DSPRelated.com
Forums
Online · Live + on demand
February 9-11, 2027
Price goes up Jan 1: register now to lock in today's rate
Register Now →

What kind of analog-to-digital converters are used in cell phones?

Started by Jerry ●February 1, 2013
On 2/5/13 11:04 PM, radams2000@gmail.com wrote:
> > This is an example of a so-called incremental sigma-delta converter. There is an internal oscillator that runs a minimum of 2100 internal cycles before an output is available. The digital filter is completely reset after each conversion. So this is not a "normal" sigma-delta. When they use the term single-cycle settling they are referring to the fact that from an external user point of view, it appears that you issue a start-convert pulse and you get an immediate answer; the fact that the internal clock runs for thousands of cycles is hidden from the user.
...
> > The theory of Incremental sigma-delta converters is pretty cool if you're into those kinds of things :)
seems sorta wasteful. ditch all of that information just so you can MUX the input? -- r b-j rbj@audioimagination.com "Imagination is more important than knowledge."
On Tue, 5 Feb 2013 20:04:24 -0800 (PST), radams2000@gmail.com wrote:

>Vlad > >This is an example of a so-called incremental sigma-delta converter. There = >is an internal oscillator that runs a minimum of 2100 internal cycles befor= >e an output is available. The digital filter is completely reset after each= > conversion. So this is not a "normal" sigma-delta. When they use the term = >single-cycle settling they are referring to the fact that from an external = >user point of view, it appears that you issue a start-convert pulse and you= > get an immediate answer; the fact that the internal clock runs for thousan= >ds of cycles is hidden from the user. > >Even in a normal sigma-delta, you can design the decimation filter to optim= >ize settling father than brick-wall anti-aliasing, which improves the rate = >at which you could mux the converter.=20 > >The theory of Incremental signa-delta converters is pretty cool if you're i= >nto those kinds of things :)
Kinky.
> > >Bob
Eric Jacobsen Anchor Hill Communications http://www.anchorhill.com
Robert

Compared to the successive approx method of data-acq converters, incremental converters are not very cycle-efficient, but if you don't need to go very fast then the benefit is that you get freedom from matching requirements.

It's funny, in the early days of sigma-delta, I would always run into the data-acq guys who couldn't see how sigma-delta could give you a solid answer, and I would wave my hands around and mumble something about how they needed to start thinking in the frequency domain.  But now I think you can actually totally analyze sigma-delta as a time-domain converging algorithm. It's just that you are converging on a sequence of input samples, and not just a single sample. If you happen to throw a sample-and-hold on the input then those samples are all the same, so it looks like a data-acq converter, assuming that you reset everything before each conversion, including the digital filter.
I need to stop myself now before I reveal my unhealthy obsession with this topic :)

Bob
On Wed, 6 Feb 2013 02:17:01 +0000 (UTC), glen herrmannsfeldt
<gah@ugcs.caltech.edu> wrote:

>Tim Wescott <tim@seemywebsite.please> wrote: > >(snip) > >> Sigma-delta ADCs don't multiplex well, so it's doubtful that >> the audio ADCs are multiplexed. > >Aren't the ADCs fast enough, relative to the sampling rate, >that you can multiplex with other slow inputs? (Or both channels >of a stereo source.) > >I remember stories from near the beginning of CD days that some >players would multiplex the DAC between two channels, so that there >was a half sample delay between them. I suppose you notice if you >wire the two channels together to get mono, but likely not in an >actual stereo room. Even with headphones it should be hard to tell.
Those DACs in earlier CD players were from a time before sigma-delta became common in DACs and ADCs. Those were simpler DACs (perhaps using R-2R ladders, I forget) that DID multiplex well. They weren't sigma-delta.
On Tuesday, February 5, 2013 8:19:12 PM UTC-7, Vladimir Vassilevsky wrote:
> "Tim Wescott" wrote: > > Sigma-delta ADCs don't multiplex well > > This is a prejudice. > Look at ADS1248. Could be multiplexed at every sample. > > Vladimir Vassilevsky
I took a quick look and this device is for measuring temperatures up to 2kSPS. So what others have said about slowness due to flushing the filter etc. is consistent with this being a slow device. Jerry