Same value Parallel Capacitors in regulator

Discussion in 'General Electronics Chat' started by beethovenj, Apr 19, 2014.

  1. beethovenj

    Thread Starter New Member

    Mar 28, 2014
    5
    0
    Hi to all!

    I found similar posts but not one explaining the doubt I have, so there it is:

    I am working with a circuit that has a high stability +3.3V Vref.

    The circuit is the following:

    +3.3V input, the a series BLM41PG102SH, to a 22u and 4x 22n to the input of a ADP3338, its output to a 22u to a seriesBLM18PG121 to 8x 22n and 2x 4.7u.

    All caps to ground.

    How same value caps help in getting high stability?

    Thanks so much!
     
  2. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    7,386
    1,605
    Well, you have to understand the PDQ relies on the NPT of the Q17 in accordance with LSMFT.

    Now, would you please try your post again? A schematic is the best way describe a circuit. You can link to the one you are looking at. I don't think you can post picture yet (you need 9 more posts), a mod could help but a link gets us there right now.

    Otherwise I have no idea what you mean.

    And welcome to the forums!
     
  3. beethovenj

    Thread Starter New Member

    Mar 28, 2014
    5
    0
    Hi,

    Yup, it wasn't very clear, was it? Sorry about that.

    In fact the circuit itself is not very relevant, it is only relevant in the sense that the doubt comes from a real circuit.

    Such circuit has:
    - a three pin regulator
    (and as by-pass capacitor uses:)
    - in the input: 4x 22n plus a 22u,
    - in its output uses 8x 22n plus two 47u

    My question is why is good to use several parallel capacitors. The circuit is no RF but it works with fast signals.

    I hope it is a bit more clear now... not brilliant for a first post! Sorry about that again!



    How
     
  4. beethovenj

    Thread Starter New Member

    Mar 28, 2014
    5
    0
    (I didn't intend to put the smiley! I just wrote the the colon and brackets... but I guess a unintended smiley is not such a bad thing!)
     
  5. AnalogKid

    Distinguished Member

    Aug 1, 2013
    4,520
    1,247
    I see two things frequently on schematics. One is clustering all of the IC decoupling capacitors in one place rather than have them clutter up the signal parts of the drawing. The other is paralleling multiple small capacitors to get a very low equivalent series resistance (ESR) and/or increased ripple current capability. This usually is done on the input and output of switching regulator circuits. Are you sure your 3-pin regulator is a linear part?

    ak
     
    #12 likes this.
  6. beethovenj

    Thread Starter New Member

    Mar 28, 2014
    5
    0
    Hi AnalogKid,

    Thanks so much for the reply.

    The capacitors are all connected directly to the regulator, no to other ICs.

    I am positive about the regulator, it is a LDO. The input voltage comes from a switching power supply, and then comes the LDO. The output of the LDO is used to power some ICs as well as a high stability voltage reference for a DAC that configures a smaller Vref for a comparator.

    For the ESR reduction I would use a single small cap, not eight of them! I'll think better about it to understand it... thanks!
     
  7. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    7,386
    1,605
    OK, now I see the circuit: series inductor, several caps to ground, 3 terminal regulator (low dropout linear), series inductor, several caps to ground.

    I would "guess" Analog Kid got it right when he suggests ALL the bypass caps on the board are drawn in one place. It's a common shorthand.

    If you check out the data sheet for the ADP3338 it will tell you what "bulk" caps are needed to keep it stable. Many regulators need some (or lots) cap to keep them stable. I was quite surprised to see how noisy a reference regulator could be without lots of cap. So caps are definitely required for low noise.

    It is also common to use a large (>= 1uF) cap together with a small (0.1-0.01) and also occasionally a tiny (< 0.001) cap when the best filtering is needed. The larger the cap the lower frequency it works to, similar to music speakers where you have a bass speaker, midrange and tweeter speakers to fill the spectrum.

    Large caps give backup for large slow changes in load, the medium caps backup for medium changes, and so on.

    A typical arrangement is one bulk cap at the regulator, and one medium (0.01-0.1) cap at each device.
     
  8. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    7,386
    1,605
    And that is why it is so highly bypassed: getting noise out so it can also serve as a reference.

    I've seen it done both ways. If you can find a single cap at a price you can live with, you use it. Sometimes that cap is so costly that a multiple of cheaper caps do the same job at 1/10th the price.
     
    #12 likes this.
  9. #12

    Expert

    Nov 30, 2010
    16,261
    6,768
    Excellent, AK. I looked at this yesterday and did not remember the convention of drawing all the capacitors in the wrong place...so I didn't answer.
    Still, that doesn't apply here so we're going to have to go with Ernie from Lon Guyland.
     
  10. alfacliff

    Well-Known Member

    Dec 13, 2013
    2,449
    428
    another reason for multple smaller caps in paralell is lower inductance, but this being a linear regulator, that would not be so important here.
     
  11. beethovenj

    Thread Starter New Member

    Mar 28, 2014
    5
    0
    Thanks to all of you! Very helpful.

    Is it not risky anyway to use a power supply voltage (use to power some fast switching digital IC) to also act as a high stability reference voltage?

    I suppose, as commented, that is why you get all those caps...
     
  12. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    7,386
    1,605
    As always, the correct answer is "it depends."

    I've worked on a few stand alone devices that need to make an accurate measurement and I've found that by measuring two points and taking the difference that the absolute value of the reference voltage drops out of the equation. That assumption holds as long as the short term stability (say under 1/10th a sec) is pretty good. It can change from day to day but as long as 2 successive measurements are the same all works nicely.

    So no gold plated .001% reference was required. Just the sloppy supply voltage was quite adequate in those specific cases.
     
Loading...