What's a buffer, how do they work and why would we want one?
- Thread starter Chuck D. Bones
- Start date
spi
Well-known member
In a similar vein, I've wondered what goes into choosing the right value for this resistor?
Chuck D. Bones
Circuit Wizard
Or posted a meme.Uh oh! Shouldn’t have quoted that whole post…
steviejr92
Authorized Vendor
Sorry taking it down! Not the place for it!
Chuck D. Bones
Circuit Wizard
spi
Well-known member
yes, resistor between a transistor and vref/gnd
Chuck D. Bones
Circuit Wizard
There are two considerations:
1. How much DC current do we need to bias the transistor?
2. What load do we want to present to the previous stage (or device)?
The base current flows thru the bias resistor and we don't want too much voltage drop.
Often, that bias resistor is part of a high-pass filter. We choose the R & C to pass the right amount of bass.
An example is in order. Let's look at the LGSM TS clone input buffer. Vref is 4.5V and R3 is the bias resistor. Q1's collector current will be around 400μA. According to the datasheet, median HFE at Ic = 400μA is around 125. That give us a base current of (let's not see all the same hands now...)
Ib = 3.2μA.
That means we drop 1.6V in R3. That voltage drop eats some of our headroom, but we have plenty of headroom on the input side because a guitar signal might peak out a 1Vp-p if you have hot pickups and hit the strings real hard. 510K is a high enough resistance that we won't load the pickup appreciably.
Now let's look at the output buffer. I included C7, C8, R11 & the VOLUME pot because I want to suggest an improvement. R12, R13 & Q2 are the same as the input buffer, so we drop 1.6V in R12. We will have the same or more headroom if we set the VOLUME for unity. But if we dime the VOLUME, we'll have less headroom. Probably not a big deal if we keep TONE below noon because the max output from the 2nd stage will be on the order of 2Vp-p. But... There is no problem with loading the volume pot a little bit, so we could reduce R12 to 100K which in turn reduces the voltage drop to 0.32V and increases the headroom. We could also use a higher HFE transistor for Q2, which would also reduce the voltage drop in R12. We could go a step further. The output of IC1.2 is biased to Vref. The VOLUME control is biased to Vref and so is Q2's base. So why do we need C7, C8 & R12? We don't. We can remove all three parts and jumper C7 & C8. Now the maximum resistance between Q2's base & Vref is 100K. As we turn VOLUME down, the resistance gets lower. Why are C7, C8 and R12 even in there? Because they are leftovers from the Tube Screamer's soft switching. There might be some slight scratching when VOLUME is turned since Q2's bias will change a little bit when VOLUME is turned. Probably not audible and even if it is, so what? How often do we turn the VOLUME knob?
1. How much DC current do we need to bias the transistor?
2. What load do we want to present to the previous stage (or device)?
The base current flows thru the bias resistor and we don't want too much voltage drop.
Often, that bias resistor is part of a high-pass filter. We choose the R & C to pass the right amount of bass.
An example is in order. Let's look at the LGSM TS clone input buffer. Vref is 4.5V and R3 is the bias resistor. Q1's collector current will be around 400μA. According to the datasheet, median HFE at Ic = 400μA is around 125. That give us a base current of (let's not see all the same hands now...)
Ib = 3.2μA.
That means we drop 1.6V in R3. That voltage drop eats some of our headroom, but we have plenty of headroom on the input side because a guitar signal might peak out a 1Vp-p if you have hot pickups and hit the strings real hard. 510K is a high enough resistance that we won't load the pickup appreciably.
Now let's look at the output buffer. I included C7, C8, R11 & the VOLUME pot because I want to suggest an improvement. R12, R13 & Q2 are the same as the input buffer, so we drop 1.6V in R12. We will have the same or more headroom if we set the VOLUME for unity. But if we dime the VOLUME, we'll have less headroom. Probably not a big deal if we keep TONE below noon because the max output from the 2nd stage will be on the order of 2Vp-p. But... There is no problem with loading the volume pot a little bit, so we could reduce R12 to 100K which in turn reduces the voltage drop to 0.32V and increases the headroom. We could also use a higher HFE transistor for Q2, which would also reduce the voltage drop in R12. We could go a step further. The output of IC1.2 is biased to Vref. The VOLUME control is biased to Vref and so is Q2's base. So why do we need C7, C8 & R12? We don't. We can remove all three parts and jumper C7 & C8. Now the maximum resistance between Q2's base & Vref is 100K. As we turn VOLUME down, the resistance gets lower. Why are C7, C8 and R12 even in there? Because they are leftovers from the Tube Screamer's soft switching. There might be some slight scratching when VOLUME is turned since Q2's bias will change a little bit when VOLUME is turned. Probably not audible and even if it is, so what? How often do we turn the VOLUME knob?
jesuscrisp
Well-known member
Removing those coupling caps plus the one between input buffer and gain stage are among my favorite TS mods!
Overall great thread on buffers, only thing I slightly disagree with is the Rat example. The JFET buffer sits after the tone stage and before the volume pot, where most likely it's job was to seperate the two so the volume knob wouldn't affect tone. However, there's still the volume knob after the JFET which would cause the output impedance of the circuit to vary depending on where the pot is set. Thus it's not doing the typical job of an output buffer, which would be to provide consistent low output impedance.
Overall great thread on buffers, only thing I slightly disagree with is the Rat example. The JFET buffer sits after the tone stage and before the volume pot, where most likely it's job was to seperate the two so the volume knob wouldn't affect tone. However, there's still the volume knob after the JFET which would cause the output impedance of the circuit to vary depending on where the pot is set. Thus it's not doing the typical job of an output buffer, which would be to provide consistent low output impedance.
Chuck D. Bones
Circuit Wizard
I guess that depends on your definition of "low." The Rat VOLUME control could be reduced to 10K, then the output impedance would vary from 0Ω to about 2.7K. In any case it's a buffer and it's at the output end of the circuit.
Hey Dave,
I have a question about buffers and why they reduce squealing when used in front of a high gain pedal or high gain chain?
In my situation I was experimenting with a TC Sentry in the loop of my Bogner XTC 101B using a 4 cable method. This worked well on the clean and blue channel but on the red it started to squeal and I had to lower the master to around 9 oc to remove the squeal.
I read about using a buffer between the guitar and the input of the Sentry, similar to using it in front of a Valhalla build.
I built a simple Cornish Buffer pedal a while ago and used it between the guitar and Sentry and the squealing is gone now.
So is that mainly because the impedance is low going into the Sentry and there's no loading of the guitar (pickups)?
Would be great if you could explain why this works exactly!?
Thanks!
I have a question about buffers and why they reduce squealing when used in front of a high gain pedal or high gain chain?
In my situation I was experimenting with a TC Sentry in the loop of my Bogner XTC 101B using a 4 cable method. This worked well on the clean and blue channel but on the red it started to squeal and I had to lower the master to around 9 oc to remove the squeal.
I read about using a buffer between the guitar and the input of the Sentry, similar to using it in front of a Valhalla build.
I built a simple Cornish Buffer pedal a while ago and used it between the guitar and Sentry and the squealing is gone now.
So is that mainly because the impedance is low going into the Sentry and there's no loading of the guitar (pickups)?
Would be great if you could explain why this works exactly!?
Thanks!
Chuck D. Bones
Circuit Wizard
This is my best guess as to why putting a buffer in front of a pedal can stop it from squealing.
The squeal we hear is oscillation. Oscillation requires gain and positive feedback at one or more frequencies. The positive feedback path is not deliberately designed into a pedal circuit, with a few notable exceptions.
Any coupling between the input and output of the pedal can provide enough positive feedback to initiate and sustain oscillation.
Any conductor (a wire, for instance) will act as antenna and as a plate on a capacitor. Put two wires in proximity and you have a potential feedback path. Just a few pF is enough coupling if the pedal has high gain and high input impedance. Imagine a pedal with a 1pF capacitor connected from input to output and no guitar plugged in. That capacitor and the input impedance of the pedal form a voltage divider. The higher the input impedance, the stronger the feedback signal. In simulation, 1pF is enough to send a Thermionic into oscillation with GAIN, PRESENCE, TREBLE & VOLUME dimed.
Why does inserting a buffer in-line ahead of the squealing pedal cure the unwanted oscillation? Because the output impedance of the buffer, which is normally very low, is now in parallel with the squealing pedal's input impedance. It causes the voltage divider mentioned in the previous paragraph to become much more lossy. That increased loss may well be enough to cure the squealing. So now you're probably thinking "why not just put a buffer in the pedal to cure the squealing?" The problem with that is now the buffer and its wiring are in proximity to the output of the pedal so we haven't really changed anything. In fact, we might have made it worse because the buffer's input impedance could be even higher than the original pedal circuit.
What can we do then? Reduce the coupling between input and output. Make sure the pots, jacks, switches, PCB, etc. are properly grounded to the enclosure. Route the input and output wires along the enclosure walls and far from each other. The wiring to the stomp switch is a place when input and output wires are in close proximity. Keep the wires from PCB to stomp switch short & direct. We can try shielding the input wire, but the shields need to be grounded at both ends with short ground wires. The board routing is also a concern. Sensitive input traces need to be separated from the output circuit. Ground planes can help to provide shielding on the board.
The squeal we hear is oscillation. Oscillation requires gain and positive feedback at one or more frequencies. The positive feedback path is not deliberately designed into a pedal circuit, with a few notable exceptions.
Any coupling between the input and output of the pedal can provide enough positive feedback to initiate and sustain oscillation.
Any conductor (a wire, for instance) will act as antenna and as a plate on a capacitor. Put two wires in proximity and you have a potential feedback path. Just a few pF is enough coupling if the pedal has high gain and high input impedance. Imagine a pedal with a 1pF capacitor connected from input to output and no guitar plugged in. That capacitor and the input impedance of the pedal form a voltage divider. The higher the input impedance, the stronger the feedback signal. In simulation, 1pF is enough to send a Thermionic into oscillation with GAIN, PRESENCE, TREBLE & VOLUME dimed.
Why does inserting a buffer in-line ahead of the squealing pedal cure the unwanted oscillation? Because the output impedance of the buffer, which is normally very low, is now in parallel with the squealing pedal's input impedance. It causes the voltage divider mentioned in the previous paragraph to become much more lossy. That increased loss may well be enough to cure the squealing. So now you're probably thinking "why not just put a buffer in the pedal to cure the squealing?" The problem with that is now the buffer and its wiring are in proximity to the output of the pedal so we haven't really changed anything. In fact, we might have made it worse because the buffer's input impedance could be even higher than the original pedal circuit.
What can we do then? Reduce the coupling between input and output. Make sure the pots, jacks, switches, PCB, etc. are properly grounded to the enclosure. Route the input and output wires along the enclosure walls and far from each other. The wiring to the stomp switch is a place when input and output wires are in close proximity. Keep the wires from PCB to stomp switch short & direct. We can try shielding the input wire, but the shields need to be grounded at both ends with short ground wires. The board routing is also a concern. Sensitive input traces need to be separated from the output circuit. Ground planes can help to provide shielding on the board.
Last edited:
Thanks for explaining that Dave, it’s very helpful!
I now understand why using a noise gate with the detection loop containing both pedals before the input of the tube amp and the tube amp preamp on high gain including quite some length of cable can work as an antenna as well when used without a buffer.
I now understand why using a noise gate with the detection loop containing both pedals before the input of the tube amp and the tube amp preamp on high gain including quite some length of cable can work as an antenna as well when used without a buffer.
Chuck D. Bones
Circuit Wizard
If the cables are shielded and the grounds connected at each end, then the cables connecting the pedals and amp should not be a problem. One wonky pedal or cable can screw up the whole system.
The cables I used were good, low capacitance, no pedals going into the amp, only guitar into Sentry noise gate and the preamp of the Bogner in the loop on the red channel was enough to trigger the squeal. With the buffer between guitar and Sentry I can max out the amp and not get a squeal. So still not sure where the squeal is produced? I would say inside the Sentry? Even with short cables between the Sentry and the Bogner it can squeal without a buffer.
Here's a simple diagram:
And the squeal-less version with a buffer (Cornish buffer) or a buffered pedal like the Strymon Compadre compressor (off/no compression/buffer on):
Last edited:
