Rules | Recent posts | topic RSS | Search | Register  | Log in

A Question about AIC23.

 
Post new topic  Reply to topic    EDAboard.com Forum Index -> Digital Signal Processing
Author Message
Circuit_seller



Joined: 23 Sep 2002
Posts: 413
Helped: 15


Post24 Jun 2004 13:09   A Question about AIC23.

Hi

I have a TMS320C6416DSK which has a AIC23 codec on it. I want to know can i apply a 12kHz signal to it?
I have read the data sheet, in the data sheet specified that a digital filter will filter digital sampling data ( as shown in below). I want to know if 3 means PI or not?

Regards
[img][/img][/u]



Sorry, but you need login in to view this attachment

Back to top
quba54



Joined: 15 Apr 2004
Posts: 9
Helped: 1


Post25 Jun 2004 18:45   Re: A Question about AIC23.

Normalized audio sampling frequency is result of equation of [signal frequency]/[sampling frequency].
So if we have for instance sampling frequency equal 44.1KHz, value 3 in real frequency domain means 44.1kHz*3=132kHz.
Back to top
Circuit_seller



Joined: 23 Sep 2002
Posts: 413
Helped: 15


Post26 Jun 2004 9:47   Re: A Question about AIC23.

Hi

This frequency response is for a digital filter. so i think 3 means pi.
Do you have expriences in working with this chip?

Regards
Back to top
dora



Joined: 11 Jun 2003
Posts: 158
Helped: 2


Post28 Jun 2004 12:31   Re: A Question about AIC23.

Hi Circuit_seller

I have not used AIC23 chip.

I just brawsed the datasheet for you and in my opinion the normalized frquency should be interpred like this.


The '3' should not be interpred as a pi
The interpolation filter in the ADC works at the frequency much higer then the input sampling rate (44.1khz for example).
In my poinion you should assume that 1 coresponds to the input samplig rate so the interpolation filter has cut frequency equal
to the Nyquist rate (half of the sampling rate)

As far as i saw the sampling rate can be adjust using 'Sample Rate Control (Address: 0001000)' register

So assume you use sampling rate 44.1 kHz ([SR3:0 BOSR] = 10001) then 1 on your figure coresponds to 44.1kHz and you
can apply frequency up to 22.050kHz as it is according sampling theorem.
All higer frequency components (the quantization noise aslo) are attenuated by the interpolation filter so you got higer amplitude resolution.

Best Regards
dora
Back to top
Circuit_seller



Joined: 23 Sep 2002
Posts: 413
Helped: 15


Post29 Jun 2004 7:31   Re: A Question about AIC23.

Hi

Thank you for your attention. But i still do not know why TI Specified a digital filter frequency response so. I should test it.

Regards.
Back to top
dora



Joined: 11 Jun 2003
Posts: 158
Helped: 2


Post30 Jun 2004 9:39   Re: A Question about AIC23.

Hi Circuit_seller

I think I know the reason of your confusion. You expect to see pi corespondent to the Nyquist frequency for this filter.
This is indeed the case for all digital filters when the frequency [rad/s] is used as x axis.

In this case however this is interpolation filter. It has clock 256*44.1khz (both coefficients 256 and 44.1 can be adjusted for this chip ... i give just example)
so the Nyquist frequency of this filter is 128*44.1khz.
So noboady is interested in this internal high frequency however and it is convenient to normalize it to the input sampling frequency 44.1khz.
So 1 coresponds to 44.1 khz .

You can easy measure this ADC filter magnitude responce.
Just use white noise (let sey up to 3*44.1khz) and made the spectrum estimation of the acquired data.
(You need white noise generator.) I can made spectrum estimation of the data for you (if you have some troubles in this respect)

For the DAC just send white noise data and measure the continious spectrum. (can send you white noise data , or just use rand in matlab or equivalent software)

You can probe the magnitude response by a sine wave also .

Note that thsi filter actually increase the one bit resolution (at 256*44.1khz) to the (16 bit at 44.1khz) so it is the very inportan part of the sigma delta ADC/DAC.

If I can help with something just let me know!

Best Regards
Dora
Back to top
Post new topic  Reply to topic    EDAboard.com Forum Index -> Digital Signal Processing
Page 1 of 1 All times are GMT + 2 Hours


Abuse
Administrator
Moderators
topic RSS 
sitemap