just looking at this interesting project, a search for those
DACs comes up with numerous things. Would it not be more cost effective to build a set of custom (but simple) boards for interfacing with the FPGA you could possibly put DAC chips and amp chips on the same board(s).
To some extent, yes.
I have tested 6, 8, and 12 driver subsections with apparent success.
Right now, I am building an entirely new set of DACs and AMPs. Unfortunately you can only achieve 0.7 cm spatial resolution with a sample rate of 48kHz before ringing becomes audible, and that just isn't enough to generate the types of wavefronts I want from a straight array with 34mm spacing. Furthermore, a computer can only connect a maximum of 128 USB devices, and I definitely don't want to run into issues there. Each hub counts as its own device, and I still haven't settled on a hub that never drops packets (without costing wayyy too much). I may have to build the usb host myself, which is certainly possible.
With these considerations in mind, I have switched to building 10 channel USB DAC/amp modules. Each board will have one usb device with 10 I2S outputs, and 10 channels of amplification.
Right now I'm testing at 192kHz/32bit and the results sound much better than 48kHz/16bit.
I am still trying to choose the right amp for the project. Any tips? I definitely don't need more than 5-10 watts, but I'm open to over-engineering. Noise from the amp is going to be a significant factor no matter what choice I make, given the sheer number of channels.
Since I would like to save money on the power supply, I have been thinking I would use just one supply for all of the boards. Maybe that would be an issue for a Class D topology?
I have tested 6, 8, and 12 driver subsections with apparent success.
Right now, I am building an entirely new set of DACs and AMPs. Unfortunately you can only achieve 0.7 cm spatial resolution with a sample rate of 48kHz before ringing becomes audible, and that just isn't enough to generate the types of wavefronts I want from a straight array with 34mm spacing. Furthermore, a computer can only connect a maximum of 128 USB devices, and I definitely don't want to run into issues there. Each hub counts as its own device, and I still haven't settled on a hub that never drops packets (without costing wayyy too much). I may have to build the usb host myself, which is certainly possible.
With these considerations in mind, I have switched to building 10 channel USB DAC/amp modules. Each board will have one usb device with 10 I2S outputs, and 10 channels of amplification.
Right now I'm testing at 192kHz/32bit and the results sound much better than 48kHz/16bit.
I am still trying to choose the right amp for the project. Any tips? I definitely don't need more than 5-10 watts, but I'm open to over-engineering. Noise from the amp is going to be a significant factor no matter what choice I make, given the sheer number of channels.
Since I would like to save money on the power supply, I have been thinking I would use just one supply for all of the boards. Maybe that would be an issue for a Class D topology?
I assume he is talking about the minimal time-resolution for processing/output-latency or the minimum delay step on the output.
Staying int the 48kHz Domain and processing it at the same frequency has you stuck to this time-domain and max freq range (nyquest) - but transforming the 48kHz signal into a faster processing matrix enables you to shift or delay the signal finer than the 48kHz would normaly provide. Of course frequency resolution isn´t improved by this step - 48kHz should suffice for what you want.
Staying int the 48kHz Domain and processing it at the same frequency has you stuck to this time-domain and max freq range (nyquest) - but transforming the 48kHz signal into a faster processing matrix enables you to shift or delay the signal finer than the 48kHz would normaly provide. Of course frequency resolution isn´t improved by this step - 48kHz should suffice for what you want.
SmarmyDong and Sabbelbacke,
I've had a chance to dig more into all this, and I can confirm that you are correct about all of this.
With this in mind, I've re-evaluated my development process. I have decreased the boundary size for the room, and I'm implementing a virtual acoustic environment in addition to the WFS.
I've found a source for AVB endpoints that is only $18 per 32 channels, but the amplifiers will need to convert TDM to I2S, and the TDM conversion is about $1 per channel, and the amplifiers are another $2.50 per channel. Still trying to find a way to cut down the cost here.
I'll keep the forum posted with progress on the project, but I've been quoted 3 months lead time on the AVB endpoints, so it might be a while.
I've had a chance to dig more into all this, and I can confirm that you are correct about all of this.
With this in mind, I've re-evaluated my development process. I have decreased the boundary size for the room, and I'm implementing a virtual acoustic environment in addition to the WFS.
I've found a source for AVB endpoints that is only $18 per 32 channels, but the amplifiers will need to convert TDM to I2S, and the TDM conversion is about $1 per channel, and the amplifiers are another $2.50 per channel. Still trying to find a way to cut down the cost here.
I'll keep the forum posted with progress on the project, but I've been quoted 3 months lead time on the AVB endpoints, so it might be a while.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.