Gainclone amp with DAC in active speakers

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm considering a new project and did a few searches on the forum.

This is what I thought: Active speaker with S/Pdif input, digital crossover filters in FPGA, one DAC for each speaker element, and finally one gainclone amp channel for each element.

I would not be too surprised if this has already been done. Do you know of any links?

I want to start by integrating DAC, lowpass filter, volume control, and gain in as few components as possible.


Greetings,

Børge
 
I would not be too surprised if this has already been done. Do you know of any links?

T+A made active speakers with digital input. XO was made with DSPs. It was possible to measure listening room influence and adjust the XO parameters in the DSPs with this data.
http://www.taelektroakustik.de/eng/ta2/ls_solitaire/a_2_d.htm

as far as i know the it was a master and a slave speaker so only one for volume control knob and digital input connector.


Volume control in digital or analog domain?

Reciver - digital xo - DACs - PGA2310s - chipamps ?

i´m very interested how you will make the xo.
 
As far as i know T+A made SPDIF into the master speaker, There is reciver chip, signal is splitted in left and right, one channel send proprietary formatted to the slave speaker, together with information like volume setting. In both speakers the signal is high/low split for each system with digital xo in the DSPs and send to the DACs.
 
I thought S/PDIF was the easiest kind of digital signal to pass to the active speakers, but any digital interface should do. Wlan, TCP/IP, anything with a receiver chip that converts it into I2S and presents me with a clock. Oh, and I need to transfer volume information....

My initial thought of the xo was 1st-order lowpass and hipass. I'm not a speaker builder, so feel free to correct me on this one. Anyway, the sum of the 1st order filters is 1.

I'd probably put the filters into a Xilinx. Then I planned to use DAC smoothing filters adapted to the sampling frequency, not the xo. That way analog circuits may be equal for bass and treble. A stereo DAC would be suitable for a two-way system I guess.

I also wanted to design digital filters to compensate for non-flat gain response of speaker element / box. But I know virtually nothing about speakers at the time.


Greetings,

Børge
 
not every speaker will work well with 1st order xo. Most need higher orders. And a set of 2 speakers satisfied with 1st order may limit the choice to a handfull. Is 2nd order that much difficult?

I think SPDIF and AES/EBU are most important digital input formats.

What about DSP for the filters, so it could be updated and modified to listening room needs, other speakers etc. I think you will have better results doing the filters in software. May also be possible to connect a measurement mic at listening position and let the software find the best filter coefficients by itself.

For the digital input have a look at the CS8412 / Cs 8414 datasheet and the AD1892 datasheet.
 
Borge,

I may not be much help, but perhaps some encouragement.

I am thinking of a similar but less ambitious project than yours.

I am looking to build a pair of self powered speakers to work as satelites with my sub in the computer room.

I am also thinking of using a gainclone amp for each speaker but using the soundcard of the PC for DAC, preamping and volume control.

I am very interested to see how your project works out.

Howard
 
Ive done miles of AES/EBU (AES3) interfaces. Its great for long distances (1000 meters) as the RS-422 like characteristics were designed for exactly this. I have heard objections from some audiophiles (true experts in digital interface design) about the quality of this interface. I dont understand exactly what this issue is. My tastes and needs seem to have been relatively industrial. My applications were mostly broadcast and post production Mix-to-Pix.

SPDIF with BNC might be the way to go. The data format is nearly identical to AES/EBU. There is also a BNC interface for AES/EBU that uses the 'enhanced' data word and a higher p-p voltage.
 
ClassD,

What I have already done is to use Verilog and Xilinx's CoreGenerator to make 8x (2x/4x) oversampling FIR filters. I develop coefficients in Matlab. The oversampler talks I2S and BB PCM1704 and works on my mono DAC board. I'll hopefully get the stereo board etched this evening.

Anyway, I can put filter parameters for the xo into the oversampler code. OR I can make IIR filters in the FPGA. The reason I talked about 1st order filters in a previous post is that 1/(Ts+1) + Ts/(Ts+1) = 1.

Here is my initial idea for a setup:

S/PDIF -> CS8414 -> digital oversampler -> digital IIR filters -> multichannel DAC -> volume control + active filter + chip amp -> speaker element

--
Børge
 
Børge,

Your project sounds very interesting. I also live in Oslo and I have some experience with speakers.

I have access to an anechoic chamber with good measurement gear, and I also have LspCAD crossover simulation software.

I'd be glad to assist you in your project if you need some help with the acoustics.

Regards,

Bjørn Magne
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.