Help with MFB second order high pass filter

[Chris Bassett]

Having hit a wall in a project, a light bulb goes off in my head and tells me to share my design challenges with a community of people with similar interests.

I'm designing a instrument amp for a friend and need to put in a series of filters, The circuit shown below works, but the output shows a few quirks.

The output from the In-amp is a smooth sine wave for both the 500 and 30 Htz scenarios shown. Any thoughts?


Note:
The circuit values were calculated from the filter section in TI's "Op Amps for Everyone".

 

[Harald Kapp]

The LM324 is a quad op-amp. What did you do with the 3 unused amplifiers in the package? I guess you left the unconnected (?). This can cause unwanted oscillations and current draw due to saturation (of the unused amplifiers) which in turn can influence the one you are using.
Start by properly terminating the unused op-amps (see e.g. here, figure 3).

 

[LvW]

Chris - your circuit is wrong. It has no dc feebdack, which is necessary for each opamp.
The 17.9k resistor must be connected directly to the inv. input.

 

[Harald Kapp]

Gain of the AD620 is set by a resistor between the RG-pins (1,8) of teh IC, see datasheet.
For teh second stage (filter), here is a design tool showing correct wiring and calculation of components.

 

[Chris Bassett]

After trying out the suggestion, the circuit's output is conciderably smoother. Thank you.

The feed back was correct on the breadboard, but drawn incorrectly.
The op-amp terminals were terminated as suggested.
The in-amp was set up with the gain resistor, but I left it out of the drawing. I also have a voltage follower after spitting the rail voltage.

At higher frequencies, a birth of saw tooth distortion occurs. The outputs of the filter and the in-amp are shown for 10000 Htz. Is this normal? Should I consider the charge/discharge time of my caps?

Cheers!
Chris

 

[Harald Kapp]

Should I consider the charge/discharge time of my caps?

Definitely.
The LM324 datasheet states on page 11 that a capacitove load of 50pf can be driven without additional measures. You should therefore reduce the capacitances in your circuit and increase the resistances accordingly.

 

[Chris Bassett]

For my own interest, any good reading on what resistive isolation is would be appreciated.

The suggestion to lower my capciance per the data sheet works well. Using 470pF the circuits output is considerably cleaner. The scopes shows 10000Htz getting superimposed over 60Htz. The feedback resistor is 10M and I suspect the high impedance is picking up the line voltage.

Do you think including a voltage divider in the feed back loop and calculating an equivalent resistance would be a good direction to head in?