0
\$\begingroup\$

I am working on a university project that requires me to build a 40kHz band pass filter for receiving an ultrasonic signal.

The filter is to be implemented into a portable device powered by a 9V PP3 battery so a single-supply op-amp (MCP601) is used.

The filter was designed using Analog Devices' filter design tool with the constraint of being single stage and 20dB gain at the center frequency. The filter circuit and values are shown below: (LT1115 used as placeholder)

BPF design schematic

LTSpice

The above LTspice simulation results show that the circuit should work as desired.

The circuit was implemented on stripboard using a LF50 voltage regulator to step down to 5V for Vdd and a voltage divider for Vin+ to give Vin+ = Vdd/2. The circuit is shown below. There is a 100nF decoupling capacitor between Vin+ and ground, as well as a 100nF and 10uF capacitor between the Vdd pin on the op-amp and ground.

enter image description here smol

The issue is when connecting to a signal generator (AFG-2105) to vary the input frequency, and connecting to a Picoscope to read the waveform, the filtering circuit appears to have no effect at all. I have performed a continuity test to confirm that I have isolated the pins of the op-amp and to check the input voltages.

Is there something I am missing?

Is it due to the filter being a high frequency?

EDIT: added full schematic

\$\endgroup\$
5
  • \$\begingroup\$ What do you measure as the DC output voltage? Did you bypass the voltage divider so the impedance at 40kHz is not so many ohms? It might be better to use two dividers, one for the + input and one that replaces R2A (Ra||Rb) = 698 ohms. \$\endgroup\$ Commented Apr 1, 2022 at 9:48
  • 2
    \$\begingroup\$ On the photo of the stripboard, MCP601 pin 6 (output) is directly connected to pin 2 (inverting input). This might make a (sort of) voltage follower. Difficult to be sure without a full schematic of what you built. \$\endgroup\$
    – GeBJT
    Commented Apr 1, 2022 at 10:07
  • \$\begingroup\$ Show us the schematic of exactly what you built, not just the partial snippet. Also show us the bottom side of your board. \$\endgroup\$
    – brhans
    Commented Apr 1, 2022 at 10:28
  • \$\begingroup\$ @GeBJT how would you suggest I connect the output to the inverting input for the feedback? \$\endgroup\$
    – LK26
    Commented Apr 1, 2022 at 12:09
  • \$\begingroup\$ What resistor values did you choose for the voltage regulator generating VREF? Could you add a 10uF decoupler to VREF as well? \$\endgroup\$
    – maszoka
    Commented Apr 1, 2022 at 12:58

2 Answers 2

2
\$\begingroup\$

Thanks for two updates: there is now sufficient clarity for me to suggest a wiring correction. The component identifiers below refer to the LTSpice schematic.

The output of the opamp should be connected to the junction of C2 and R2 (currently, on the topside stripboard photo, the otput is connected to opamp pin 2).

Pictorial representation of modifications below. Cut existing wire at red X. Reconnect to the opposite end of R2, as green line.

enter image description here

Hope this helps clarify my earlier comment!

\$\endgroup\$
2
  • \$\begingroup\$ Thank you, this has gotten me much closer to my desired result. I have a couple of quick questions: the shape of the magnitude plot is now what I would expect, however it is inverted and has negative dB values. Secondly, if I touch any of the exposed metal of the components, the graph changes a bit (to more how I want it) - could this be due to a poor ground connection? Thank you so much. \$\endgroup\$
    – LK26
    Commented Apr 1, 2022 at 13:38
  • \$\begingroup\$ @lewiskell Good to hear about positive progress! I have put a response to the 1st point in a new answer. On the second point:- i) Your body acts as an antenna to pick up EM interference which will be coupled into your circuit at the point you touch it. ii) Simplifying, your finger "looks" like a capacitance (to ground) to a circuit so will alter e.g. filter response. Look up the Human Body Model for ESD (Electrostatic Discharge). Extra decoupling close to the opamp supply pins wouldn't hurt. \$\endgroup\$
    – GeBJT
    Commented Apr 1, 2022 at 17:09
0
\$\begingroup\$

Shape / level of filter response (comment/question on earlier answer)

I'll answer this with reference to LTspice simulation, which you can then apply to your built circuit.

Going back to your simulation result, the green frequency response for V(out):-

The above LTspice simulation results show that the circuit should work as desired

Hmmm... not quite. By default LTspice displays Volts^2 on the vertical axis referenced to 1V rms. So, for example, 0dB = 1V; 20dB = 10V; 1mV = -60dB. (This will explain negative dB results.) Your results imply that your opamp is outputting 10V rms (from a 5V supply!). This can happen in LTspice but not in real life.

For exploration / learning try these:-

a) re-run the simulation with a 1mV input (this might be closer to your actual ultrasonic input)

b) run a new simulation plotting V(out)/V(in) (the out-to-in ratio is more meaningful when you are designing amplifiers and filters, plus you get a phase plot; similar numbers apply as in dB values above: e.g. when V(out)/V(in) = 1, this is displayed as 0dB, etc.

c) as @maszoka has suggested, increase the decoupling cap on VREF (currently 100nF) 1uF, 10uF or more.

\$\endgroup\$
2
  • \$\begingroup\$ I see a few misconceptions in there. The Y axis is not in V^2, but in 20*log10(V), hence the dB scale. To get it linear, RClick on the Y axis and select Linear. Then, the .AC analysis in all SPICE programs (since 50 years ago) linearizes the circuit, therefore there is no clipping involved (that would be a nonlinearity). As such, AC 1m and AC 1k will both yield the same out/in ratio. Feeding a source with AC 1 means there is no need for a ratio since the input is unity, already. \$\endgroup\$ Commented Apr 1, 2022 at 20:52
  • \$\begingroup\$ @aconcernedcitizen Thank you for bringing more clarity to my answer. \$\endgroup\$
    – GeBJT
    Commented Apr 1, 2022 at 22:17

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.