction.

You can improve your chances of getting useful information from a microphone analysis by using a very long, skinny shotgun microphone which is highly directional, but the problem with a shotgun mic is that they are most effective at higher frequencies and less so at lower frequencies. To make a shotgun mic work well at low frequencies requires a mic extension assembly that is several meters long.