AD8436

Rev. 0 | Page 12 of 20

APPLICATIONS INFORMATION

USING THE AD8436

This section describes the power supply and feature options,

as well as the function and selection of averaging and filter

capacitor values. Averaging and filtering options are shown

graphically and apply to all circuit configurations.

Averaging Capacitor Considerations—RMS Accuracy

Typical AD8436 applications require only a single external

capacitor (CAVG) connected to the CAVG pin (see Figure 30).

The function of the averaging capacitor is to compute the mean

(that is, average value) of the sum of the squares. Averaging

(that is, integration) follows the absolute value circuit, where

the polarity of negative input current components is reversed

(rectified) prior to squaring. The mean value is the average

value of the squared input voltage over several input waveform

periods. The rms error is directly affected by the number of

periods averaged, as is the resultant peak-to-peak ripple.

The result of the conversion process is a dc component and a

ripple component whose frequency is twice that of the input. The

rms conversion accuracy depends on the value of CAVG, so the

value selected need only be large enough to average enough periods

at the lowest frequency of interest to yield the required rms

accuracy. Figure 27 is a plot of rms error vs. frequency for various

averaging capacitor values. For Figure 27, the additional error

was 0.001% at 40 Hz using a 10 μF metalized polyester capacitor.

Larger values yield diminished returns because the settling time

increases with negligible improvement in rms accuracy.

To use Figure 27, determine the minimum operating frequency

and accuracy of the application and then find the suggested

capacitor value on the chart. For example, for –0.5% rms at 100 Hz,

the capacitor value is 1 μF.

Post Conversion Ripple Reduction Filter

Input rectification included in the AD8436 introduces a

residual ripple component that is dependent on the value

of CAVG and twice the input signal frequency. For sampling

applications such as a high resolution ADC, the ripple component

may cause one or more LSBs to cycle, and low value display

numerals to flash.

Ripple is reduced by increasing the value of the averaging capacitor,

or by postconversion filtering. Ripple reduction following

conversion is far more efficient because the ripple average value

has been converted to its rms value. Capacitor values for post-

conversion filtering are significantly less than the equivalent

averaging capacitor value for the same level of ripple reduction.

This approach requires only a single capacitor connected to the

OUT pin (see Figure 25). The capacitor value correlates to the

simple frequency relation of ½ π R-C, where R is fixed at 16 kΩ.

OUT

16kΩ

OGND

CORE

CLPF

DC OUTPUT

9

8

Figure 25. Simple One-Pole Post Conversion Filter

As seen in Figure 26, CAVG alone determines the rms error,

and CLPF serves purely to reduce ripple. Figure 26 shows a

constant rms error for CLPF values of 0.33 μF and 3.3 μF; only

the ripple is affected.

FREQUENCY (Hz)

1

0

100

1k

–1

–2

10

–3

–4

–5

–6

–7

–8

–9

–10

CAVG = 10µF

CLPF = 0.33µF OR 3.3µF

CAVG = 1µF

CLPF = 0.33µF OR 3.3µF

Figure 26. RMS Error vs. Frequency for Two Values of CAVG and CLPF

(Compare the effects of CAVG and CLPF, and

note that CLPF does not affect rms error result.)

1k

FREQUENCY (Hz)

100

–0.5

–1.5

0

–1.0

10

–2.0

47µF

10µF

CAVG = 0.22µF

1µF

2.2µF

0.47µF

4.7µF

22µF

Figure 27. Conversion Error vs. Frequency for Various Values of CAVG