Audio Beamforming with Speakers from AVR

[lgarbo]

I'm trying to use between 8 and 16 speakers to beamform an audio signal to direct it in a certain direction. The sounds I'm generating aren't particularly complex (I'm currently generating the sound using PWM on a ATTiny). I know that the audio is low enough frequency I could phase shift digitally with 8 to 16 DACs, however that seems like overkill and a major unnecessary expense in terms of board space, processor time, and components.

Talking to a friend, I found out there are RF phase shifters that work in the GHz range. I searched around and couldn't find anything below 1GHz. This seems to be exactly what I'm looking for except in the AF frequency range. Does such a component exist? If not am I missing a simple solution? Some sort of buffer?

[1design]

Delay lines are impractical at audio wavelengths. In audio all-pass filters are used as a passive solution and variable buffers as the active solution. You can buy a mid range car stereo and it has time shifts for each speaker so you can do phase/time alignment in a car environment. They are done by buffering the channel for N number of samples.

BR

[LaserSteve]

Analog AF phase shifters are often known as All Pass Filters...   Analog  Circuits for  Fixed shifts of 90 degrees for creating SSB radio signals over limited voice frequencies  are readily available, but wide range phase shifting has little use outside of Sonar , Some concert grade speakers, car audio, and a  few crazy music synthesizer folks.   

Your probably up a creek looking for discrete modules.

Steve

[ehughes]

Back up for a second.     Is this a narrow band tone you are trying to beamform?     If not,  your approach will not be functional


   What is the geometry of your array?   How did you get to this geometry?   How are are planning on computing your phase shadings?

Depending on what you are trying to do,   You could use amplitude shading only.

I think your time would be better spend doing some simulations in Julia to assess what you you really before you build any hardware.

That and get a copy of Kindler/Frey (Acoustics) and study the chapter on array/beam forming.

[snarkysparky]

https://en.wikipedia.org/wiki/Bucket-brigade_device

https://www.gearslutz.com/board/geekslutz-forum/615501-bucket-brigade-delay-chip.html

[lgarbo]

Wow, lots of great info!

@ehughes, the signals are discrete tone so beamforming should generally be possible. I'm intrigued by the amplitude shading and I'll certainly read more.

Otherwise, it seems the best option would probably be a multichannel DAC chip. From what I can tell the all pass filters would end up being more complex!

[NiHaoMike]

What about a FPGA driving some MOSFET output stages, with the PWM implemented in the FPGA itself?

[mark03]

Another thing to keep in mind is that unless you're setting this up in an anechoic chamber, or at least a room 99% covered in drapes and carpet, you're not going to notice a huge beamforming effect.  The reflections from hard surfaces, particularly walls, floor, ceiling, ...  will fill in the sound.  Add to that the ear's logarithmic response and, it's actually quite difficult to hear a difference between supposed peaks and nulls in the acoustic field.

To get a feel for this you could set up two speakers playing the same sinusoid in phase, then walk a line broadside to them.  Can you hear the interference peaks and nulls predicted by theory?  If not, you won't be hearing them in your more complex setup either.

[barry14]

Analog delay lines at audio frequencies to do beamforming was once widespread in sonar designs.  The advent of digital signal processing brought an end to that.  However, analog delay lines do work well and have the advantage of no power required, low noise since it uses passive L's and C's, and the ability to form 2 beams in opposite directions from one delay line.  We took advantage of that in a sonar system that I worked on which had an operating band of 2 to 8 kHz.  We had five delay lines which formed a total of 10 beams across a 180 wide sector. Another advantage of the analog delay line is that it is easy to implement amplitude shading (to reduce sidelobe levels) internal to the delay line.  The main disadvantage of analog delay lines is that they generally must be custom made for each application which, of course, can make them relatively costly.

[LaserSteve]

After building some 4 driver, 39 inch long, line  arrays for my church, I found there is only a limited frequency range where I could hear actual near field  lobing at.

 Bass for example is at least a twenty foot long wave in air.  So walk-by lobe testing is tricky in compact arrays unless you find the right frequency range.

I started using Holm Impulse software and a reference mic so I could really dial the space in. Holm Impulse is freeware, and worth playing with.

Steve