4
\$\begingroup\$

I've been attempting to design a distortion pedal for use with my bass guitar and amplifier. The simulations in LTspice were promising, I moved to implement it and, in short, it did not work.

Circuit

Note: The amplifier used in the circuit is merely illustrative. I am actually using an OPA2134. The sheet is here: https://www.ti.com/lit/ds/symlink/opa2134.pdf

This is the basic circuit, a bipolar power supply to power the op-amp and provide a 0V reference, the bass below it, with its impedance accounted for, the amplifier to its right, with soft clipping on top and hard clipping on the right. The bass amplifier is represented by the 1M input resistance.

From what I know, soft clipping is basically logarithmic amplification (high voltages become not so high) with a capacitor in parallel to lower the slew rate. Hard clipping is just cutting the wave once it goes above 0.7V or so (depending on the diode of course).

Now, this works fine as is (simulation-wise), and here are two examples of soft and hard clipping, respectively:

soft clipping

hard clipping

Note: Green is the output, red is the input scaled by 10. The original signal should have an amplitude of around 40mV, I believe. Also, for hard clipping to happen the soft clipping part of the circuit must be disconnected. This would be done with a switch in practice.

Seeing this, I build a little test circuit, with only the bipolar supply and the amplifier. First I used a 220 Ohm input resistance and a 440 Ohm for feedback (so the signal would be amplified by a factor of 3). It worked, but the noise when I played low frequency notes was noticeable.

After that, I switched the feedback resistor for a 10k potentiometer (wired up as a variable resistor). The amplifier produced a very loud, low frequency noise at all times. The bass was (just barely) audible, but it sounded very quiet considering the ~50x amplification factor.

So far, my questions are:

  1. Is the bipolar power supply adequate? Would input biasing be better in this situation? If so, why?
  2. Is the value of the input and feedback resistors adequate? Should they both be higher/lower?

I thought filtering the input signal might help, so I moved to design a pair of simple filters with the components I had available. This, however, distorted the signal massively. Here are the filters:

input filters

I must've done something really dumb (filters are supposed to be quite simple) because the signal looks like this:

result of the filters

As you see, the input signal now has a DC offset, and the output looks like that.

Here's another pair of questions:

  1. Are the filters correctly designed?
  2. Is there some sort of incompatibility with the bipolar supply?

I hope I was clear enough with the questions and examples, if you need any extra information let me know. Thank you in advance for your time!

SOLUTION

As described in the answer, the solution was using a voltage follower in the power supply to provide a much more stable ground node. The new power supply is below:

solution

Note: The wires that intersect but don't have the blue square are not actually connected, so if you have the same issue, don't connect your output to the op-amps power supply :)

\$\endgroup\$
  • \$\begingroup\$ You have -100mV of input offset probably due to input bias current and 1Meg or something . Null the offset somehow. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Dec 31 '19 at 17:57
  • \$\begingroup\$ You're not clipping at all on the positive side. Could a diode be smoked, or in backwards? \$\endgroup\$ – Scott Seidman Dec 31 '19 at 18:23
  • \$\begingroup\$ nah , its all due to input DC offset, but very crude Bass Synth compared to soundcloud.com/jamie-laboz/iron-pterodactyl \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Dec 31 '19 at 18:30
  • \$\begingroup\$ @ScottSeidman I disconnected the clipping modules for the filter experiment, so it's normal \$\endgroup\$ – Alexandre Rodrigues Jan 1 at 15:13
  • \$\begingroup\$ To my experience, the acoustic impression of distortion is very sensitive to small changes of the parameters. Since you are referring to your impressions in your question, you should probably first find out what characteristics are due to deviations from your required distortion curve, and which are due to you misunderstanding the psychoacoustic consequences of the required curve. So record your guitar with your PC and then use e.g. Audacity and try out different distortion curves to understand what it sounds like. \$\endgroup\$ – oliver Jan 2 at 18:27
3
+50
\$\begingroup\$

Your GND which is made by voltage division to 4,5V halves is weak, the first pulse throws it off from the midpoint of your 9V. The output clipper shouldn't be connected to it. Make more stable GND with an opamp voltage follower. Use the resistor-halved version only for opamp signal inputs. In addition reduce RG1 and RG2 to smaller resistances, say 47kOhm to recover from minor unsymmetry faster.

The signal input is now connected nearly directly to the output of the opamp through the signal source. That's positive feedback. The operation is very difficult to predict, because the input also gets signal and noise. I guess the guitar pickup works as resonant circuit and you have an oscillator.

Make a simulation where those points which are connected to your resistor halved GND are connected to each other with a wire and let the simulation GND be the minus pole of your battery. The resulted output is quite different than a perfect output plus 4,5V DC offset.

| improve this answer | |
\$\endgroup\$
  • 1
    \$\begingroup\$ Thank you, the voltage follower seems to have worked! I understand it provides a much more stable ground node, the simulation looks like how it should now. I'll try to implement it when I find the time, and I'll award you the bounty when I can (10 min cooldown) \$\endgroup\$ – Alexandre Rodrigues Jan 3 at 18:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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