how necessary are input and output buffers in an overdrive?

[therealfindo]

After the success of a stereo booster I made, and learning a little about op amps in the process, I want to keep extending that.. I've been reading the 'cook your own distortion' article, which I've found really helpful, and I want to try putting together a simple op amp overdrive / distortion 'from scratch' (i.e. do the maths etc.) I was going to use a TL071, as that's the one I'm familiar with.
However, I also had a look at some other circuits to identify the sections, and I see that things like the TS have input and output buffers - I understand the concept of a standalone buffer, having recently built an op amp buffer too, but I'm wondering how important they are in an overdrive circuit? After all, it's the same op amp!

Would it hurt not to have input and output buffers in a simple overdrive circuit? Or, is it worth using a dual op amp and using the second op amp half as an input buffer - especially if it will be a kind of 'always on' low gain drive near the start of the chain?

Does that make sense?

[dwmorrin]

The tube screamer has an input buffer to split the signal into dry and wet paths, a requirement for its FET switching scheme.
If you're not using FET switching, but instead a mechanical switch, then you can just make the first opamp high impedance to get a buffering effect.
Check out the MXR distortion+ schematic.  You can sub in a TL071 for the 741.
This simple 1 opamp stage has low impedance output and high impedance input.
If you left it on all the time, you'd get the buffering effect.
I have a 2 standalone buffers I will sometimes put at the beginning or end of a circuit I am breadboarding "just to see" how it may affect the circuit, but generally it is not required to bother with extra buffers, especially when you're already using an opamp.

[Seljer]

You have the runoffgroove tube reamer, which is a minimalized tubescreamer without the buffers and it still functions and sounds perfectly fine http://www.runoffgroove.com/tubereamer.html

[Thecomedian]

http://www.petecornish.co.uk/case_against_true_bypass.html

Some pedals have an input impedance which is far from high in real terms; the input impedance of the vast majority of amps is 1 Megohm (one million ohms) and in my experience there are few effects pedals that have the same input impedance. A load on the guitar of less than 1 Megohm will reduce the volume and high frequency content of the pickup signal thus giving rise to complaints that "this pedal looses tone/volume" etc. Many effects I have tested have an input impedance of less than 100 Kilohms (ie: only one tenth of the amp input impedance) and cause serious signal losses in the effects chain.

http://www.beavisaudio.com/techpages/Buffers/

Buffer or true bypass? The question is as old as some of my socks. If you hang out on forums or google the topic, you will find endless debates about buffers vs. true-bypass. The good news is that it is not a binary decision. Here's what I've found over countless tone experiments moderated by liberal amounts of beer:

    An pedal chain of all true-bypass devices is not always the best thing
     
    A pedal chain of lots of buffered pedals is not always the best thing
     
    A buffer at the beginning of your chain of true-bypass pedals, and perhaps one at the end is a good thing
     
    Your ears, as applied to your setup, are the ultimate guide. Try, experiment, listen, form your own conclusions.

http://www.calculatoredge.com/electronics/BJT.htm

http://screaminfx.com/tech/why-and-when-to-use-a-guitar-buffer-pedal.htm

What this means is that if Rout is large and Rin is small, then the voltage at the input of the pedal is small (which you don't want). An example where this is a problem is a fuzz face pedal, where the input resistance may be 15kOhm or lower and is shown in Figure 2. Your guitar output voltage is about 100mV and the output impedance is about 15kOhm (It's actually a little different because of frequency response but let's save that for a different time). So, this means that:

Vpedal = Vguitar (Rin / (Rin + Rout) = 100mV * (15k / (15k + 15k)) = 100mV * (1/ 2) = 50mV

Only half the guitar voltage actually appears at the input of the fuzz face!

all the figures visualize the issue very well. It makes me believe you'd want the ability to place a buffer anywhere, because perhaps you want the loading interaction that happens between a wah and guitar, or a fuzz and guitar, but you want to then transform the non-ideal output impedance of those pedals lower to preserve the signal into another pedal.

An input buffer is a previous pedal's output buffer, and an output buffer is the next pedal's input buffer.

I'm also of the opinion that it's good to buffer when possible, because you can always take stuff out of the signal chain with EQ, but you can't get that same stuff back without boosting, and boosting closes the noise:signal gap when there's losses.

http://forums.guitarnoise.com/viewtopic.php?f=40&t=47939

[Thecomedian]

http://www.petecornish.co.uk/case_against_true_bypass.html

Some pedals have an input impedance which is far from high in real terms; the input impedance of the vast majority of amps is 1 Megohm (one million ohms) and in my experience there are few effects pedals that have the same input impedance. A load on the guitar of less than 1 Megohm will reduce the volume and high frequency content of the pickup signal thus giving rise to complaints that "this pedal looses tone/volume" etc. Many effects I have tested have an input impedance of less than 100 Kilohms (ie: only one tenth of the amp input impedance) and cause serious signal losses in the effects chain.

http://www.beavisaudio.com/techpages/Buffers/

Buffer or true bypass? The question is as old as some of my socks. If you hang out on forums or google the topic, you will find endless debates about buffers vs. true-bypass. The good news is that it is not a binary decision. Here's what I've found over countless tone experiments moderated by liberal amounts of beer:

   An pedal chain of all true-bypass devices is not always the best thing
   
   A pedal chain of lots of buffered pedals is not always the best thing
   
   A buffer at the beginning of your chain of true-bypass pedals, and perhaps one at the end is a good thing
   
   Your ears, as applied to your setup, are the ultimate guide. Try, experiment, listen, form your own conclusions.

http://www.calculatoredge.com/electronics/BJT.htm

http://screaminfx.com/tech/why-and-when-to-use-a-guitar-buffer-pedal.htm

What this means is that if Rout is large and Rin is small, then the voltage at the input of the pedal is small (which you don't want). An example where this is a problem is a fuzz face pedal, where the input resistance may be 15kOhm or lower and is shown in Figure 2. Your guitar output voltage is about 100mV and the output impedance is about 15kOhm (It's actually a little different because of frequency response but let's save that for a different time). So, this means that:

Vpedal = Vguitar (Rin / (Rin + Rout) = 100mV * (15k / (15k + 15k)) = 100mV * (1/ 2) = 50mV

Only half the guitar voltage actually appears at the input of the fuzz face!

all the figures visualize the issue very well. It makes me believe you'd want the ability to place a buffer anywhere, because perhaps you want the loading interaction that happens between a wah and guitar, or a fuzz and guitar, but you want to then transform the non-ideal output impedance of those pedals lower to preserve the signal into another pedal.

An input buffer is a previous pedal's output buffer, and an output buffer is the next pedal's input buffer.

I'm also of the opinion that it's good to buffer when possible, because you can always take stuff out of the signal chain with EQ, but you can't get that same stuff back without boosting, and boosting closes the noise:signal gap when there's losses.

> output Z = 1K  (nice and low)

How low is low?

Guitar pickup is 5K in lows and 100K at top resonance. For minimal loading everywhere, we like guitar inputs to be 500K-1Meg. (Your 100K hardly loads the lows but cuts the highs down to half.)

Higher output impedance, if flat, just leads to more loss... since our boxes have ample gain, loss may be utterly tolerable even at 500K output impedance...

Hi-Fi interfaces today are often 100-1K out and 10K-100K in...

...Output impedance is a consideration but NOT all that important.

http://forums.guitarnoise.com/viewtopic.php?f=40&t=47939

[GibsonGM]

I simply let my ears do the 'talking'.   IMO there's almost no computational way to know if you should buffer.  Too HIGH of an input impedance sounds, to me --  "glassy" and irritating, robbing the guitar of its full body of tones, if you will.   I only buffer when necessary.  That said, my chain always does contain something with a high input/low output Z, almost accidentally, due to good design by those who came before us. 

Pretty much all existing designs are going to have an appropriate input/output Z, without you messing with things (real FF being an exception, altho many WANT the low input Z.  With a volume bypass cap, I LIKE the changing output Z of my guitar as I turn down....).   Now, say you have one or two pedals that have 'iffy' or 'just passing' input Z....and you get a tiny bit of "tone sucking".    The effect CAN add up, and you will hear it, so you then realize you need a buffer - so you add one!    I just see no reason to automatically follow a Boss pedal, for example, with an input buffer connected to an already-reasonably well-designed overdrive/distortion.

I find that running a crybaby after my Les Paul, and before my (high Z) Guv'nor, cuts some obnoxious highs out that otherwise can be annoying.  So, as with a lot of FF circuits, we can't demonize impedance as always bad!   But maybe should view it as something to be aware of, and manipulated like any other electronic phenomenon....good news is, you can ALWAYS just try it out, see what it sounds like, before you build it on the PCB!!

[amptramp]

The issues that haven't been mentioned yet are interaction of controls and changes to equipment.  If everything is buffered and all controls are safely isolated between input and output buffers, you can make changes and not have them affect anything else.  But if you have an output with a tone control or two going to a volume control straight out of the box, then you may have to adjust the tone depending on the volume or depending on whatever is attached to it downstream.  In some cases, if you bypass the subsequent pedal, you can't get the same response because the subsequent stage becomes part of the response.  Similarly, if you change pickups on your guitar, there may be additional effects from the change if there is no buffer at the output of the guitar.  Some people have settled on a particular combination that works without buffering, but they should be aware that controls may interact and changes may have more of an effect than they wanted.

I would rather add a tone control to a buffer than use input impedance as a way of tuning the response of the guitar, because I can get any response I want with a control, even if I change pickups.  But if I use a new pickup and I was depending on the input impedance of the first stage to tune it, I get a combination of effects from both the new pickup and the impedance it is driving.

[ashcat_lt]

I think amptramp is closest, though honestly Thecomedian is saying about the same thing in a more tl;dr fashion. 

If you want your device to perform consistently with the widest range of sources and loads connected to it, then you need to buffer any impedance sensitive parts of the circuit.  And that's pretty much all there is to it.

[Thecomedian]

Buffering everything is the engineer's dream. Playing with all the unique tonal qualities of mixing different boxes' unbuffered impedances is the artist's dream. PRR made a good analogy with his support-beams-for-house-porch analogy in terms of how much do you really need something vs how much of a waste is it to overdo it.

The Fuzz may never have gotten it's reputation or people being able to find a good sound in it by fiddling practice if it had been buffered on the input.

I think these reasons alone may keep completely switchable isolated buffers or separate simple buffer pedals viable forever. The user can decide whether to buffer your OD/distortion or not, and make their own artistic vision.