Sonar Part 1: Phased Array Design

In this video, I work through the first real sonar design sprint for the underwater drone project.

After talking to suppliers, the problem shifted into something more specific: I needed to actually design the phased array geometry.

That meant figuring out the element width, epoxy gap, pitch, operating frequency, array size, and field of view. To do that, I built an optimization pipeline that combines FEA, beamforming math, PyTorch phase optimization, and Optuna geometry optimization.

The result is the first array design I’m going to use for testing: an 8x8 piezocomposite phased array operating around 500 kHz.

Chapters:
00:00 What I thought sprint one would be
00:50 Talking to sonar material suppliers
01:25 First reality check: element size and wavelength
02:20 Rethinking the piezocomposite array
03:01 Cost shock and narrowing the design
03:58 The supplier asks for a drawing
05:02 Designing the array with optimization
05:08 Inner loop: optimizing phase delays
06:28 Outer loop: optimizing geometry
07:22 Building the FEA mesh
08:45 Harmonic FEA analysis
10:00 Piezoelectric equations and weak form
11:04 Real and imaginary displacement maps
13:32 Finding resonance in the simulation
15:31 Building the aperture maps
16:38 Understanding the far-field beam map
18:17 Precomputing the beamforming math
20:43 Forward pass with phase delays
21:04 Random steering angles and field of view
22:40 Choosing a sinc function for the loss
25:26 Watching the beam optimization
26:38 Why the inner loop matters
27:22 Optuna outer-loop optimization
29:07 Comparing optimization experiments
30:04 Field of view and array pitch
31:04 Choosing the final field of view
31:52 Final array dimensions
32:51 Ordering the sample batch
32:59 Why build the full optimizer?
33:26 Future goal: aperiodic sparse arrays
34:54 What this sprint accomplished

Видео Sonar Part 1: Phased Array Design канала Tanner Rhymes