Q
Are there any applications pitfalls to avoid when using AD8237?
A
The high-bandwidth mode setting could be a significant stumbling block becauseusers could get the wrong impression. The AD8237 was designed to be a high
dc-precision amplifier. There is some residual ripple from the chopping
frequency at 27kHz. If the high-bandwidth mode is used at a gain of 10, the
part still has 100kHz bandwidth which is well beyond the chopping frequency so
the ripples are gained by 10 as well. The low bandwidth mode helps to reduce
the ripples somewhat in this case and it may still be necessary to employ
additional external filtering or to place a capacitor from OUT to FB to reduce
the gain at high frequencies. The real value-add of the high-bandwidth mode is
that it greatly improves bandwidth and settling time for high gains, such as
G=1000.
One effect of the ALS circuit is that there is a bandwidth and slew-rate for
the common-mode signal. The main effect of this is in the large signal
frequency response (see figures 25 and 26 in the AD8237 data sheet) The part
performs much better when the AC common-mode of the two input pairs (+IN, –IN,
and FB, REF) are matched, for example if only one input is driven with an ac
signal, it should be the positive input because FB will move with +IN, creating
a matched ac common-mode condition.
The bias current of the AD8237 is typically very small, but the input impedance
can cause currents that are considerably larger than the nominal bias currents
to flow. See figures 15 and 16 in the AD8237 data sheet. If these currents are
a significant source of error in a given application, matched resistances at
all four terminals cancels the input impedance errors mathematically. To do
this practically, consider the source resistance and the gain-setting
resistance. If the source resistance is negligibly small compared with the
gain-setting resistance, add resistors equal to the parallel combination of the
two gain setting resistors in series with the other inputs (+IN, -IN, and REF).
If source resistance is larger than the gain-setting resistance, let’s say it’s
at the +IN terminal, then add resistors equal to the source resistance in
series with –IN and REF, and a resistor equal to the source resistance minus
the parallel combination of the gain-setting resistances in series with FB.
The AD8237 cannot measure “floating inputs”, that is, without a dc return path
to the AD8237 power supply. Common examples are inputs with dc blocking
capacitors and thermocouples, both of which need a resistor for AD8237 bias
current to flow to ground, but the floating input problem can also happen if
the input signal source is referred to a separate ground than the AD8237 power
supply if there is nothing to relate the two voltages.