Connect with us

ADC, Do I need a sample and hold chip?

Discussion in 'Electronic Design' started by [email protected], Sep 22, 2007.

Scroll to continue with content
  1. Guest

    Dear all,

    I have a question about the ADC.

    If I use a ADC chip and a MUX etc., and the ADC has the internal MUX
    and internal sample and hold, do I still need a sample and hold after
    the MUX before the ADC?

    Many thanks,

  2. Eeyore

    Eeyore Guest

    Depends on various things. Even if it's a succesive approximation converter you
    can calculate if it needs a s/h by examining the maximum input slew rate. I've
    managed without one.

  3. Guest

    Thanks, Graham.
    Can I say if I put a S/H, it will be safer?
    Could you tell me how do you calculate it?

  4. John Larkin

    John Larkin Guest

    No; it already has a s/h, right? But anyhow, if the chip has an
    integrated mux and s/h, how could you get between them to add another

    The two common integrated adc architectures, these days, are
    delta-sigma and charge-balance successive approximation, and neither
    neeeds an external s/h.

    A little anti-aliasing filtering is always prodent.

  5. Rich Grise

    Rich Grise Guest

    If "the ADC has the internal MUX...", then why do you need a separate
    MUX at all?

  6. Guest

    Thanks for your response.
    The ADC is not able to give enough channels and I need to condition
    the signal before the ADC. The maximum ADC channel I can have for one
    ADC is 32 single end channels. But there are 24 differential channels
    to sample.
    To save the cost and the number of componennts, I have to use several
    multiplexers in the front.

  7. Eeyore

    Eeyore Guest

    If it doesn't need a s/h in the first place it doesn't need to be 'safer'.

    What's your application ? What are you measuring. What chip are you using ?

  8. Guest

    The application is to measure DC voltages. I am going to use the
    ADG509F (Analog Device) - MUX, and ADS8326 (TI) - ADC.
    There will be a low pass filter before the ADS8326.

  9. Eeyore

    Eeyore Guest

    What frequency LPF ? That virtually assures me that you won't need a s/h from first
    principles since the filter will limit the rate at which the input voltage can change.

  10. Guest

    Not sure yet. For DC voltage measurement, what freq. do you think is
    right? Probably 30Hz, which will filter out the mains 50 or 60Hz.
  11. The low pass filters should go _before_ the mux. The signal path
    between mux and ADC has to pass the mux switching frequency - if you
    put a LP filter there, you have to scan slow enough that the filter
    will settle between readings, which would mean, with a 30 Hz LP, the
    switching frequency of the mux will have to be significantly below 30
    Hz to get any sensible readings.

    Peter Bennett, VE7CEI
    peterbb4 (at)
    new newsgroup users info :
    GPS and NMEA info:
    Vancouver Power Squadron:
  12. Rich Grise

    Rich Grise Guest

    As long as your MUX can get the signal to the ADC in time (i.e., before
    the sample window) and is stable enough for the sample interval that it
    doesn't disturb the reading, then I'd say, you probably don't need another

    You'd switch your MUX to the channel in question, tell the ADC to make
    the reading, and _its_ S/H should operate normally.

    So, no, you shouldn't need an additional sample & hold.

    Good Luck!
  13. Eeyore

    Eeyore Guest

    If you want to filter out 50 or 60 Hz you'll need a very high order filter @ 30Hz. It all
    depends on the characteristics of what you're measuring of course.

    It seems to me that you're going at this blind. You ought to have a better idea of your
    actual requirements.

  14. Eeyore

    Eeyore Guest

    That's very true.

    Condition the signals and then mux them.

  15. Hal Murray

    Hal Murray Guest

    Putting the filter after the mux avoids duplicating the filters.

    It works find as long as your sample rate is slow enough.
  16. Guest

    Thanks for all your answers.
    I will probably put the 30Hz filter before the MUX, since the RC
    filters aren't going to add too much on the budget. Apart from the
    cost, the other reason is the MUX has to switch quickly enough to get
    all 24 channels samples at least less than 0.05 sec.

  17. YD

    YD Guest

    Late at night, by candle light, penned this immortal
    Looks like you might want to use diff to single amps on the inputs.
    Then you get to use them as filters/conditioners as well. Depends on
    what kind of overhead you can tolerate, 24 of them do take up a bit of
    real estate and spare change. It's always nice to have an extra layer
    between the DAC and the cruel, cold outside world.

    - YD.
  18. On Sat, 22 Sep 2007 19:31:51 -0500,
    True, but for a 30 Hz low pass filter, the scan rate would have to be
    Bloody Slow - a half second per channel might work. And the OP now
    says he wants to scan all 24 channels in 50 mS.

    Peter Bennett, VE7CEI
    peterbb4 (at)
    new newsgroup users info :
    GPS and NMEA info:
    Vancouver Power Squadron:
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