The fifth prototype amplifier is designed for use in acquiring EEG signals, which
have considerably lower input levels than do EKG signals. The basic design of this
amplifier is the same as that of amplifier
4. In addition, another instrument amplifier chip has been added to increase
the overall gain sufficiently to acquire EEG signals.
This amplifier consists of two Analog Devices AD623AN
instrument amplifier ICs, and active 4th-order Butterworth lowpass filtration
using a custom-made equal-component Sallen-Key filter. The lowpass cutoff for
this filter is still at ~30 Hz (the same value as was used to acquire EKG
signals), which may be somewhat low for EEG signals. Nevertheless, this filter
band is wide enough to begin to acquire interesting and useful EEG signals (such
as the alpha band), although the cutoff frequency may be increased to about 100
Hz in future designs.
The gain of the first instrument amplifier chip is fixed at 1000x, the
equal-component Sallen-Key filter adds a gain of about another 2.5x, the gain of
the second instrument amplifier IC is set at about 8x, and the output attenuator
drops the signal by about 0.2, giving an overall effective gain of about 4000x.
This is perhaps slightly small, delivering an output signal of about 0.25 VPP
depending on the electrical contacts used and the strength of the underlying
scalp signal. In future it may be desirable to increase the second stage
amplifier gain to about 16x or more, in order to expand the output signal to the
full range of the Mac sound input port.
- Neither the power supply, nor the chip bypass, components are shown. See the
power supply page for schematics. The AD623 is bypassed
by two capacitors: a 100n film, and a 10u. The op-amps are bypassed by a single 100n
- The op-amps used in this implementation (power supply and filter) are OP295s
instead of OP213s. I don't think this makes any difference. Any decent
single-supply op-amp should work fine. However, I usually prefer to use Analog
Devices ICs if possible.
- The input is AC coupled using bipolar 1u capacitors combined with 100K
gate current resistors. This provides a highpass cutoff (-3db, 1st order) at
about 1.6 Hz, which is quite low enough for EEG signals. The input can be DC
coupled, although depending on the ambient electrical environment this may
preclude any acquisition of a signal due to supply rail clipping. The input to
the second stage amplifier chip is also AC coupled using a 1u BP capacitor and a
100K resistor. Since the negative input to the second stage is connected to V0, the
second stage amplifier is essentially single-ended, and the use of an instrument
amplifier IC is perhaps overkill.
- The gain of the first AD623 was fixed at 1000x, its maximum. According to the
datasheet, this provides the maximum common mode rejection for up to 100 Hz.
- The intermediate signal is filtered by an active 4th-order Butterworth lowpass
equal-component Sallen-Key filter. The basic design for this filter was taken
from Don Lancaster's Active Filter Cookbook, which I highly recommend.
This book can be ordered from Amazon for under $30. The cutoff for this filter
is ~30 Hz, and the damping of the two 2nd order stages is 1.848 and
0.765. This filter provides about another 2.5x in gain.
- The output is referenced with respect to V0. The output is
attenuated using a voltage divider constructed from 2K and 500 Ohm resistors. This
is because while the output of the amplifier can be up to 9 VPP (or whatever the
battery happens to be that day), the sound input of the Mac clips at 1.5 VPP.
The gain and phase response of this circuit (actually, of amplifier 4) at
frequencies of interest (~2-80 Hz) is shown here. The frequency
response of external electronics can now be quickly and easily acquired using the
Mac and component software instead of an external oscilloscope.
©Copyright Sky Coyote, 2001-2002.