LTC1563-3_15 Datasheet, PDF (10/20 Pages) Linear Technology – Active RC, 4th Order Lowpass Filter Family

LTC1563-2/LTC1563-3
APPLICATIONS INFORMATION
Functional Description
The LTC1563-2/LTC1563-3 are a family of easy-to-use,
4th order lowpass filters with rail-to-rail operation. The
LTC1563-2, with a single resistor value, gives a unity-gain
filter approximating a Butterworth response. The
LTC1563-3, with a single resistor value, gives a unity-gain
filter approximating a Bessel (linear phase) response. The
proprietary architecture of these parts allows for a simple
unity-gain resistor calculation:
R = 10k(256kHz/f_C)
where f_C is the desired cutoff frequency. For many appli-
cations, this formula is all that is needed to design a filter.
For example, a 50kHz filter requires a 51.2k resistor. In
practice, a 51.1k resistor would be used as this is the
closest E96, 1% value available.
The LTC1563-X is constructed with two 2nd order sec-
tions. The output of the first section (section A) is simply
fed into the second section (section B). Note that section
A and section B are similar, but not identical. The parts are
designed to be simple and easy to use.
By simply utilizing different valued resistors, gain, other
transfer functions and higher cutoff frequencies are
achieved. For these applications, the resistor value calcu-
lation gets more difficult. The tables of formulas provided
later in this section make this task much easier. For best
results, design these filters using FilterCAD Version 3.0 (or
newer) or contact the Linear Technology Filter Applica-
tions group for assistance.
Cutoff Frequency (f_C) and Gain Limitations
The LTC563-X has both a maximum f_C limit and a mini-
mum f_C limit. The maximum f_C limit (256kHz in High Speed
mode and 25.6kHz in the Low Power mode) is set by the
speed of the LTC1563-X’s op amps. At the maximum f_C,
the gain is also limited to unity.
A minimum f_C is dictated by the practical limitation of
reliably obtaining large valued, precision resistors. As the
desired f_C decreases, the resistor value required increases.
When f_C is 256Hz, the resistors are 10M. Obtaining a
reliable, precise 10M resistance between two points on a
printed circuit board is somewhat difficult. For example, a
10M resistor with only 200MΩ of stray, layout related
10
resistance in parallel, yields a net effective resistance of
9.52M and an error of – 5%. Note that the gain is also
limited to unity at the minimum f_C.
At intermediate f_C, the gain is limited by one of the two
reasons discussed above. For best results, design filters
with gain using FilterCAD Version 3 (or newer) or contact
the Linear Technology Filter Applications Group for
assistance.
While the simple formula and the tables in the applications
section deliver good approximations of the transfer func-
tions, a more accurate response is achieved using FilterCAD.
FilterCAD calculates the resistor values using an accurate
and complex algorithm to account for parasitics and op
amp limitations. A design using FilterCAD will always yield
the best possible design. By using the FilterCAD design
tool you can also achieve filters with cutoff frequencies
beyond 256kHz. Cutoff frequencies up to 360kHz are
attainable.
Contact the Linear Technology Filter Applications Group
for a copy the FilterCAD software. FilterCAD can also be
downloaded from our website at www.linear.com.
DC Offset, Noise and Gain Considerations
The LTC1563-X is DC offset trimmed in a 2-step manner.
First, section A is trimmed for minimum DC offset. Next,
section B is trimmed to minimize the total DC offset
(section A plus section B). This method is used to give the
minimum DC offset in unity gain applications and most
higher gain applications.
For gains greater than unity, the gain should be distributed
such that most of the gain is taken in section A, with
section B at a lower gain (preferably unity). This type of
gain distribution results in the lowest noise and lowest DC
offset. For high gain, low frequency applications, all of the
gain is taken in section A, with section B set for unity-gain.
In this configuration, the noise and DC offset is dominated
by those of section A. At higher frequencies, the op amps’
finite bandwidth limits the amount of gain that section A
can reliably achieve. The gain is more evenly distributed in
this case. The noise and DC offset of section A is now
multiplied by the gain of section B. The result is slightly
higher noise and offset.
156323fa