Louadspeaker crossower - digi in/out - no analouge

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

TNT

Member
Joined 2003
Paid Member
Hallo!

Any pointers to DIY projects for a X-over without any analouge in and outputs. Idon't need them as I have all the DAC capacity I need. Just digital in/out but with all the slope/freq/delay features that is required for advanced Xovering.

Thanks for any hints to schemas or even existing products!

/J
 
Member
Joined 2004
Paid Member
TNT said:
Hallo!

Any pointers to DIY projects for a X-over without any analouge in and outputs. Idon't need them as I have all the DAC capacity I need. Just digital in/out but with all the slope/freq/delay features that is required for advanced Xovering.

Thanks for any hints to schemas or even existing products!

/J

There are two different approaches that come to mind. First, you can use a general-purpose DSP chip like those from TI, Freescale, Analog Devices, Cirrus or several others and write your own filter routines or build them from the development tool libraries. This approach makes sense if you are already familiar with the DSP device and if you already own the development tools. However, the development tools usually aren't cheap, and these devices typically have a fairly steep learning curve.

Second, you can use a dedicated audio processor such as the TAS3103 or AD1953 that has biquads and delays. This approach is usually fairly cheap because you don't need development tools for most of these devices, the audio processing board is fairly simple, and overall the learning curve is less steep. The challenge in this approach is developing the code to generate the coefficients for the biquads and developing a controller to program the audio processor registers.

I've been working on some tools for this second approach, but right now I've only got code developed for the TAS3004 chip and the biquads in the STA308 chip. However, next up is the TAS3103, which would be a good device for making a general-purpose loudspeaker crossover. That is the chip that Ground Sound uses in their electronic crossover. It's got a lot of flexibility plus good numeric precision.

You can download an early version of the code for the TAS3004 processor at: http://www.audiodevelopers.com/plateamps/BCD.zip

The program will balk at not having the target hardware, but if you disable the USB interface it will let you design crossovers using a math model of the TAS3004. If you had the right hardware, you would be able to hear any change you make in the crossover in near real time (it takes one or two seconds on a fast PC to calculate the coefficients). This program also does something rather nice, in that you can read in frequency response measurement data ("FRD" files) for your drivers and the program will let you interactively tweak the crossover.

At the rate I am going this code will be fully complete sometime around 2015. However, donations, collaboration or other factors, such as my retirement, may shorten this time.
 
BCD program quiestion

Hello,
I built TAS3004 experiment board some time ago. My idea was to make some integrated amplifier and to use TAS3004 for attenuation and equalization. I tried it using spreadsheet calculator which was also uploaded here by Mr. Neil Davis, I think. Everything was OK but... I put aside TAS3004 board as I didn't have enough time to finish that hobby project.

Now I start to think about some DIY home biamp. amplifier's construction with digital crossovers. So I am planning to use TAS3004 for that task.
I have just tried Neil's program with my board. I use data from results window to program 24C16 EEPROM on my board. It seems that biquad calculations were accurate - I got accurate frequency response (measured with PC and Sb card) but there was some annoying hiss sound around 15KHz on each channel.
So is it also somehow calculated (i.e. demo program limitation) or I have some other problem(s)?

Thanks for advice in advance!!!
 
Member
Joined 2004
Paid Member
Re: BCD program quiestion

yugaja said:
I have just tried Neil's program with my board. I use data from results window to program 24C16 EEPROM on my board. It seems that biquad calculations were accurate - I got accurate frequency response (measured with PC and Sb card) but there was some annoying hiss sound around 15KHz on each channel.
So is it also somehow calculated (i.e. demo program limitation) or I have some other problem(s)?

Thanks for advice in advance!!!

Nothing intentionally crippled--just a screw-up that got fixed in a later version.

I'm still working on the BCD code, but shifting focus to other chips that either have better numerical precision than the TAS3004 or that are part of digital power amps. I'm working on an STA328 amp right now, and the AD1953 would be the next after that. There is also a companion program that will control the 6 to 8 channel amps such as the STA308 (which is partially working) or the TAS5518 chip.

The last TAS3004-based amplifier that I built used the AD1994 chip, which appears to be discontinued (*sigh*). The amp is actually analog input so it doesn't really fit in this thread, but I included a picture anyway.

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