Help with audio common mode rejection circuit

Hi all,

I am making an audio amplification circuit for a violin pickup and I was wondering if my circuit could get a look see before I start building it. I have limited knowledge in CMRR and I couldn’t figure out what the best pre-amplification opamp to use, but with the searching and research I could do I attempted to build and test a circuit in Spice before I started wasting money and patience.

The general design parameters:

  • audio input is from a pvdf piezo film type transducer hence the input stage
  • single positive voltage source
  • output to arduino due adc pin at 3.3v
  • no restriction on power consumption

Below I have attached a photo of the circuit, if you spot anything peculiar or an optimization that I haven’t performed, please let me know! :slight_smile: Thank you for your time.

Interesting. Why must it be differential?

Edit - seeing how this might be going... piezo transducers work perfectly well single ended. The preamp circuit for that will have about four times fewer components. It will also be easier to design for a single supply.

I think you can do all of this with a single op-amp stage. Or maybe two... i.e. You can make a differential amplifier with gain and offset, and some filtering too.

If you are using a standard piezo pickup, all of them I've seen are single ended (one signal wire and a ground). So if you are using a commercial pickup, you'll probably have to re-wire it for a balanced connection. The cable shield can be left open at the pickup and grounded at the amplifier end.

Piezo pickups are high impedance, but I'm not sure you need 10M Ohms. I believe most piezo instrument pickups plug into a normal guitar/instrument input (~1M Ohm). If you haven't done so already, try experimenting with lower impedance to see what it does to the signal and tone. Lower impedance means less noise pick-up. (And, I mean really try it, don't simulate it. ;) ) Your differential stage effectively has two 400K resistors in series. That's nearly 1M and your circuit may be fine without the input (buffer) stage.

With the input buffer, your differential amplifier (technically it's not a comparator) can have much lower impedance, maybe 5K would be good.

You don't need all those 3.3V/2 bias circuits. I'd bias the op-amps at half the 12V supply. C6 will block that higher bias and R15 & R18 will re-bias to 3.3V/2.

You probably need a volume/gain control in there somewhere, at least during the development process until you know how much signal you're getting out of the pickup.

You need a resistor in series with C6 (maybe 1K) to limit the current out of the op-amp when/if your protection diodes kick-in.

DVDdoug: If you are using a standard piezo pickup, all of them I've seen are single ended (one signal wire and a ground). So if you are using a commercial pickup, you'll probably have to re-wire it for a balanced connection. The cable shield can be left open at the pickup and grounded at the amplifier end.

The pvdf film device I am using is just a generic device which has a very very high output impedance. I had to crank my amp for straight amplification without a pre-amp. Then the problem became that the amplification was SOO high, that all the background radiation was amplified instead of the signal, even with proper shielding (on the sleeve, still have to make a copper jacket for the device. I wouldn't have thought the surface area of the pvdf would have been big enough to make that much noise.)

Also, the PVDF film doesn't like to be stretched too far without amplification. On the instrument I have a piezo ceramic crystal that is attached through 20 feet of cable before being amplified and it works fine, the pvdf on the other hand is still a whisper even with the amp AND a pre-amp cranked way high. I am trying my best not to have to put a amplifier on the instrument for weight concerns.

DVDdoug: With the input buffer, your differential amplifier (technically it's not a comparator) can have much lower impedance, maybe 5K would be good.

Good point... I had a bunch of 400kohm smd resistors laying around and I guess I tried forcing myself into using them. Luckily I found a reel of 5k's though.

DVDdoug: You don't need all those 3.3V/2 bias circuits. I'd bias the op-amps at half the 12V supply. C6 will block that higher bias and R15 & R18 will re-bias to 3.3V/2.

Doh... Forgot about the blocking resistor on the output stage. That would explain why the output is still a little jittery in simulation.

Just to verify though, using a voltage in the middle of the voltage range of the opamp will give it better frequency response correct? Because its allowed to transition from 6v to 0, instead of 1.65v to 0? Something to do with capacitance and me not wanting to do the calculations.

DVDdoug: You need a resistor in series with C6 (maybe 1K) to limit the current out of the op-amp when/if your protection diodes kick-in.

The TL072 are designed for continuous short circuit current source and sink, or is it still good practice to do this? Ill have to check but the output of the TL072 is well below the 1N4148's.

aarg: Interesting. Why must it be differential?

The hint is in the title of the thread. That's how you remove common-mode noise.

daviddein: I am trying my best not to have to put a amplifier on the instrument for weight concerns.

Therein is your problem.

You are pushing the proverbial uphill.

The pre-amp must be directly attached to the sensor. It is simply not negotiable. Correct your most basic design blunder and all will be well. :grinning:

Cable capacitance removes high frequencies (20feet/10Megohm is bad).
Piezo loading (<10Meg) removes low frequencies.

Put a Jfet preamp in the instrument, and phantom power it over the wire.
Just like a lapel microphone does.

Google “tillman preamp”.
Leo…

Wawa: Put a Jfet preamp in the instrument, and phantom power it over the wire. Just like a lapel microphone does.

Good point. This proposal in in fact, exactly what an Electret microphone is and of course, the amplifier is always built into the microphone.

Paul__B: Therein is your problem.

You are pushing the proverbial uphill.

The pre-amp must be directly attached to the sensor. It is simply not negotiable. Correct your most basic design blunder and all will be well. :grinning:

:( I know. I was trying to dream.

MarkT: The hint is in the title of the thread. That's how you remove common-mode noise.

I'm aware of that. But another way is shielding. That's been successful in most audio applications for almost 100 years now.

aarg: I'm aware of that. But another way is shielding. That's been successful in most audio applications for almost 100 years now.

I took the advice of actually building the circuit and testing it rather than simulating. Shielding was a bigger issue than trying to implement fancy amplification schemes like common mode rejection.

DVDdoug: I think you can do all of this with a single op-amp stage. Or maybe two... i.e. You can make a differential amplifier with gain and offset, and some filtering too.

I took your suggestion of two opamps and it worked just fine, and no problems with amplification even with an input impedance to the differential amplifier at around 600kohm. The component number is pretty small so a small ic board won't weigh too much I guess. The extra wire I was worried about was a non-issue which is why I wanted to keep the pre-amp off of the instrument in the first place.