Output buffers - Aren't they a MUST-DO thing?

[Labaris]

Hi everyone.

Many pedals have the level knob (a.k.a. volume) right at the end of the circuit.
This I think is not the best choice, since it makes the output impedance variable (not sure about that anyway).
I had problems with this in Marshall Guv'nor for example. With different pedals connected to its outputs, different behaviors were observed (more or less volume). I put an output buffer to it, the one in AMZ high gain overdrive, and now the pedal is really "independent", with tons of volume if needed (before the mod the vol. knob was always at max.)

Opinions? Thanks!

[merlinb]

Opinions? Thanks!

I agree with you, variable (and high) output resistance is not acceptable in a modern pedal. As you observe, it makes performance unpredictable, and totally rapes the signal/noise ratio. A simple opamp buffer costs nothing to implement. Pedal designers should take more cues from mixing desk designers.  

(I have a Marshall Guv'nor too. I found it to be the most pathetic, reedy-sounding fuzz box ever. Nothing like the "high quality distortion tones" it advertises. Did I just get a bad apple?)

[slacker]

I think the problem with the Guvnor is more to do with the fact that the whole tone stack is hanging off the end of the circuit and is unbuffered, so you will get interaction with whatever you plug it into especially that has a lowish input impedance.

If you just have a volume pot on the end of a circuit driven by something with a low output impedance like an opamp then I don't think there's a problem.

[R.G.]

It all depends on the circuits before and after the ending pot..

A pot presents a variable impedance to both what drives it and what follows it. When the pot is fully up, the pot resistance is simply in parallel with the external load following the pot. As the pot is turned down, it quickly becomes a T network with the upper part of the pot and the load resistance being the top two series resistors and the bottom part of the pot being the leg of the T. To the load, the pot goes from simply being in parallel with the following load to being the parallel combination of the two sections of the pot as defined by the wiper position, that being in series with the output impedance of the effect driving it.

The rule of 10:1 applies. In general, if the following impedance (that is, whatever is connected to the pot) is more than 10X the pot resistance, you can ignore it in terms of loading either the pot wiper or the effect driving the top of the pot. That's not perfectly true, but it generates only small errors. If what follows the pot is 1M or over, and the pot is 100K or less, you can ignore the circuit following the pot wiper with only minor errors.

Likewise, if the output impedance of the effect driving the pot is less than 1/10 of the pot value, you can ignore the loading of the pot. If the effect has an output impedance of 10K or less, any pot of 100K or more can be ignored as not loading the effect significantly.

If those 10:1 conditions aren't true, the loading will generate more than 10% errors and you would be well advised to at least think about the results.

Putting a buffer on the output of an effect amounts to saying "modify the circuit with more parts to reduce its output impedance to negligibly small".  This makes the 10:1 approximation true for pot loading of quite small pots. Putting a buffer after a pot makes the 10:1 approximation true for even quite small loads on the wiper of the pot.

All that being said, there is no substitute for understanding your circuit, what its output impedance is, what loading is connected to it, and what input impedance it drives. Details matter in a way that generalities simply can't.

So are output buffers a MUST-do thing? No.
Is knowing that you can drive whatever is most likely to be connected to the output with small losses a MUST-do thing? I think so. There is always a value to knowing.

[Labaris]

Thanks to everyone for your opinions.

merlinb:

Do you have a stock Guv'nor or a DIY version?
Mine sounds great now. I did some other mods to it (tonestack included) that maybe you'd like to try.
Is yours SO bad?

slacker

I think you're right. All those pots coming from the clipping stage, getting again into another pot... that's not a good thing

R.G.

Thanks for your explanation Mr. Keen.
And I understand that details matter a lot. What is connected before and after the output.
But I'm still thinking it would be valid to search for a general rule for analysing outputs behavior, when designing them or when trying to modify/fix something in an existing circuit. Because you will not always know what exact circuit (load) is connected to every pedal.
So, how can I "prepare" an output stage to be as good as I need, no matter the connections between circuits? Could output buffers be a wrong solution in some cases?

[R.G.]

But I'm still thinking it would be valid to search for a general rule for analysing outputs behavior, when designing them or when trying to modify/fix something in an existing circuit. Because you will not always know what exact circuit (load) is connected to every pedal.
So, how can I "prepare" an output stage to be as good as I need, no matter the connections between circuits? Could output buffers be a wrong solution in some cases?

That's an interesting question; it carries inside it some assumptions you probably don't realize you've made.

For instance: "you will not always know what exact circuit (load) is connected to every pedal."
That's absolutely true. And in fact you can NEVER know what circuit might be connected to every pedal. Someone might just plug in their 2 ohm speaker cabinet. Or a "clever" guitarist/hacker may design an absolutely great amplifier with an input impedance of 10 ohms and want your pedal to work with it. Your mind is right now saying "but that's silly; no one would ever do those or expect them to work properly if they did." And you're right. So implicit in that last thought is the idea that whomever uses your pedal will connect a reasonable load to it - but not what a reasonable load is.

We can only guess. Guitar amps are usually 1M resistances in parallel with a vacuum tube grid. Unless they're solid state amps from the Dreadful Solid State Era, and are 100K or less. And PA or other amps may be 10K to 100K. Do you put in a buffer just in case?

Then there's buffer performance. Many players are horrified if you put a buffered bypass into a pedal, so there's some aversion to buffers in general. And a buffer in front of some well known pedals make them sound different than expected. The buffer in front of a pedal can be the buffer at the output of your pedal, so your pedal gets a reputation for not working well with a Fuzz Face or clones, just to note one of the pedals that doesn't like a buffer in front. And many pedals have an up-front buffer of 1M or more, so they don't care whether your pedal has a buffer or not.

It's not a simple question. Well, OK, it is a simple question, but the answer is is NOT simple.

"as good as I need, no matter the connections between circuits".

And that's another one of those questions. How good is "as good as I need"? And in what units do we measure "goodness"? Enough treble or bass without too much? The now-infamous "tranparent"? Enough volume? Exactly the right amount of distortion?

It is indeed valid to look for a general rule. Unfortunately that rule is "know as much as possible and get most of the cases right if you can", because the things you're trying to make better sometimes conflict. Maybe you need two outputs - one buffered, one not. That costs you a switch or another output jack, and now you have the opportunity to set it up incorrectly too.

[merlinb]

I disagree with RG, I say buffers should be mandatory.

The fact that some pedals don't like to be driven by buffers (e.g., Fuzz Face) is their problem. They are primitive, out-of-date designs; dinosaurs in the modern music environment. They should be should be brought up to date, so that they have their own input buffer followed by the necessary "good sounding" impedance right before the actual fuzz circuit. The whole thing should be one, good sounding, self contained box. FX boxes are inherently building blocks; it is unfair to the musician to design them so that an individual effect sounds bad when combined with another particular FX box, or good when combined with a different box. I shouldn't have to 'match' my effects pedals, and I shouldn't expect certain combinations to sound bad since that restricts the combinations that are available for use.

Also, RG didn't mention the noise issue. Loading is only half the story. Sure the signal level from a 10k output pot is not going to be terribly affected by a 1M load, but extraneous hum and noise is going to find it easy to couple into the cable. By buffering the output, any noise is shunted away, preserving the signal to noise ratio.

If all pedals had buffers then we wouldn't need to 'fix' the sound with noise gates and true bypass. I don't understand why guitarists don't expect the same level of performance from their gear as sound engineers do.

[earthtonesaudio]

Also, RG didn't mention the noise issue. Loading is only half the story. Sure the signal level from a 10k output pot is not going to be terribly affected by a 1M load, but extraneous hum and noise is going to find it easy to couple into the cable. By buffering the output, any noise is shunted away, preserving the signal to noise ratio.

I'd love to see this noiseless buffer you speak of.   

I don't understand why guitarists don't expect the same level of performance from their gear as sound engineers do.

I think you could find at least one answer to that enigma by examining the Fuzz Factory.  Here is a pedal with a bunch of "interactive" controls, which is difficult to set the same way twice and therefore encourages experimentation.  I think many guitarists like being surprised by how their gear works (or doesn't work).  Maybe it gives them a sense that there's something alchemical going on, something magical.  Maybe they just don't know any different. 


I might be in the minority of people here, but I find bufferless, impedance-mismatching devices just as useful as pedals with "proper" input and output buffering.  I don't find it a hindrance to achieving a particular sound, because I already know what to expect.  I think in discussions like these, for better or worse, it is easy to slip into a conversation where we are discussing what is best for "guitarists in general" and not so much what is best for "me."

[petemoore]

  Margueritaville.

[anchovie]

I might be in the minority of people here, but I find bufferless, impedance-mismatching devices just as useful as pedals with "proper" input and output buffering. 

I'll join your minority!

I don't use any output buffers and when I'm playing with PT2399 circuits I don't bother with an input buffer. Some people have done so much work with those chips to filter the high end in order to avoid excessive noise in the delay line - I just let the IC suck some tone naturally!

[R.G.]

I disagree with RG, I say buffers should be mandatory.

Differences in opinion are welcome.

My first recollection of a statement of how effects "ought to be" is Craig Anderton's exposition of a (as I recall it) "FX Standard" involving input impedance, output impedance, noise standards, frequency response, etc. These were all good ideas for interoperability - and the "standard" was universally ignored, partly because there are no Effects Police and partly because the developing gotta-have-a-real-vintage-pedal movement engendered a hunt for precisely those pedals which didn't match up.

The fact that some pedals don't like to be driven by buffers (e.g., Fuzz Face) is their problem. They are primitive, out-of-date designs; dinosaurs in the modern music environment. They should be should be brought up to date, so that they have their own input buffer followed by the necessary "good sounding" impedance right before the actual fuzz circuit.

I agree. They oughta be. However, that isn't what sells best, and worse, the world is littered with originals and copies which are still primitive and out-of-date. Changing the way the world already is is harder than making a few things work right.

FX boxes are inherently building blocks; it is unfair to the musician to design them so that an individual effect sounds bad when combined with another particular FX box, or good when combined with a different box. I shouldn't have to 'match' my effects pedals, and I shouldn't expect certain combinations to sound bad since that restricts the combinations that are available for use.

Once again, I agree with the motivation, but the real world doesn't go along. When I design an effect, the input impedance is uniformly 1M or greater; it has pull down resistors on both ends of the effect even though it would not pop on switching as a preventative against external switching setups; the output impedance is under 10K for driving cables but more than 1K to protect against shorts; the inputs are immune to most forms of RF; the power is immune to DC reversal and the connection of AC power adapters; the thing does not go into oscillation at any control setting, power supply level or with its input open (unless it's supposed to do that!); it doesn't send junk back up the power line, and is immune to a certain amount of junk sent to it on the power line - and in some instances, the ground line; the list goes on. My pedals don't have to be matched.

However - imagine trying to convince Eric Johnson that his hand picked Fuzz Faces would be improved by adding in a buffer. And imagine trying to convince Eric's fans and imitators that Fuzz Faces are better when they're modified to be "better" when that's not what EJ uses. Or Hendrix imitators that the eccentricities of connecting a FF and a wah are just wrong. In many cases, making a effect work technically better is not what the naive guitarist wants. He wants it to be "original" more than he wants it to be "right".

Also, RG didn't mention the noise issue. Loading is only half the story. Sure the signal level from a 10k output pot is not going to be terribly affected by a 1M load, but extraneous hum and noise is going to find it easy to couple into the cable. By buffering the output, any noise is shunted away, preserving the signal to noise ratio.

That's true. On the other hand, buffers themselves generate noise, unavoidably. Modern parts are better than they used to be, and you can use advanced biasing techniques and low noise resistors, but self generated noise in the buffer itself is always there. I personally design for driving cables with low impedances, generally about 1-2K, for the reason of avoiding treble losses and noise pickup, but buffers can cause some interesting noise issues when put in front of a very high gain chain of distortion pedals.

If all pedals had buffers then we wouldn't need to 'fix' the sound with noise gates and true bypass. I don't understand why guitarists don't expect the same level of performance from their gear as sound engineers do.

Whenever I see people doing things that don't make sense to me, I think that I don't know all the rules of whatever "game" they're playing. Guitarists are almost the opposite of sound engineers. They don't care what the equipment ought to be doing as long as it sounds "killer" or "... just like Eric Johnson" or "just like (insert other famous guitarist here)" and not only don't understand that there are technical explanations for how that happens, but actively don't want to know in some cases. They're basically emotional in their response to putting out music - and that's good because good music is about emotion.

[merlinb]

Well it sounds like RG and I at least share the same expectations, but perhaps he is a little more liberal then me. I would rather start producing classic clones that have all the mod-cons, and simply not tell the guitarist what's inside (beyond what he expects). Assuming I can design it so that there is no negative change in performance. Eventually we would get to the utopia where everyone is so used to the better, modern performance, that the vintage-o-phile mentality will be tempered, and you can then safely advertise the pedal as being up to date and generally sooper-dooper-better-than-the-original.

That's true. On the other hand, buffers themselves generate noise, unavoidably.

Yes, but so what? An opamp buffer has a noise figure that is so vanishingly small as to make point moot (and I'm not sure why you bring it up?). The noise contributed by even simple buffers is truly puny compared with the noise you get by not using a buffer. I would never suggest a buffer is noiseless; what I am saying is that it cures a lot more noise than it creates, on every count.

[Jay]

Roger Mayer has been remaking all his pedals with buffered (or true bypass) outputs for a while now:

http://www.roger-mayer.co.uk/guitar_effects.htm

And he does do a Fuzz Face with buffered input as well as output.

http://www.roger-mayer.co.uk/classicx.htm

So I wonder how well these sell, versus his 'original' non-buffered version?  i.e. do his customers prefer to buy the 'real deal, warts and all' version or the 'improved' one?

Interestingly, for those wanting the classic Jimi Fuzz Face sound, Roger used buffers around them in the Studio anyway - so  you're hearing a buffered effect on the first couple of Hendrix albums.

[liquids]

I'm not entering the debate.  Just going to share what I do, because it works for me.  At this point, currently everything I build has a buffer (or buffering) at the output, especially following passive voltage divider type volume controls.  Most of what I've built also has a buffer right at the beginning of the circuit for consistency, as well (though I may no longer need to do that), as I'm currently convinced that is what is 'best' for what I do and want from the guitar, the issues notwithstanding.  

Recently though, I got around to building and soldering up that "alwayson" buffer (with an on-off toggle) at the beginning of my chain.   It turns turns out I really didn't appreciate the difference in tonality the buffer right at the input afforded me (even dead clean).  Unfortunately, I thought the "alwayson" buffers that I had a bit further "down stream" in my chain were doing the job in the meantime...but I was wrong!  

So I've already built some stuff around that (no input buffer) tonality with that incorrect assumption.  So I ended up 'correcting' my simple input buffer with an additional passive R-C filter / buffer stage, so that in terms of high frequencies, I can simulate what I like about the high end 'suck' I've become accustomed to (!) when I compare the now buffered to old unbuffered signal, with the benefits of driving everything with a low output impedance signal, which actually helps make the tone it seem more 'punchy' in the lows, to my ears, and of course, more consistent.  

Downstream I've got at least one more effective alwayson buffer, and then every time a DIY pedal is on it's buffering too.  This is the most tonal consistency I've had yet, and I love it.  

It taught me again/further that most guitarists have become quite used to the sound of passive pickups/loading/tone suck in many cases, reliant upon it, and we like that sound (it's near and dear to our hearts).  So even though I've been known to chant the mantra here and elsewhere that buffers "don't add high end, they merely restore your tone," practically, adding some filtering in there at the beginning helps to persuade the ear - or at least mine.

I'm happy I build my own stuff for myself and currently for no one else, because this issue really causes headaches and tonal inconsistency, given how many variables there are, and most players are completely (and understandable) unaware unless they have the right schem to every piece of gear they have, let alone understand it.  At least I know what I've got, sort of know what I want, and can build around that.  Helps me sleep at night.  

[merlinb]

 So even though I've been known to chant the mantra here and elsewhere that buffers "don't add high end, they merely restore your tone," practically, adding some filtering in there at the beginning helps to persuade the ear - or at least mine.

Here here.

Buffers prevent tone suck, which you might otherwise want, but they give you the ability to control the suck. It is easy to artificially add some tone suck with a filter, to simulate you favourite impedance combination (but with less noise into the bargain), but you can't 'add back' fidelity which has been lost with an unbuffered set up.

If I need some tone suck to get where I want to be, then I want to have full control over it. I don't want to be at the random whim of unknown input/output impedances, different cable capacitances and so forth. I demand repeatability, and also access to tones that are unobtainable with an unbuffered set up. Buffers only broaden the available palate, they never reduce it.

[B Tremblay]

He wants it to be "original" more than he wants it to be "right".

In a word, mojo.