Hello,
I have been using C6713 DSK. My question is; Is there antialiasing filter in
front of AIC23 codec of DSK. If yes, can bypass (omit, cancel) it?
In other words, can I sample over 48 kHz frequencies with aliasing
intentionally.
Tnx.
---------------------------------
Expecting? Get great news right away with email Auto-Check.
Try the Yahoo! Mail Beta.
bypass antialiasing filter of TLV320AIC23
Started by ●June 17, 2007
Reply by ●June 19, 20072007-06-19
Mursel-
>
> I have been using C6713 DSK. My question is; Is there antialiasing filter in front
> of AIC23 codec of DSK. If yes, can bypass (omit, cancel) it?
>
> In other words, can I sample over 48 kHz frequencies with aliasing intentionally.
You have to be specific. What exact frequency range do you want to alias?
The AIC23B device is a sigma-delta codec with sampling rates up to 96 kHz. Sigma-delta technology implies inherent
anti-alias filtering (based on built-in FIR filters) up to the over-sampling rate (from 250 x fs to 384 x fs for this
device). Anti-aliasing above those rates has to be done externally (usually a simple RC stage). I don't think there
is any way to turn off the internal FIR filters.
For example, if want to deliberately alias a 48 kHz bandwidth starting at 12.288 MHz, then you could omit the external
stage (or set it slightly higher). But if you want a frequency range from 48 to 96 kHz to alias, that's not going to
happen.
All the information you need to figure this out is in the device data sheet:
http://focus.ti.com/lit/ds/symlink/tlv320aic23b.pdf
-Jeff
>
> I have been using C6713 DSK. My question is; Is there antialiasing filter in front
> of AIC23 codec of DSK. If yes, can bypass (omit, cancel) it?
>
> In other words, can I sample over 48 kHz frequencies with aliasing intentionally.
You have to be specific. What exact frequency range do you want to alias?
The AIC23B device is a sigma-delta codec with sampling rates up to 96 kHz. Sigma-delta technology implies inherent
anti-alias filtering (based on built-in FIR filters) up to the over-sampling rate (from 250 x fs to 384 x fs for this
device). Anti-aliasing above those rates has to be done externally (usually a simple RC stage). I don't think there
is any way to turn off the internal FIR filters.
For example, if want to deliberately alias a 48 kHz bandwidth starting at 12.288 MHz, then you could omit the external
stage (or set it slightly higher). But if you want a frequency range from 48 to 96 kHz to alias, that's not going to
happen.
All the information you need to figure this out is in the device data sheet:
http://focus.ti.com/lit/ds/symlink/tlv320aic23b.pdf
-Jeff
Reply by ●June 19, 20072007-06-19
There is no specific range. Because I want to proof only subsampling.
I have generated AM signal carrying single tone (1 kHz) by a signal generator. . I have put carrier at integer multiples of sampling frequency . I could hear very very weak tone and I have detected tone changings when I change tone frequency.
(An Agilent digital signal generator I have used: it couldn't generate under 100k carrier for AM and FM. Over 100 kHz, I have tried any integer multiples of 8 Khz)
There is no channel noise, and attenuation. I have applied directly C6713 DSK's microphone input. So, tone has to be heard purely and strongly. Then I have suspected from anti aliasing filter. I have learnt that there is one on AIC23 codec. And also I have learnt that it could not be bypassed. You have been saying same thing. Isn't it?
But you say different and interesting thing: If I want to sample between
12.288.000---12.336.000 Hz
i.e 256x48000---257x48000 Hz
The antialiasing filter could not affect that band. Is it true? How can I extend, generalize this rule for whole spectrum and also AIC23 all sampling rates? Please answer these question specially.
where did you learn this knowledge? from here? http://focus.ti.com/lit/ds/symlink/tlv320aic23b.pdf
I have read this document. But unfortunately, I could not understand some concepts? Like "256xfs", like "sidetone insertion" ??
I have written www.spectrumdigital.com (the manufacturer of C6713 DSK) they only suggest to read that document :)))
I am sure, when you see this message you will say: "not much detailed maaaan" :)))
Thanks for your patience.
Jeff Brower wrote:
Mursel-
>
> I have been using C6713 DSK. My question is; Is there antialiasing filter in front
> of AIC23 codec of DSK. If yes, can bypass (omit, cancel) it?
>
> In other words, can I sample over 48 kHz frequencies with aliasing intentionally.
You have to be specific. What exact frequency range do you want to alias?
The AIC23B device is a sigma-delta codec with sampling rates up to 96 kHz. Sigma-delta technology implies inherent
anti-alias filtering (based on built-in FIR filters) up to the over-sampling rate (from 250 x fs to 384 x fs for this
device). Anti-aliasing above those rates has to be done externally (usually a simple RC stage). I don't think there
is any way to turn off the internal FIR filters.
For example, if want to deliberately alias a 48 kHz bandwidth starting at 12.288 MHz, then you could omit the external
stage (or set it slightly higher). But if you want a frequency range from 48 to 96 kHz to alias, that's not going to
happen.
All the information you need to figure this out is in the device data sheet:
http://focus.ti.com/lit/ds/symlink/tlv320aic23b.pdf
-Jeff
---------------------------------
Looking for a deal? Find great prices on flights and hotels with Yahoo! FareChase.
I have generated AM signal carrying single tone (1 kHz) by a signal generator. . I have put carrier at integer multiples of sampling frequency . I could hear very very weak tone and I have detected tone changings when I change tone frequency.
(An Agilent digital signal generator I have used: it couldn't generate under 100k carrier for AM and FM. Over 100 kHz, I have tried any integer multiples of 8 Khz)
There is no channel noise, and attenuation. I have applied directly C6713 DSK's microphone input. So, tone has to be heard purely and strongly. Then I have suspected from anti aliasing filter. I have learnt that there is one on AIC23 codec. And also I have learnt that it could not be bypassed. You have been saying same thing. Isn't it?
But you say different and interesting thing: If I want to sample between
12.288.000---12.336.000 Hz
i.e 256x48000---257x48000 Hz
The antialiasing filter could not affect that band. Is it true? How can I extend, generalize this rule for whole spectrum and also AIC23 all sampling rates? Please answer these question specially.
where did you learn this knowledge? from here? http://focus.ti.com/lit/ds/symlink/tlv320aic23b.pdf
I have read this document. But unfortunately, I could not understand some concepts? Like "256xfs", like "sidetone insertion" ??
I have written www.spectrumdigital.com (the manufacturer of C6713 DSK) they only suggest to read that document :)))
I am sure, when you see this message you will say: "not much detailed maaaan" :)))
Thanks for your patience.
Jeff Brower wrote:
Mursel-
>
> I have been using C6713 DSK. My question is; Is there antialiasing filter in front
> of AIC23 codec of DSK. If yes, can bypass (omit, cancel) it?
>
> In other words, can I sample over 48 kHz frequencies with aliasing intentionally.
You have to be specific. What exact frequency range do you want to alias?
The AIC23B device is a sigma-delta codec with sampling rates up to 96 kHz. Sigma-delta technology implies inherent
anti-alias filtering (based on built-in FIR filters) up to the over-sampling rate (from 250 x fs to 384 x fs for this
device). Anti-aliasing above those rates has to be done externally (usually a simple RC stage). I don't think there
is any way to turn off the internal FIR filters.
For example, if want to deliberately alias a 48 kHz bandwidth starting at 12.288 MHz, then you could omit the external
stage (or set it slightly higher). But if you want a frequency range from 48 to 96 kHz to alias, that's not going to
happen.
All the information you need to figure this out is in the device data sheet:
http://focus.ti.com/lit/ds/symlink/tlv320aic23b.pdf
-Jeff
---------------------------------
Looking for a deal? Find great prices on flights and hotels with Yahoo! FareChase.
Reply by ●June 19, 20072007-06-19
Mursel-
> There is no specific range. Because I want to proof only subsampling. I have
> generated AM signal carrying single tone (1 kHz) by a signal generator. . I have
> put carrier at integer multiples of sampling frequency . I could hear very very
> weak tone and I have detected tone changings when I change tone frequency.
To 'hear' a tone, you are running audio loopback program on the DSK board, with 8 kHz
sampling rate, and you listen on headphone or line-level output? If so, there might
be RF coupling with the components on the DSK and cabling to/from the card and you're
hearing audio-range harmonics. For a better test of the AIC23B's internal filtering
capability, I suggest to run the signal generator directly at 16 kHz. Do you still
hear a tone? You can bump the sampling rate to 32 kHz and run the same test for
comparison.
> There is no channel noise, and attenuation. I have applied directly C6713 DSK's
> microphone input. So, tone has to be heard purely and strongly. Then I have
> suspected from anti aliasing filter. I have learnt that there is one on AIC23
> codec. And also I have learnt that it could not be bypassed. You have been saying
> same thing. Isn't it? But you say different and interesting thing: If I want to
> sample between
>
> 12.288.000---12.336.000 Hz
> i.e 256x48000---257x48000 Hz
>
> The antialiasing filter could not affect that band. Is it true? How can I extend,
> generalize this rule for whole spectrum and also AIC23 all sampling rates? Please
> answer these question specially.
You have to be careful with the concept of "anti-aliasing" inside a sigma-delta
converter. The device is over-sampling (in this example at 256x the output sampling
rate), and digital filters are applied internally to attenuate frequencies above the
output rate. These are really low-pass filters, not anti-alias filters. The actual
anti-alias filter is applied outside the device, to attenuate frequencies above the
over-sampling rate. My main point was, if you don't apply the external filter, then
aliasing can occur, but only for frequencies above the over-sampling rate.
As for the AIC23B itself, it has four (4) oversampling rate options. I don't know if
these apply to all possible sampling rates or not. You have to read the data sheet.
There is not going to be a general rule -- just the finite set of oversampling rate +
sampling rate options the device allows.
> where did you learn this knowledge? from here?
> http://focus.ti.com/lit/ds/symlink/tlv320aic23b.pdf I have read this document. But
> unfortunately, I could not understand some concepts? Like "256xfs", like "sidetone
> insertion" ??
"256xfs" means 256 times the sampling rate (Fs, or Frequency-sampling).
Sidetone insertion simply means routing a small fraction of mic input directly to
line level and/or headphone output -- sort of a telephone handset emulation thing.
You can figure these things out using the web.
> I have written www.spectrumdigital.com (the manufacturer of C6713 DSK) they only
> suggest to read that document :))) I am sure, when you see this message you will
> say: "not much detailed maaaan" :)))
Figuring out audio codec data sheets is never easy -- takes time, patience and
additional research. These are complicated devices.
-Jeff
> Jeff Brower wrote:
>
> Mursel-
> >
> > I have been using C6713 DSK. My question is; Is there antialiasing
> filter in front
> > of AIC23 codec of DSK. If yes, can bypass (omit, cancel) it?
> >
> > In other words, can I sample over 48 kHz frequencies with aliasing
> intentionally.
>
> You have to be specific. What exact frequency range do you want to alias?
>
> The AIC23B device is a sigma-delta codec with sampling rates up to 96
> kHz. Sigma-delta technology implies inherent
> anti-alias filtering (based on built-in FIR filters) up to the
> over-sampling rate (from 250 x fs to 384 x fs for this
> device). Anti-aliasing above those rates has to be done externally
> (usually a simple RC stage). I don't think there
> is any way to turn off the internal FIR filters.
>
> For example, if want to deliberately alias a 48 kHz bandwidth starting at
> 12.288 MHz, then you could omit the external
> stage (or set it slightly higher). But if you want a frequency range from
> 48 to 96 kHz to alias, that's not going to
> happen.
>
> All the information you need to figure this out is in the device data
> sheet:
>
> http://focus.ti.com/lit/ds/symlink/tlv320aic23b.pdf
>
> -Jeff
> There is no specific range. Because I want to proof only subsampling. I have
> generated AM signal carrying single tone (1 kHz) by a signal generator. . I have
> put carrier at integer multiples of sampling frequency . I could hear very very
> weak tone and I have detected tone changings when I change tone frequency.
To 'hear' a tone, you are running audio loopback program on the DSK board, with 8 kHz
sampling rate, and you listen on headphone or line-level output? If so, there might
be RF coupling with the components on the DSK and cabling to/from the card and you're
hearing audio-range harmonics. For a better test of the AIC23B's internal filtering
capability, I suggest to run the signal generator directly at 16 kHz. Do you still
hear a tone? You can bump the sampling rate to 32 kHz and run the same test for
comparison.
> There is no channel noise, and attenuation. I have applied directly C6713 DSK's
> microphone input. So, tone has to be heard purely and strongly. Then I have
> suspected from anti aliasing filter. I have learnt that there is one on AIC23
> codec. And also I have learnt that it could not be bypassed. You have been saying
> same thing. Isn't it? But you say different and interesting thing: If I want to
> sample between
>
> 12.288.000---12.336.000 Hz
> i.e 256x48000---257x48000 Hz
>
> The antialiasing filter could not affect that band. Is it true? How can I extend,
> generalize this rule for whole spectrum and also AIC23 all sampling rates? Please
> answer these question specially.
You have to be careful with the concept of "anti-aliasing" inside a sigma-delta
converter. The device is over-sampling (in this example at 256x the output sampling
rate), and digital filters are applied internally to attenuate frequencies above the
output rate. These are really low-pass filters, not anti-alias filters. The actual
anti-alias filter is applied outside the device, to attenuate frequencies above the
over-sampling rate. My main point was, if you don't apply the external filter, then
aliasing can occur, but only for frequencies above the over-sampling rate.
As for the AIC23B itself, it has four (4) oversampling rate options. I don't know if
these apply to all possible sampling rates or not. You have to read the data sheet.
There is not going to be a general rule -- just the finite set of oversampling rate +
sampling rate options the device allows.
> where did you learn this knowledge? from here?
> http://focus.ti.com/lit/ds/symlink/tlv320aic23b.pdf I have read this document. But
> unfortunately, I could not understand some concepts? Like "256xfs", like "sidetone
> insertion" ??
"256xfs" means 256 times the sampling rate (Fs, or Frequency-sampling).
Sidetone insertion simply means routing a small fraction of mic input directly to
line level and/or headphone output -- sort of a telephone handset emulation thing.
You can figure these things out using the web.
> I have written www.spectrumdigital.com (the manufacturer of C6713 DSK) they only
> suggest to read that document :))) I am sure, when you see this message you will
> say: "not much detailed maaaan" :)))
Figuring out audio codec data sheets is never easy -- takes time, patience and
additional research. These are complicated devices.
-Jeff
> Jeff Brower wrote:
>
> Mursel-
> >
> > I have been using C6713 DSK. My question is; Is there antialiasing
> filter in front
> > of AIC23 codec of DSK. If yes, can bypass (omit, cancel) it?
> >
> > In other words, can I sample over 48 kHz frequencies with aliasing
> intentionally.
>
> You have to be specific. What exact frequency range do you want to alias?
>
> The AIC23B device is a sigma-delta codec with sampling rates up to 96
> kHz. Sigma-delta technology implies inherent
> anti-alias filtering (based on built-in FIR filters) up to the
> over-sampling rate (from 250 x fs to 384 x fs for this
> device). Anti-aliasing above those rates has to be done externally
> (usually a simple RC stage). I don't think there
> is any way to turn off the internal FIR filters.
>
> For example, if want to deliberately alias a 48 kHz bandwidth starting at
> 12.288 MHz, then you could omit the external
> stage (or set it slightly higher). But if you want a frequency range from
> 48 to 96 kHz to alias, that's not going to
> happen.
>
> All the information you need to figure this out is in the device data
> sheet:
>
> http://focus.ti.com/lit/ds/symlink/tlv320aic23b.pdf
>
> -Jeff
