Chuck D. Bones
Circuit Wizard
Part 2 deals with the topic of how opamps are used in pedal circuits.
There are four basic building blocks made from opamps and used in pedals. Buffer; Amplifier; Limiter; Filter.
We'll use examples from the Friedman BE-OD since it contains all four building block types.
Here's a unity gain buffer. We use negative feedback from the output (pin 1) to the inverting input (pin 2) to control the gain. Since we have 100% feedback, the gain is unity. We use a buffer when we need a building block with high input impedance, low output impedance, flat frequency response and 0dB gain. It does essentially the same thing as an emitter follower or source follower. In this example, it buffers Vref.
Next we have a non-inverting amplifier. It too has a high input impedance, low output impedance and flat frequency response, but the gain is greater than unity. Assuming the opamp has very high open-loop gain, the gain of this circuit is 1+R5/R4 = 2.773 = 8.86dB. The signal is not invertered, i.e. the phase of the output signal is the same as the input signal. The gain of a non-inverting amplifier can never be less than unity (0dB).
We can build an inverting amplifier like this. The gain is -(R15+Rtrim)/R14. The gain is negative because the amplifier inverts the signal. If the input moves in the positive direction, the output moves in the negative direction. The gain of an inverting amplifier can be greater or less than unity. The input impedance of an inverting amplifier is equal to R14. We have to take this into account when we design circuits using inverting amplifiers.
We can make a limiter by putting diodes in the feedback loop. This is a non-inverting limiter. It shows up in many distortion circuits, such as the Tube Screamer. When the diodes start to turn on, they increase the amount of feedback which decreases the gain. The larger the signal, the more the diodes limit the signal swing at the output. The gain varies throughout the waveform, which distorts the signal. Non-inverting limiters have a built-in clean bleed because the output signal is the sum of the voltage between pins 6 and 7 plus the input signal. The limiting is slightly softer because of this and as a result there are more dynamics.
We can also make an inverting limiter, like this. Inverting limiters do not normally have a clean bleed, unless we deliberately put one in. R12 serves that purpose because no matter how much the diodes conduct, R12 keeps the instantaneous gain from going to zero. If R12 was zero, the compression would be extreme and there would be very little dynamics. Sometimes that's what we want.
Finally, we have an example of an active filter. This is actually a form of bandpass filter, but it's tuned to make the boost in the bass region.
We can also make inverting filters and the Baxandall tone stack is an example of that.
Something all of the circuits have in common is a DC path from the + input to Vref and a DC path from the - input to the output. This biases the opamp such that the output is centered at Vref.
One question that comes up is how do we select component values? Gain is set by resistor ratios and cutoff freq is set by the product of capacitance and resistance. There are practical limits to the resistor values. Make the resistances too small and the feedback circuit loads down the output. Make the resistances too large and the output won't be centered at Vref. An ideal opamp has zero DC current flowing in or out of the + and - inputs. A real opamp has DC current flowing thru the input pins. That current is called bias current. Opamp manufacturers endeavor to keep the bias currents small and equal, but they are never zero or perfectly balanced. The nominal bias current in an MC1458 is typically 80nA. 80nA flowing in a 1M resistor produces an 80mV voltage drop. Not too bad and it doesn't eat enough headroom to worry about. Put in a 10M resistor and now the voltage is nearly 1V. In this way, the bias currents cause the opamp's output to not be centered at Vref. If it's off a little bit, no problem. The typical bias current in an LM833 is 500nA. Put that thru a 1M resistor and we get a 500mV offset. That is consuming headroom and we certainly don't want to go any higher than 1M. FET-input opamps such as TL072, LF351, CA3130, to name a few, have bias currents measured in pA. The typical bias current in a TL072 is 20pA. That makes a 20uV offset in a 1M resistor. This means we can use very high resistor values with FET-input opamps and not have to worry about bias current effects.
Real opamps don't have infinite gain. If we try to make a very high gain amplifier, the opamp's limited open-loop gain comes into play. My rule of thumb is don't go over 60dB (1000x) in a single stage. I usually keep the max gain to 46dB (200x). If you need more than that, then make the gain in stages. That's what Friedman, Revv and Diezel do in their pedals. The effect of limited gain and bandwidth are very apparent in the Rat. With DIST dimed, the high freq gain would top out over 71dB (3,450x). In reality, it peaks out around 65dB and falls off rapidly above 800Hz because the opamp's open-loop gain falls off as frequency goes up.
I think that's enough for now. Questions?
There are four basic building blocks made from opamps and used in pedals. Buffer; Amplifier; Limiter; Filter.
We'll use examples from the Friedman BE-OD since it contains all four building block types.
Here's a unity gain buffer. We use negative feedback from the output (pin 1) to the inverting input (pin 2) to control the gain. Since we have 100% feedback, the gain is unity. We use a buffer when we need a building block with high input impedance, low output impedance, flat frequency response and 0dB gain. It does essentially the same thing as an emitter follower or source follower. In this example, it buffers Vref.
Next we have a non-inverting amplifier. It too has a high input impedance, low output impedance and flat frequency response, but the gain is greater than unity. Assuming the opamp has very high open-loop gain, the gain of this circuit is 1+R5/R4 = 2.773 = 8.86dB. The signal is not invertered, i.e. the phase of the output signal is the same as the input signal. The gain of a non-inverting amplifier can never be less than unity (0dB).
We can build an inverting amplifier like this. The gain is -(R15+Rtrim)/R14. The gain is negative because the amplifier inverts the signal. If the input moves in the positive direction, the output moves in the negative direction. The gain of an inverting amplifier can be greater or less than unity. The input impedance of an inverting amplifier is equal to R14. We have to take this into account when we design circuits using inverting amplifiers.
We can make a limiter by putting diodes in the feedback loop. This is a non-inverting limiter. It shows up in many distortion circuits, such as the Tube Screamer. When the diodes start to turn on, they increase the amount of feedback which decreases the gain. The larger the signal, the more the diodes limit the signal swing at the output. The gain varies throughout the waveform, which distorts the signal. Non-inverting limiters have a built-in clean bleed because the output signal is the sum of the voltage between pins 6 and 7 plus the input signal. The limiting is slightly softer because of this and as a result there are more dynamics.
We can also make an inverting limiter, like this. Inverting limiters do not normally have a clean bleed, unless we deliberately put one in. R12 serves that purpose because no matter how much the diodes conduct, R12 keeps the instantaneous gain from going to zero. If R12 was zero, the compression would be extreme and there would be very little dynamics. Sometimes that's what we want.
Finally, we have an example of an active filter. This is actually a form of bandpass filter, but it's tuned to make the boost in the bass region.
We can also make inverting filters and the Baxandall tone stack is an example of that.
Something all of the circuits have in common is a DC path from the + input to Vref and a DC path from the - input to the output. This biases the opamp such that the output is centered at Vref.
One question that comes up is how do we select component values? Gain is set by resistor ratios and cutoff freq is set by the product of capacitance and resistance. There are practical limits to the resistor values. Make the resistances too small and the feedback circuit loads down the output. Make the resistances too large and the output won't be centered at Vref. An ideal opamp has zero DC current flowing in or out of the + and - inputs. A real opamp has DC current flowing thru the input pins. That current is called bias current. Opamp manufacturers endeavor to keep the bias currents small and equal, but they are never zero or perfectly balanced. The nominal bias current in an MC1458 is typically 80nA. 80nA flowing in a 1M resistor produces an 80mV voltage drop. Not too bad and it doesn't eat enough headroom to worry about. Put in a 10M resistor and now the voltage is nearly 1V. In this way, the bias currents cause the opamp's output to not be centered at Vref. If it's off a little bit, no problem. The typical bias current in an LM833 is 500nA. Put that thru a 1M resistor and we get a 500mV offset. That is consuming headroom and we certainly don't want to go any higher than 1M. FET-input opamps such as TL072, LF351, CA3130, to name a few, have bias currents measured in pA. The typical bias current in a TL072 is 20pA. That makes a 20uV offset in a 1M resistor. This means we can use very high resistor values with FET-input opamps and not have to worry about bias current effects.
Real opamps don't have infinite gain. If we try to make a very high gain amplifier, the opamp's limited open-loop gain comes into play. My rule of thumb is don't go over 60dB (1000x) in a single stage. I usually keep the max gain to 46dB (200x). If you need more than that, then make the gain in stages. That's what Friedman, Revv and Diezel do in their pedals. The effect of limited gain and bandwidth are very apparent in the Rat. With DIST dimed, the high freq gain would top out over 71dB (3,450x). In reality, it peaks out around 65dB and falls off rapidly above 800Hz because the opamp's open-loop gain falls off as frequency goes up.
I think that's enough for now. Questions?
. 
