>Remember that when you are comparing SAR versus sigma-delta converters you
=
>need to add the delay of external analog anti-alias filters required for
th=
>e SAR converter to the calculation, unless the signal itself is naturally
b=
>and-limited to the point where you can convince yourself you don't need
any=
> more filtering.=20
>
>Bob
>
Good point, Bob. Thanks for that!
Can anyone recommend a well supported development board with really low
delay?
More than two channels in and out would be good as well...
Thanks!
T.
Reply by Robert Adams●November 28, 20122012-11-28
Remember that when you are comparing SAR versus sigma-delta converters you need to add the delay of external analog anti-alias filters required for the SAR converter to the calculation, unless the signal itself is naturally band-limited to the point where you can convince yourself you don't need any more filtering.
Bob
Reply by Thomas6666●November 27, 20122012-11-27
Hi Steve,
Thank you for your opinion on this.
>>>> Does anyone know a device with a fast SAR ADC (that could be used for
>>>> audio
>>>> processing)? A board with suiting anti-aliasing filters, floating
point
>
>>>> DSP
>>>> and fast DAC would be perfect.
>>>> Have you got an idea, what the minimum expected delay could be?
>>
>>> And if you're using it for audio processing, why do you care so much
>>> about delay?
>>
>>There are applications like adaptive noise canceling where processing
>delay
>>is crucial.
>
>True, but is the delay through most modern sigma-delta converters
>intolerable? Two or three sample delay converters are widely available.
>
>>> Randy Yates
>>Lamer.
>Steve
Unfortunately every extra sample hurts and reduces the performance of the
algorithm.
Which eval board would you suggest, that has a low overall delay (from
analog in to analog out) with Sigma-Delta converters? Easiest would
probably be a common one that is well supported (I made some bad experience
with boards with really bad documentation).
Thank you in advance!
T.
Reply by steveu●October 7, 20122012-10-07
>
>"Randy Yates" <yates@digitalsignallabs.com> wrote in message
>news:874nmor0n3.fsf@randy.site...
>> "Thomas6666" <61878@dsprelated> writes:
>>
>>> it's been a while ago since someone asked here for an DSP eval board
(for
>>> audio processing) that's got a SAR analog-digital converter.
>>> Theoretical it should be possible to reach very low delay between an
>>> audio
>>> input and the output after processing by a DSP.
>>> Does anyone know a device with a fast SAR ADC (that could be used for
>>> audio
>>> processing)? A board with suiting anti-aliasing filters, floating point
>>> DSP
>>> and fast DAC would be perfect.
>>> Have you got an idea, what the minimum expected delay could be?
>
>> If you're concerned about delay in conversion, why not consider flash
>> converters?
>
>Find a flash converter with more then 12 bits of resolution.
Flash converters have mostly given way to several flavours of pipelined
converters, which break the high resolution problem down into a number of
coarser resolution steps.
>> I don't know of a lot of successive-approximation converters
>> still in use.
>
>Most of MHz class sampling A/Ds are charge redistribution SARs.
Yep, SAR is still *the* mainstream technique for conversion, and most
shuffle charge in one way or another.
>> And if you're using it for audio processing, why do you care so much
>> about delay?
>
>There are applications like adaptive noise canceling where processing
delay
>is crucial.
True, but is the delay through most modern sigma-delta converters
intolerable? Two or three sample delay converters are widely available.
>> --
>> Randy Yates
>
>Lamer.
Steve
Reply by Thomas6666●October 6, 20122012-10-06
>> If you're concerned about delay in conversion, why not consider flash
>> converters?
>
>Find a flash converter with more then 12 bits of resolution.
>
>> I don't know of a lot of successive-approximation converters
>> still in use.
>
>Most of MHz class sampling A/Ds are charge redistribution SARs.
>
>> And if you're using it for audio processing, why do you care so much
>> about delay?
>
>There are applications like adaptive noise canceling where processing
delay
>is crucial.
>
>> Randy Yates
>
>Lamer.
Thanks Vladimir "Lamer" Vassilevsky. That are the points.
And that's why I expected that there should be such a device somewhere.
Some audio processing algorithms need low delay and high resolution.
The algorithm that I would like to implement could be implemented in such a
way, that it processes sample by sample.
T.
Reply by Vladimir Vassilevsky●September 24, 20122012-09-24
"Randy Yates" <yates@digitalsignallabs.com> wrote in message
news:874nmor0n3.fsf@randy.site...
> "Thomas6666" <61878@dsprelated> writes:
>
>> it's been a while ago since someone asked here for an DSP eval board (for
>> audio processing) that's got a SAR analog-digital converter.
>> Theoretical it should be possible to reach very low delay between an
>> audio
>> input and the output after processing by a DSP.
>> Does anyone know a device with a fast SAR ADC (that could be used for
>> audio
>> processing)? A board with suiting anti-aliasing filters, floating point
>> DSP
>> and fast DAC would be perfect.
>> Have you got an idea, what the minimum expected delay could be?
> If you're concerned about delay in conversion, why not consider flash
> converters?
Find a flash converter with more then 12 bits of resolution.
> I don't know of a lot of successive-approximation converters
> still in use.
Most of MHz class sampling A/Ds are charge redistribution SARs.
> And if you're using it for audio processing, why do you care so much
> about delay?
There are applications like adaptive noise canceling where processing delay
is crucial.
> --
> Randy Yates
Lamer.
Reply by Randy Yates●September 23, 20122012-09-23
"Thomas6666" <61878@dsprelated> writes:
> Hello everyone,
>
> it's been a while ago since someone asked here for an DSP eval board (for
> audio processing) that's got a SAR analog-digital converter.
> Theoretical it should be possible to reach very low delay between an audio
> input and the output after processing by a DSP.
> Does anyone know a device with a fast SAR ADC (that could be used for audio
> processing)? A board with suiting anti-aliasing filters, floating point DSP
> and fast DAC would be perfect.
> Have you got an idea, what the minimum expected delay could be?
Hi Thomas,
If you're concerned about delay in conversion, why not consider flash
converters? I don't know of a lot of successive-approximation converters
still in use.
And if you're using it for audio processing, why do you care so much
about delay?
--
Randy Yates
Digital Signal Labs
http://www.digitalsignallabs.com
Reply by rickman●September 21, 20122012-09-21
On 9/21/2012 11:16 AM, Thomas6666 wrote:
> Hello,
>
> Thanks to everyone for your replies.
> Sorry for my late response. I've been on a vacation and had limited
> internet access...
>
> The ADAU1772 does not work for this idea. I'd like to implement an adaptive
> algorithm. That's why I am looking for a programmable device.
>
> Al Clark's module might be an option - depending on price and needed effort
> to make it work ;-)
>
> Unfortunately I am not that experienced in DSP programming. An
> out-of-the-box solution gorgeous...
What problem are you looking for a solution to? That is pretty
important if you are looking for a solution out of the box.
As to your original question about micros doing DSP, yes, there are some
micros that are suitable for DSP. I believe that ARM considers the
Cortex M4 to approach DSP speeds of 1 MAC per clock (or maybe per 2
clocks, not sure) once all the optimizations are working. It has been
awhile since I saw this presentation, but I believe they claimed you
could do this from standard C code without special libraries or compiler
features.
When you talk about a low delay between input and output, much of that
depends on your algorithm. If you need to accumulate a block of data to
process, you can't even start the processing until the block is
complete. But if you are using an algorithm that processes data on the
fly then you have a decent shot. Remember that analog anti-alias
filters can also add delay.
Rick
Reply by Thomas6666●September 21, 20122012-09-21
Hello,
Thanks to everyone for your replies.
Sorry for my late response. I've been on a vacation and had limited
internet access...
The ADAU1772 does not work for this idea. I'd like to implement an adaptive
algorithm. That's why I am looking for a programmable device.
Al Clark's module might be an option - depending on price and needed effort
to make it work ;-)
Unfortunately I am not that experienced in DSP programming. An
out-of-the-box solution gorgeous...
T.