Mehdi Hosseini-

>> Note: One thing I forgot to point out earlier is that a microphone-
>> based system probably will not be susceptible to sample phase issues,
>> because the time delays due to the distance between the mic and the
>> sound source will vary and will far outweigh the sample latency. In
>> other words, this is most likely a non-issue. In fact, that may
>> explain why the data sheets do not even mention the timing of the
>> sample process.
>>
>> Brian Willoughby
>
> You are right, but the sample phase issues is so important for me because I want to obtain time delay of arrival
> (TDOA) between 2 microphones.
>
>>These are good points. Unless the chip data sheet specifically talks about
>>simultaneous sampling performed by multiple devices then sampling phase >alignment is likely to be an issue.
>>And if the data sheet doesn't talk about attaching multiple devices to >the same serial port bus or "TDM bus", then
>> it's not going to work >anyway. The device must specifically support a "multichannel" or "chain" >mode of some type.
>
>>-Jeff
>
> Thanks Jeff, but I don't understand your good points! You mean
> that attaching multiple devices to the same serial port
> we can get simultaneous samples? If so, How?

You have to find a device that supports multichannel and/or multi-chip operation, and clearly says so in its data
sheet. Suggest to search Cirrus Logic and Analog Devices for "multichannel" and "daisy chain".

-Jeff

Mehdi,

On Oct 10, 2009, at 10:08, s...@gmail.com wrote:
> > Note: One thing I forgot to point out earlier is that a microphone-
> > based system probably will not be susceptible to sample phase
> issues,
> > because the time delays due to the distance between the mic and the
> > sound source will vary and will far outweigh the sample latency. In
> > other words, this is most likely a non-issue. In fact, that may
> > explain why the data sheets do not even mention the timing of the
> > sample process.
> >
> > Brian Willoughby
>
> You are right, but the sample phase issues is so important for me
> because I want to obtain time delay of arrival (TDOA) between 2
> microphones.
In your case, I suggest that you research available ADC chips until
you find one which is capable of sampling 2 channels in sync with
each other. I'm sure there will be a way to interface such a chip to
the C5509, but you must first find the right chip.

I believe that you originally mentioned a 3-channel setup, so if your
3rd channel does not need to be perfectly aligned, you can
potentially use a separate chip. If you need all 3 channels to be
sampled in sync, then you're going to have a hard time finding a chip
already designed for that, and in that case you'll have to design a
more complex analog circuit for sample-and-hold.

> >These are good points. Unless the chip data sheet specifically
> talks about
> >simultaneous sampling performed by multiple devices then sampling
> phase >alignment is likely to be an issue.
> >And if the data sheet doesn't talk about attaching multiple
> devices to >the same serial port bus or "TDM bus", then it's not
> going to work >anyway. The device must specifically support a
> "multichannel" or "chain" >mode of some type.
>
> >-Jeff
>
> Thanks Jeff, but I don't understand your good points! You mean that
> attaching multiple devices to the same serial port we can get
> simultaneous samples? If so, How?
>

No, there's probably no way to communicate to two devices
simultaneously on the same serial port, at least not without severe
hacking.

You could potentially tie the DX pin (e.g. DX0) to the serial input
on two ADC chips which align their sample time to the incoming serial
commands. However, you'd need two separate DR pins (e.g. DR0 and
DR1) in order to obtain the separate sample values from the two chips.

I think that what Jeff is alluding to is that there are several
different ways of addressing an SPI chip. Some chips have a Chip
Select pin which must be active the entire length of a command and
data transfer. Other chips require the Chip Select pin to only be
active for each individual word of the transfer, and thus there would
be multiple select transitions during a multiple word command and
data transfer. Still other chips have an address coded into the
serial command and do not use a Chip Select pin at all - these chips
often have some way to wire each chip differently so that you can
assign a different address to each chip.

The bottom line is that you need to study the data sheet for the
individual serial ADC chip and make sure that you understand how the
chip is selected or addressed. If you don't, and you attach multiple
chips to the same port, then they'll trip over each other and the
data will be corrupted.

My point was that most serial ADC chips use the serial command to
align the sample time. You cannot communicate to multiple devices
simultaneously on a serial port - they must be accessed serially,
thus the name. Therefore you cannot sample simultaneously unless you
use multiple serial ports in parallel. You still have to make sure
that the technique you select is compatible with the addressing and
sample timing scheme for which your chosen ADC chip is designed.

Brian Willoughby

Sound Consulting

Hi friends

> Note: One thing I forgot to point out earlier is that a microphone-
> based system probably will not be susceptible to sample phase issues,
> because the time delays due to the distance between the mic and the
> sound source will vary and will far outweigh the sample latency. In
> other words, this is most likely a non-issue. In fact, that may
> explain why the data sheets do not even mention the timing of the
> sample process.
>
> Brian Willoughby

You are right, but the sample phase issues is so important for me because I want to obtain time delay of arrival (TDOA) between 2 microphones.

>These are good points. Unless the chip data sheet specifically talks about
>simultaneous sampling performed by multiple devices then sampling phase >alignment is likely to be an issue.
>And if the data sheet doesn't talk about attaching multiple devices to >the same serial port bus or "TDM bus", then it's not going to work >anyway. The device must specifically support a "multichannel" or "chain" >mode of some type.

>-Jeff

Thanks Jeff, but I don't understand your good points! You mean that attaching multiple devices to the same serial port we can get simultaneous samples? If so, How?

Regards
Mehdi Hosseini

Brian-

> On Sep 24, 2009, at 12:06, Jeff Brower wrote:
>> I would guess that as long as all three (3) AIC12 (or AIC12K)
>> devices receive the same master clock oscillator input, then they
>> will sample simultaneously.
> The problem is phase. The master clock runs at a much higher rate
> than the sample clock, and thus there are many different points at
> which the chip can sample even if it shares a master clock. Many
> serial ADC chips start the acquisition of a sample on a specific
> event, such as the 4th bit of the serial command word, or on the
> falling edge of the chip select signal. In either of those cases,
> which correspond to chips I have designed with, the sample time for
> each chip will be different because the communication timing is
> different.
>
> In this particular case, with one serial port sharing two or three
> ADC chips, it's most likely that they will NOT sample simultaneously
> unless you design very elaborate circuitry. The DSP will read one
> chip at a time, and as a result only one chip will be sampling at a
> time, and thus none of them will sample simultaneously.

These are good points. Unless the chip data sheet specifically talks about simultaneous sampling performed by
multiple devices then sampling phase alignment is likely to be an issue. And if the data sheet doesn't talk about
attaching multiple devices to the same serial port bus or "TDM bus", then it's not going to work anyway. The device
must specifically support a "multichannel" or "chain" mode of some type.

-Jeff

> Note: One thing I forgot to point out earlier is that a microphone-
> based system probably will not be susceptible to sample phase issues,
> because the time delays due to the distance between the mic and the
> sound source will vary and will far outweigh the sample latency. In
> other words, this is most likely a non-issue. In fact, that may
> explain why the data sheets do not even mention the timing of the
> sample process.
>
> Brian Willoughby
> Sound Consulting

On Sep 24, 2009, at 12:06, Jeff Brower wrote:
> I would guess that as long as all three (3) AIC12 (or AIC12K)
> devices receive the same master clock oscillator input, then they
> will sample simultaneously.
The problem is phase. The master clock runs at a much higher rate
than the sample clock, and thus there are many different points at
which the chip can sample even if it shares a master clock. Many
serial ADC chips start the acquisition of a sample on a specific
event, such as the 4th bit of the serial command word, or on the
falling edge of the chip select signal. In either of those cases,
which correspond to chips I have designed with, the sample time for
each chip will be different because the communication timing is
different.

In this particular case, with one serial port sharing two or three
ADC chips, it's most likely that they will NOT sample simultaneously
unless you design very elaborate circuitry. The DSP will read one
chip at a time, and as a result only one chip will be sampling at a
time, and thus none of them will sample simultaneously.

Note: One thing I forgot to point out earlier is that a microphone-
based system probably will not be susceptible to sample phase issues,
because the time delays due to the distance between the mic and the
sound source will vary and will far outweigh the sample latency. In
other words, this is most likely a non-issue. In fact, that may
explain why the data sheets do not even mention the timing of the
sample process.

Brian Willoughby
Sound Consulting

>If so, then it's probably going to work the way you want.
>But, as Brian mentioned, to be sure you should ask TI guys
>on E2E forum.
>-Jeff

Thanks for introducing this forum to me, I will try it as soon as possible.

--
Seyed Mehdi Hosseini

Mehdi-

>>I am going to use the TLV2544 for reading 3 analog input which come from >3microphone.
>>I want to read each channel periodically and whit specific sampling rate.
>>How can I do this task by using the DMA in C5509?
>>Thanks
>>--
>>Seyed Mehdi Hosseini
>>
> Thanks for your attention Jeff and Brian.
> After more research about ADCs, Codec and Audio ADCs, I decided
> to use an Audio ADC or Codec such as PCM1870 or
> TLV320AIC12 which have microphone bias circuit and internal
> filter. These devices specially designed for audio task.
> According to datasheets we can use 3 TLV320AIC12 in a bus for
> getting signals from 3 microphones, but there is some
> problem! I want to get samples on each microphone in the same
> time, I mean the synchronous sampling. There is no detail in
> datasheet about this.

I would guess that as long as all three (3) AIC12 (or AIC12K) devices receive the same master clock oscillator input,
then they will sample simultaneously. You can check the bus mode timing diagram in the data sheet to verify the
following:

-each device shares the same McBSP signals; i.e.
only one McBSP port is used

-each device should have its own "time slot" on
the McBSP interface. Specifically, this means
a) that in bus mode each device knows which
position it has, maybe using pull-up or pull-down
Rs on a couple of pins, and b) that each device
should have a time period when its serial output
(Tx) line (Rx for the DSP) is in tri-state or
high impedance, so it won't conflict with the
other devices

If so, then it's probably going to work the way you want. But, as Brian mentioned, to be sure you should ask TI guys
on E2E forum.

-Jeff

On Sep 24, 2009, at 09:46, s...@gmail.com wrote:
> After more research about ADCs, Codec and Audio ADCs, I decided to
> use an Audio ADC or Codec such as PCM1870 or TLV320AIC12 which have
> microphone bias circuit and internal filter. These devices
> specially designed for audio task.
The TLV320AIC12 is not recommended for new designs. The TLV320AIC12K
is similar, with a headphone amplifier, but that may be more than you
need.

> According to datasheets we can use 3 TLV320AIC12 in a bus for
> getting signals from 3 microphones, but there is some problem! I
> want to get samples on each microphone in the same time, I mean the
> synchronous sampling. There is no detail in datasheet about this.

The PCM1870 diagram implies that left and right are sampled
synchronously, but the data sheet does not mention how to precisely
control the timing of the sample acquisition. I did not take the
time to look at the TLV320AIC12x data sheets.

These questions are not really related to the C55x. You would
probably be better off signing up for the Texas Instruments Engineer
to Engineer forum at http://e2e.ti.com/ where they have forums for
analog-to-digital conversion.

Brian Willoughby
Sound Consulting

>I am going to use the TLV2544 for reading 3 analog input which come from >3microphone.
>I want to read each channel periodically and whit specific sampling rate.
>How can I do this task by using the DMA in C5509?
>Thanks
>--
>Seyed Mehdi Hosseini
>
Thanks for your attention Jeff and Brian.
After more research about ADCs, Codec and Audio ADCs, I decided to use an Audio ADC or Codec such as PCM1870 or TLV320AIC12 which have microphone bias circuit and internal filter. These devices specially designed for audio task.
According to datasheets we can use 3 TLV320AIC12 in a bus for getting signals from 3 microphones, but there is some problem! I want to get samples on each microphone in the same time, I mean the synchronous sampling. There is no detail in datasheet about this.
How can I do this?
thank you.
--------------
Mehdi Hosseini