Connect with us

Sound chips that natively go to DC?

Discussion in 'Electronic Design' started by Joerg, Jan 1, 2014.

Scroll to continue with content
  1. Joerg

    Joerg Guest


    The subject pretty much says it. Are there any that can at least be
    coaxed into DC input and output by disabling internal cancelers and
    such? Should not be too drifty. The digital interface doesn't matter
    much, USB, SPI, whatever. Preferably not I2C but even that would be ok.

    16 bit would suffice. Of course, more is better. Input to output
    synchronicity is important, as usual. Also synchronicity between the
    stereo input channels.
  2. The all do dc, you need to jump the cap inside..

    as for output, that may not be such a good idea because many of
    them do not operate dual rail supplies, they have dc isolation

    Best thing to do is what I did years ago., AM in AM out. write the
    software to decode the modulation on the input and modulate AM for the
    output. You need to make an interface board, but that isn't really that

    I did it via a 324 chip, one unit for the CW oscillator, a unit for
    the modulator, one input from the OSC and the other input from your
    reference and finally, an AM detector that uses a third unit with a
    integrator in the feed back and that forms your detection. The last
    unit can be used as a unity stage.

  3. Guest

    Do you mean PC sound cards?
    I doubt that any sound cards use dual rails. In any case, these will
    have lousy DC characteristics (offset). It would take some work to
    straighten that out.
    If you're going to go that far, why PWM from a digital port, of some
  4. Not exactly sure what you mean there? are you saying creating a custom
    digital port (USB for example) ? If so, that would me you need software
    drivers. Something that gets messy..

    Recently I was experimenting with inserting 2 FM carriers via a simple
    board interface. Using both channels.

    Looks interesting any ..

  5. Joerg

    Joerg Guest

    T'is analog somewhere's in there.
  6. Joerg

    Joerg Guest

    I found that to be problematic in the ones I looked at, some sort of
    DC-scrubber circuit that couldn't be disabled.

    That would be no problem at all.

    I was contemplating the old HC4066 trick for chopping and synchronous
    rectification but it does get old.
  7. Joerg

    Joerg Guest

    Almost, I mean the chips on there.

    That would be ok, I can take care of that. As long as the offset is very
    stable and determined, like a bandgap or two.

    Well, that jinxes it :-(

    Sure, and I could also just use regular ADCs and DACs on another port.
    But it makes a mess of things. Lots more parts, muxing (signals go via
    the same connector), and most of all now the processor must have this on
    a high priority interrupt because unlike sound chips there's no buffer.
    Cost is not a big concern on this project but simplicity and size is.
  8. many of them have a small DC current put at the mic input that allows
    you to power up a microphones

  9. Joerg

    Joerg Guest

    Often only on one input and I can deal with that. But if, as Keith
    hinted, the DC characteristics in general are lousy then I can't use
    sound chips anyhow.
  10. Nico Coesel

    Nico Coesel Guest

    Regular sound chips (codecs) have no buffer either. The buffering
    occurs in the digital interface which is part of the host chip. SPI
    and I2C are likely not to work. You'd need to look for a typical
    DSP-style codec interface like AC97, I2S, etc.

    Removing the DC blocking capacitors should make a codec go to DC but
    you'll need some differential amplifiers to make the codec deal with
    ground centered signals. If you are lucky the codec outputs a
    reference voltage which is used to reference the signals.
  11. Joerg

    Joerg Guest

    Goes in: Signals from DC to 10kHz, amplitude up to 3Vpp but this gets
    divided down. Input swings around ground so in single-supply chips
    there'd have to be an external DC-shifter (no big deal).

    Goes out: Signal of very same frequency and phase as goes it but slowly
    amplitude modulated. Amplitude range much lower, generally a few
    millivolts. DC stability must be impeccable. But siganl can be divided
    down from FS in order to ease that requirement a bit. Modulation BW less
    than 50 Hoitz and modulation depth no more than 50%.

    Synchronicity between in- and output is critical and ideally the
    processor should not have to take care of this. Sound chips have buffers
    to deal with USB transmission gaps and the like.
  12. Joerg

    Joerg Guest

    They usually do. For example, right now one from this family is in there:

    Quote "When receiving the audio data, the PCM2900C/2902C stores the
    first audio packet, which contains 1-ms audio data, into the internal
    storage buffer. The PCM2900C/2902C starts to play the audio data when
    detecting the next start of frame (SOF) packet, as illustrated in Figure

    Unfortunately it has the usual in there, poorly documented digital
    high-pass filters.

    It usually doesn't because they put digital HPFs in there.
  13. Nico Coesel

    Nico Coesel Guest

    SOF comes from USB. So yes a USB codec needs some buffering. But a
    codec intended to be connected to a SoC or DSP doesn't have buffering.
    At least I never came across one that did.
    I've Never seen one with an HPF. Its not necessary because they are
    intended to be used with DC blocking capacitors. A HPF would impair
    the boom-boom sound.
  14. boB

    boB Guest

    Most (if not all) of the Delta-Sigma A/D converters for audio that I
    have seen and used have a HPF built in but allow you to bypass them.
    For instance, this one from TI, similar to the earlier link posted but
    this data sheet shows it in the block diagram...

    I'm pretty sure that the reason they put the HPF in there is because
    their DC accuracy or drift is not very good. I have wanted to use one
    of these regular old audio ADCs for DC before but I have gone to
    slower delta-sigma converters that are optimized for DC.

    I think that the old Crystal semiconductor and AKM delta-sigmas had
    some kind of HPF that could not be defeated but can't remember for
    sure now. The newer ones I think can be bypassed for DC operation.

    One of the converters I use for DC does a DC offset calibration
    in-between every conversion which works great. Probably not so easy
    to do with a regular audio ADC but I'd love to be able to make one
    work or hear how well an audio ADC works for someone at DC.

  15. Interesting chip. that PCM1804. my Dell XPS has the same options
    on it for PCM/Digitial with the same sample rates and bits.

    I'll have to check my dell next time I boot it to see what the sound
    chip set is.

  16. boB

    boB Guest

    My new (ish) dell laptop uses an IDT codec and is a company I am not
    familiar with. I think I found a block diagram of it somewhere but I
    find that it is usually very difficult to find real documents for the
    high volume chipsets that computer makers use these days.
    This is especially true for Broadcom chips where they do not want the
    everyday joe to be able to program them... Graphics chips too like
    Nvidia. Not like it used to be where you could just get the document
    for a chip and be able to program it. That's why I stick with smaller
    embedded microcontroller applications where I know what is going on

    Happy new Year from the yechy cold and rainy pacific northwest !

  17. Joerg

    Joerg Guest

    Therein lies the main problem. Even If I found a trick to disable the
    HPF the DC specs are either not properly given at all or they are very
    dismal. Like with the internal "reference" in some ATMegas where I
    seriously wondered how someone could possibly botch an analog circuit so

    An alternative would be some other AD/DA device that isn't called sound
    chip but relieves the processor of having to baby its timing.

    Hey, send us some of that rain. We are dying for it here, seriously.
  18. boB

    boB Guest

    I hear you ! I do not use Atmel products anymore unless I absolutely
    have to after they almost put our last company out of business (early
    ATmega32's) Didn't know they had a problem with V-references though.

    So, one way to use an audio board for DC but with questionable upper
    bandwidth is to use an (external) FM modulator. I think there was a
    thread about this just recently here on SED ?

    I now see why the elderly migrate south for the winter.
    Yes, I'll trade you ! For a while anyway.
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day