I have a noisy and low amplitude (0-50mV amplitude) signal from a hall effect sensor (TI DRV5053) that has a ~1V DC offset. The signal will be in the range of 95-105 Hz and is used to determine the amplitude of piston motion using a small magnet.

I needed to get this into something useful for our data system so I went to the AD website and generated a bandpass filter with a gain of 100 and narrow passband around 100 Hz. THis would get rid of the DC offset and give the signal a good amplitude to connect to a data system and remove the unwanted high frequency noise.

Filter Circuit

I sloppily soldered two of these active filters together on a small proto board and while one works exactly how I expected it to, the second one does not. Whenever I connect the function generator with a 20mV 100Hz sine wave there is a high frequency oscillation the input at about 350 kHz, however the output is oscillation free and at about 1/2 of the amplitude that I expect.

Unfortunately I lack the experience necessary to deduce where the problem may lie. I have inspected everything I could and it looks like it's connected properly. Because the output is not oscillating I feel like that means that the second stage of the filter is working and the first stage is not. There have been brief moments when connecting the input signal where there was no oscillations but it quickly starts oscillating again. I haven't been able to consistently reproduce this.

Does anyone have any idea about what might be causing this oscillation and what I might want to investigate in order to fix this.

Here are images of the working and non-working filters on an oscilloscope.

Oscillating signal 40mV pk-pk input is purple, and the filtered output is yellow

Properly Working Active Filter.

  • 2
    \$\begingroup\$ Pictures of your board? And why did you use two separate ICs for the op amps, when you have two op amps on each chip anyway? \$\endgroup\$
    – Hearth
    Commented Mar 14, 2019 at 17:40
  • \$\begingroup\$ On a breadboard? Apply those +/- DC supply bypass capacitors directly from the IC supply pin to the local ground where opamp "+" is grounded, and R2A, R2B are grounded. Some folks apply these capacitors on the two breadboard strips used for supply distribution - may be too far away. Could also be improper ground point for your generator: ground loop + gain makes this grounding point critical. This may involve your 'scope ground too. \$\endgroup\$
    – glen_geek
    Commented Mar 14, 2019 at 18:04
  • 1
    \$\begingroup\$ Need pictures of the implementation, the problem is probably in parasitic component values from the board \$\endgroup\$
    – Voltage Spike
    Commented Mar 14, 2019 at 18:25

1 Answer 1


Although the CMOS dual Op Amps are supposed to be unity gain stable with 55 deg phase margin at 24V supply, that is with no load capacitance from a cable from sig. gen and probe.

The Multiple-feedback BPF filter you chose, if you imagine the caps as short circuits, shunts the feedback to unity gain , which you want to avoid for CMOS OA's. It oscillates with a reactive (xx pF) load at the unity gain-BW frequency since the output impedance is much higher than BJT types. Your resonance is just below the minimum of spec range due to small loading effects. (0.4MHz Min).

Therefore, I recommend you use the Sallen-Keys feedback BPF configuration, which I believe, does not have this problem as the attenuation is done on the input with positive feedback for filtering and negative feedback for gain.

enter image description here

Here I would choose min Rf at 100k and use 0.5% parts with a Bessel response and a BW of 15 Hz and Av=40 dB.

So scale all these 0.5% parts of R's x100 and reduce C/100

enter image description here

  • \$\begingroup\$ You have made a good point about the loading of the second stage, if perhaps Mr. Goodell is using a coax cable to the scope rather than a probe. However, the multiple feedback band pass is an excellent topology, and I doubt that that switching to Salen Key is worthwhile. However, if he rebuilds the filter, he may inadvertently fix whatever problem he really has. \$\endgroup\$
    – user69795
    Commented Feb 19, 2020 at 4:51
  • \$\begingroup\$ @user69795 on 2nd thought , measurement error , corrected by a slower sweep might indicate line noise from supply or floating pickup. \$\endgroup\$ Commented Feb 19, 2020 at 10:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.