What's a buffer, how do they work and why would we want one?

I have always used buffers whenever they seemed advisable - which was often the case.
A buffer is just a circuit that relieves the preceding circuit of any load, while (ideally) not imparting any sonic signature of its own.

It is a gross misunderstanding of what is going on if people think they "hear" the buffer. No. They hear the driving circuit, how it sounds, when it has no load.

The interaction of a circuit and a load, say a pickup with some cable capacity, is not necessarily bad.
A pickup may sound ice-picky without a capacitance that modifies its corner frequency and/or the right load resistance, and sound better with the right cable and the right load afterwards. The problem is that both of them introduce a lot of imponderability downstream.

Take the classic: a fuzz face sounds unexpected when driven with a buffer. Impedance mismatch. Insert a series resistance. All of that can be easily controlled to the point of becoming independent from having to arrange your pedals or cables in a certain sequence.

With that all sorted out, behold the pristine and powerful tone quality of a correctly buffered guitar. This is unsurpassed.

That all stems from a huge lack of understanding of the bigger picture.

But there are also some caveats with buffers: Pete Cornish advocates buffering all pedals. I do not subscribe to that. He comes from
gigantic stages that have endless cable runs and unpredictable connections. (Note: he talks of a buffered bypass...)

As mentioned above, you have to re-think your pedal train. All of that can be controlled, but maybe not so easy with commercially acquired pedals. It is certainly not done with inserting buffers here and there indiscriminately.

As you guys have said before, all buffers should be the same theoretically, but only if they are behaved.
Unfortunately, the often seen unity gain buffer is the one that is most prone to instability (see Ask The Application Engineer #32)
But it applies to any output architecture. This can by the way also happen to discrete devices, so it is said.
The cure for that is basically a simple compensation resistor, often called build-out resistor. This is so dirt cheap that it makes me wonder that it is not used more often.

So back to the question: do you need a buffer? Not if you are happy with what you have and if you enjoy a ball-and-chain.
 
Last edited:
Certainly. I try to say it in my words, since I am no specialist in this field. The stability of the feedback loop depends on the quality of the fed back signal phase-wise. This is ideally 180 degrees or does not deviate much from it. Logically, 0° feedback is feed forward - a perfect oscillator.

External load capacities may introduce some phasing, which in turn may compromise the phase for the feed back signal. A resistor, usually of some 100 Ohms, that goes in series with the op-amp output, together with any parasitic capacities, forms a pole that is sufficiently far away to be of any harm.
This resistor should be close to the op-amp. I have seen a follower buffer made out of a 324 with just a 1m piece of wire at its output. It oscillated wildly and went into momentary latch-up and inversion. Ugly.
More on this in the above referenced document.

I believe (like in many cases, and this has been approved by Doug Self himself) most such oscillations are not of the permanent kind, they are spurious. More felt like ringing and a perceived unnatural "brightening".

As humble as it comes about, I is unlikely that anyone re-visits the buffer stage after the design to look for its stability. I don't, but one should particularly if changing the op amp type.
 
Last edited:
Agreed. If you don't know what comes after the buffer (cable of indeterminate length, another pedal, an accidental short circuit) then a series resistor is a very good idea. 100Ω is sufficient if you're driving a cable. 1K is more appropriate if you're driving a FF or protecting against driving a short circuit. If the buffer is driving an internal circuit, such as most of the TS derivatives, then the series resistor is not required.
 
Just so I understand correctly, if I have a buffer before a FF (say a buffered bypass tuner) I counter the effect of the buffer by putting a small resistor in series before the input of the FF?
 
Exactly. Try it. I suppose you can even tinker up a makeshift "cable" that does this. 1k sounds very small, I always used higher values.
A typical pickup`s DC impedance is about 5-6kOhm. Try a 10k. This way your driving circuit becomes more of a current drive.
Others have suggested 50k, but that may be over the top.

For build-out resistors I have seen values as low as 100R indeed. I have also seen builders use substantially larger resistors, but for a different purpose (see what Mr. Black has to say about What makes switches pop). I see no reason why it should not serve both purposes.

I have described in depth what is going on with buffering and fuzz faces on my web pages: here, here and here.

Ah and yes, those pages (and what I have said above) naturally refer to a circuit's outside boundaries and not to internal parts sensitive to impedance.
 
Last edited:
The FF was specifically designed to be directly driven by a passive guitar. If you're not using that way, then either you like the tone you're getting or you need to add a series resistor, either at the driving end (the preceding pedal), or the receiving end (the FF input). I recommend the latter. In the case of the FF, 100Ω is not large enough to solve the problem. I'd put in at least 1K. You may have noticed that some FF derivatives have either a pot or trimpot between the input and Q1's base.

A notable feature of the original FF is that Q1's base connects directly to the input (OK, there is a DC blocking cap, but you follow my meaning). This configuration is directly responsible for the FF's strong response to the guitar's controls and for RF pickup. Adding the right resistor value, along with an RF filter cap, reduces the RF pickup without substantially altering the response to guitar setting and playing style.
 
You must log in or register to reply here.
Back
Top