array of micro speakers with shading in ATH4 horn

this looks very interesting: https://www.renesas.com/us/en/docum...-audio-algorithms-rev100?language=en&r=502821

any idea of how much DSP you will be able to fit onto the chip?
Not a lot but enough for delay, gain and a filter. Let me set something up and post an example. The PWM outputs can also be delayed by the ns too so the load on the PSU can be spread too.

"AudioCanvas"
https://www.renesas.com/us/en/docum...-audio-algorithms-rev100?language=en&r=493746
 
Last edited:
For a mono channel I can squeeze in

left/right mixer - > gain -> biquad(z) -> 31 band EQ > biquad(z) > 12 channels with delay to 170ms, two biquad filters & gain.

Maxed out the DSP!


1668513461439.png
 
Using the 12 channels in this way you would need to have external inverting buffer for pwm low or I could drop down to 9 channels and generate high and low from the D2 as we have 18 powered outputs.

This would allow me to increase the delay to outermost to 350ms as it would free up resources.
 
Last edited:
Latest release of ath4 has allowed me to shade & delay the flat array. I now have a somewhat flatter vertical response out to around 20deg. WIP.
I think that frequency shading makes more sense than amplitude and delay shading, but anything that reduces the effective length has a similar effect.
As I said before a narrow vertical response in these simulations is not a bad thing for a line but it will make the crossover to a woofer more abrupt when the directivity suddenly explodes. A coupe of smaller drivers either side of the small array would smooth that out.