Forums

dsPic adc for audio signals

Started by Unknown April 15, 2006
Hi.
With a dsPic (ie dsPic30F4013) I need to process two voice band
channels (for telephony, fs=8kHz).
Can there be problems to use the embedded adc, oversampling 8 times and
then filtering?
Do you think this may waste more than one half of dsp bandwidth?
Why is not usual to use the embedded adc for audio signals?
Thank you.

pp

pinco_pallo@jumpy.it wrote:

> Hi. > With a dsPic (ie dsPic30F4013) I need to process two voice band > channels (for telephony, fs=8kHz). > Can there be problems to use the embedded adc, oversampling 8 times and > then filtering? > Do you think this may waste more than one half of dsp bandwidth? > Why is not usual to use the embedded adc for audio signals? > Thank you. > > pp >
The ADCs that come in microprocessors aren't usually very good. The process necessary to make a good microprocessor isn't the process necessary to make a good ADC, so if you really want the performance you need to get them on two different pieces of silicon. I remember using the old Intel 80196 -- they only made windowed parts in a really early silicon rev, and the ADC suffered from a lot of internal crosstalk. We'd get our software to the point that we thought it was good, then do final performance checks with the OTP parts that had much better ADC performance. -- Tim Wescott Wescott Design Services http://www.wescottdesign.com Posting from Google? See http://cfaj.freeshell.org/google/
pinco_pallo@jumpy.it wrote:
> Hi. > With a dsPic (ie dsPic30F4013) I need to process two voice band > channels (for telephony, fs=8kHz). > Can there be problems to use the embedded adc, oversampling 8 times and > then filtering? > Do you think this may waste more than one half of dsp bandwidth? > Why is not usual to use the embedded adc for audio signals? > Thank you.
It's very easy to interface a codec to the dsPICs, giving excellent speech quality. Why bother with the on-chip ADC? Leon
Tim Wescott wrote:
> pinco_pallo@jumpy.it wrote: > >> Hi. >> With a dsPic (ie dsPic30F4013) I need to process two voice band >> channels (for telephony, fs=8kHz). >> Can there be problems to use the embedded adc, oversampling 8 times and >> then filtering? >> Do you think this may waste more than one half of dsp bandwidth? >> Why is not usual to use the embedded adc for audio signals? >> Thank you. >> >> pp >> > The ADCs that come in microprocessors aren't usually very good. The > process necessary to make a good microprocessor isn't the process > necessary to make a good ADC, so if you really want the performance you > need to get them on two different pieces of silicon.
Rubbish. The ADCs in lots of MCUs are poor, but don't tar them all with the same brush. There are MCUs where a high performance ADC is the centre-piece of the chip. That said, the ADC in the dsPIC doesn't exactly seem to be a stellar performer.
> I remember using the old Intel 80196 -- they only made windowed parts in > a really early silicon rev, and the ADC suffered from a lot of internal > crosstalk. We'd get our software to the point that we thought it was > good, then do final performance checks with the OTP parts that had much > better ADC performance.
Didn't covering the window properly fix that? Steve
Steve Underwood wrote:

> Tim Wescott wrote: > >> pinco_pallo@jumpy.it wrote: >> >>> Hi. >>> With a dsPic (ie dsPic30F4013) I need to process two voice band >>> channels (for telephony, fs=8kHz). >>> Can there be problems to use the embedded adc, oversampling 8 times and >>> then filtering? >>> Do you think this may waste more than one half of dsp bandwidth? >>> Why is not usual to use the embedded adc for audio signals? >>> Thank you. >>> >>> pp >>> >> The ADCs that come in microprocessors aren't usually very good. The >> process necessary to make a good microprocessor isn't the process >> necessary to make a good ADC, so if you really want the performance >> you need to get them on two different pieces of silicon. > > > Rubbish. The ADCs in lots of MCUs are poor, but don't tar them all with > the same brush. There are MCUs where a high performance ADC is the > centre-piece of the chip. That said, the ADC in the dsPIC doesn't > exactly seem to be a stellar performer.
I would still be mighty careful before I trusted one.
> >> I remember using the old Intel 80196 -- they only made windowed parts >> in a really early silicon rev, and the ADC suffered from a lot of >> internal crosstalk. We'd get our software to the point that we >> thought it was good, then do final performance checks with the OTP >> parts that had much better ADC performance. > > > Didn't covering the window properly fix that?
Not if I recall correctly. At the time I was told it was the rev of the silicon, but you could run the system with covers on and get the same problems.
> > Steve
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com Posting from Google? See http://cfaj.freeshell.org/google/
>Rubbish. The ADCs in lots of MCUs are poor, but don't tar them all with >the same brush. There are MCUs where a high performance ADC is the >centre-piece of the chip.
the Analog Devices microcontrollers have A/D's that are as good as any stand alone A/D, but this is a rare exception