Figure 42 in the AD8236 datasheet shows an ECG application circuit that I'd like to use. However the cutoff frequency of the low pass filter in the circuit is not what I'd like. How do I change it?

Figure 42 in the AD8236 datasheet shows an ECG application circuit that I'd like to use. However the cutoff frequency of the low pass filter in the circuit is not what I'd like. How do I change it?

The AD8236 includes a 2nd order low pass Bessel filter as the last stage in the schematic. The filter is composed of the following components: 24.9 kohm resistor, 4.02kohm resistor, 220 nF capacitor, 680 nF capacitor, and AD8609 op amp.

Since we will be changing some of these values in our discussion, we need some reference designators. I am going to label the components as follows:

24.9 kohm: R1

4.02 kohm: R2

220 nF: C1

680 nF: C2

For a low pass Sallen Key architecture with a gain of 1 (as shown in the figure 42) the equations are as follows:

f0 = 1/(2*PI*tau).

Q = tau/(R1*C1+R2*C1)

Where tau= sqrt(R1*C1*R2*C2)

We use a Bessel filter in the ECG schematic. Unlike the Butterworth or Chebyshev topologies, this type of filter has a linear phase response. In other words, the filter shifts different frequency components the same amount in time. This is important for applications such as ECG, where the filtered results are viewed in the time domain: you want your high frequency and low frequency components to line up on the screen.

For a 2nd order Bessel filter, we need a Q of 0.58.

Let's say we want a new cutoff frequency of 500 Hz. Normally when engineers mention the cutoff frequency, they have in mind the frequency where the output is attenuated 3 dB. For a Bessel filter this is not the same thing as f0. For a 2nd order Bessel filter, attenuation at f0 is only 1 ½ dB. If we want 3dB attenuation at 500 Hz, a lower f0 is needed: specifically 500/1.36 = 368 Hz. (All of this information comes from the excellent textbook, "Design of Analog Filters" by Van Valkenburg).

So for the following:

f0 = 368 Hz

Q = 0.58

The following standard component values get us pretty close:

R1: 24.9kohm

R2: 49.6 kohm

C1: 10 nF

C2: 15 nF

Since we have 4 variables to play with, but only two equations, many different combinations of component values will work. However it takes some trial and error to get something pretty close with standard capacitor values.

Or if you don't want to mess with the equations you can just have the analog filter wizard compute it for you.