LTC2484 External Caps

LTC2484 datasheet figures 15 through 18 suggest large CREF decoupling capacitors cause gain error.  I am puzzled by this.  I would expect the opposite.  The second paragraph on page 28 is confusing.  The first sentence almost seems to suggest that the external CREF is part of the sampling capacitance (internal 12pF || CREF ?).  The second sentence says "such" (small ?) values of CREF deteriorate offset and gain performance.  But that contradicts the figures.  Can you please clarify this.  What am I missing here?  TIA

  • 0
    •  Analog Employees 
    on Mar 19, 2019 7:49 PM

    Hi Andrew,

    Figures 15-18 deal with the effect of source impedance of the voltage source. See Figure 12 of the data sheet. Because the analog inputs and reference inputs are part of a switched capacitor (see Figure 11of the data sheet) network if there is any source impedance on these inputs and sufficient time is not allowed for settling errors can result.

    The second paragraph on page 28 is saying what Figures 15-18 show graphically. Figure 11 shows that the bypass caps are in parallel with the sampling capacitor.

  • Figures 15-18 also deal with the effect of CREF, which is the subject of my question.  The last sentence of the second paragraph contradicts the graphs.  I think this sentence is misplaced.  It would make more sense at the end of the third paragraph, since large CREF deteriorates FS accuracy according to the graphs, and also in my opinion.  I have an LTSpice sim which illustrates this, which I could share.  Sampling causes a dip in the voltage on the VREF pin.  It recovers quickly and fully, if CREF is small; however, if CREF is large, it smooths-out the current pulses (as mentioned in the 3rd para) producing a DC drop across Rsource.

  • 0
    •  Analog Employees 
    on Mar 20, 2019 5:17 PM in reply to AndrewHolme

    I guess it comes down to how you interpret the phrase " Such values for CREF". Since it has just been stated that small values (<1nF) result in only small errors I interpreted it to mean larger values. I understand that it could also be interpreted to mean the small values that were previously mentioned but as you state that doesn't make sense.

  • Datasheet improvement suggestion:

    Replace "Such" with "Large"