Buffers after fuzz = no-no

[DWBH]

I've built a modified fuzz face and while it sounds good on its own, it sounds terrible when there are buffered pedals after it - very trebly, and louder.
I think it may have to do with low/high impedance mismatch. Is there any way to fix this issue?

[Processaurus]

The only thing that could be happening is that your amp has a lower impedance than the buffers in the pedals, as the input stage of your amp is also a buffer.  You could measure it, and then build an output buffer with that impedance into your fuzz pedal, then it would sound the same with whatever impedance you present the pedal's output with.

[DWBH]

Humm... but don't always use the same amp.
Is there an impedance converter? I mean, most buffer pedals (Boss, Behringer, etc) have high input impedance (1Mohm). Can't I *convert* the low (I'm assuming it has a low) output impedance on the fuzz face and make it a high output impedance?

Here's the schem of the fuzz:
http://analogguru.an.ohost.de/001/schematics/Dam_Meathead.gif

[liquids]

If part of your dry 'tone' is a weak, dark signal than yeah.

I used to run long cables through a TB board, compressor on all the time.

I started to learn to live with the compressor off, and re-did my dry tone for clarity, bass content, etc.

Now, when I turned on the compressor (which was buffering my signal) it seemed it was "adding lots of high end sheen."

It wasn't a buffer persay, but was acting similar to a buffer.  My dry tone was very dark with lots of loading.  The compressor on was actually restoring my tone, but since I had already compensated for the dry and loaded down tone, it made it sound unbearably bright and unnatural.

My advise is to tweak around a buffer some where in your chain, as early as you can.  A bufer after a fuzz shouldn't hurt anything, unless it's tone is similarly dependant on loading down whatever comes after it....which is probably just bad design anyway.   

[earthtonesaudio]

The cable between your fuzz and whatever comes after it has capacitance.  Together with the 500k output pot, this creates a low-pass filter which reduces harsh high frequencies.  Using a shorter cable followed by a buffer minimizes this effect, so you get the full force of all those brittle trebles on your poor eardrums.  You could probably add a small-ish cap from the wiper of the output level control to ground and make everything right.

More info: http://www.muzique.com/news/boosters-are-not-buffers/

[JDoyle]

In your circuit the highest output impedance occurs at the halfway setting on the 'dirt' (or volume) control and is a whopping 125k give or take. That's a lot. In fact a buffered pedal with an input impedance of 1M will just barely be adequate to ensure no loading on the signal from the pot itself.

To be honest, lower that pot's value to something in the 50k to 100k range (the low freq. cut off will remain low enough, at least to check the easy mod, and is roughly 72 Hz, and 144 Hz, respectively - and may be a subjective improvement) and see if that changes anything.

A high value output pot like that encounters a lot of issues in regards to output impedance, if it is unbuffered; stray capacitance between wiper and contacts, resulting in a parasitic tone control along the travel of the level control; and lastly, a lot of unnecessary added noise.

And this isn't a knock on the designer - but if possible you may want to check the output on a scope for oscillation. The caps C3 and C4 in the schematic you linked to represent a very direct path for high frequency positive feedback.

Anything fed though C4 at 47pF will easily make it through the 470pF of C3 to the base of the input Q, which has a very low input impedance, being only that of a forward biased diode. This means everything fed back from Q2's collector to its base and is high enough of a frequency where the 470pF looks like a short (which will be most of what was fed back as it had to go through a cap 1/10th the size) will see essentially two paths that it will split into - the two transistor bases, both having essentially the same input impedance. It's other options are the collector of Q1 - way too high an impedance, - and the 18k collector resistor of Q1 - still too high in comparison to the far less than 100 ohms or so of the two parallel transistor bases. In fact, adding the output level pot to the mix, if one looks at where a signal 'wants' to go from the collector of Q2 in this circuit, if it is of a frequency where C3 and C4 can be considered shorts, a conservative guess would say 90% of that signal is split between the two transistor bases. One of which is intended, the other, not so much; and bad things could happen. Even if it is above the audible range, oscillation can really screw things up - and rather than causing actual tonal degradation when a buffer follows, sticking a high impedance effect after this circuit may simply leave the unloaded circuit to its own devices/issues and start some nasty things within, causing the poor tone.

I personally don't think this looks like a very stable design and would first move one of the caps, if not both, to another arrangement, such as bypassing the collector resistors, to bleed off treble - rather than use feedback to alter frequency response.

The reason I say this is, all things being equal, following any circuit with a high impedance should lead to increased BASS, not increased treble so much; everything will be increased due to decreased loading, but the bass respose should also improve (e.g. the -3db curve of the output cap and level pot||buffer Zin). The increased treble could be ultrasonic oscillation causing peaking, ringing, God knows what, etc.

Regards,

Jay Doyle

[liquids]

DBHW would have to tell us how he typically runs the volume pot, but my assumptions were along the lines of most fuzz faces, which, with that kind of biasing--'tapping' signal less that 1K off the V+ (820R here) rather than right off the collector, the volume pot full up is often just above unity, give or take.  Most pedals are at unity with far more series resistance on the pot, sure, but fuzz faces like this are typically different...

And my understanding and experience is that C3 and C4 instead smooth/cut treble, rather than provide a "direct path for high frequencies."

[DWBH]

Indeed, those two caps smooth it out, and keep the circuit from sh*tting itself into oscillation and *uncontrolled* mayhem.

[DWBH]

To be honest, lower that pot's value to something in the 50k to 100k range (the low freq. cut off will remain low enough, at least to check the easy mod, and is roughly 72 Hz, and 144 Hz, respectively - and may be a subjective improvement) and see if that changes anything.

I took Jay's advise, and lower the pot down to 100k, but unfortunately it didn't change anything.

[DWBH]

The cable between your fuzz and whatever comes after it has capacitance.  Together with the 500k output pot, this creates a low-pass filter which reduces harsh high frequencies.  Using a shorter cable followed by a buffer minimizes this effect, so you get the full force of all those brittle trebles on your poor eardrums.  You could probably add a small-ish cap from the wiper of the output level control to ground and make everything right.

More info: http://www.muzique.com/news/boosters-are-not-buffers/

So I have no pedalboard at home right now. All the gigging gear I use is at rehearsal room.
I do have the fuzz, a patch cable, a long cable, an amp and an extra guitar.
So I plugged the fuzz to the amp. Long cable to the pedal, patch cable to the amp. Brittle, rapsy sound.
Then, I swapped cables. Patch cable from guitar to pedal, long cable from pedal to amp. More full now.
So I guess the capacitance of the long cable is smoothing out those frequencies. As per earthtonesaudio suggestion, I add a smallish cap (20pF - 470pF) to simulate the long cable capacitance.

[JDoyle]

And my understanding and experience is that C3 and C4 instead smooth/cut treble, rather than provide a "direct path for high frequencies."

They do both. They provide a 'direct path for high frequencies' at the collector to feedback to the base, and as they are inverted with regard to the signal at the base, they cut treble via negative feedback.

The cap from collector to base is standard issue in hifi amplifier design. It is simple, effective and cheap; and as long as the diff. amp that precedes it has the current reserve available to charge it and avoid slew rate limiting, it is pretty much the best way to compensate a power amplifier.

It is by no means a bad practice for a single stage, but two in a row is playing with fire in my opinion.

DWBH - if your circuit oscillates without the caps, the problem lies with the physical layout, not the circuit itself.

Regards,

Jay Doyle

[DougH]

A lot of fuzz face circuits have pretty high output impedance. I've never completely understood that. IME, high Zout volume controls will give you a built-in high frequency rolloff that, as you found, will change the tone depending on the cable, cable length, and what is plugged in after the circuit. They can also provide inconsistent frequency response, depending on how high the volume is set. (I think Jay may have mentioned that.)

This is all okay if you basically always set the volume the same, always use the same cable, always plug the same thing in after the effect. But if you change any of those variables- all bets are off. It's better to ensure some decent current drive from the output by either using a smaller pot or buffering it. Then provide whatever rolloff is needed to compensate, elsewhere in the circuit. Another problem with the "direct path" of C3/C4 that Jay mentioned is that in general it's best to avoid having two caps in series IMO. This can cause weird phase-shift problems, and given that it's in a feedback loop that's at least partially a positive feedback loop, who knows what that can be doing to you.

[edvard]

I'm pretty sure Doug is right, as I've encountered something like this twice last week with some stuff I was breadboarding.
Once in a two-stage cascaded mu-amp FET circuit I was testing, and again with a Harmonic Percolator mod.
Both those circuits have a rather high output impedance and I read on this board somewhere that when you load a high-output-impedance circuit by following it with a mid to low-ish impedance, it will cut the gain of the last stage in the pedal.

So what is probably happening is that you're used to the 'raw' sound that comes from going guitar->pedal->amp with the amp providing a reasonable but not terribly high impedance input.
This is probably how the designer tested it, so it sounds 'normal' in that configuration.
When you follow it with a very high-impedance pedal (like a buffer, which typically provides 1 Megohm or better) suddenly your high impedance output isn't loaded anymore and UP goes the gain and everything else with it.

I saw this very thing happening when I plugged my Percolator circuit into a chorus pedal with 470k input impedance.
Suddenly my sound was weaker and 'flabbier', not the tight scrunchiness I was used to.
Just for giggles, I put in a FET source-follower in between and suddenly I had a TON of gain and brightness out of the Perc.
I had to put 47k on the gate of the FET to bring it back down to reasonable levels.
In the cascaded mu-amp circuit I noticed a good sized gain loss between stages unless I put in a high value pot or a source-follower.

It looks like your circuit is also a high-output-impedance variety, evidenced by the 500k volume pot.
I think Jay's advice to reduce the pot value is sound advice, especially if you plan on using your buffered pedal in that chain all the time, but try even lower values.
Or you could add an emitter-follower between Q2 and the pot so YOU can set the impedance that sounds right and the volume pot can be something reasonable.

[JDoyle]

I read on this board somewhere that when you load a high-output-impedance circuit by following it with a mid to low-ish impedance, it will cut the gain of the last stage in the pedal.

Remember that though the gain of a transistor stage is normally defined as Rc/Re (Rd/Rs for FETs), in reality Rc'=Rc||Rload; where Rc' is the effective load at the collector/drain (the actual value that should be used in the formula), Rc is the collector resistor, and Rload is the input impedance of the stage following. So, if you have a high input impedance following the stage (say 1M), in most cases that won't change the gain all that much (but note that it always will change it some), but if the input impedance of the following stage is the same value as Rc, then the gain of the stage is cut in half, for example.

All single stage high gain circuits suffer from this because in order to get high gain, you need to have a high impedance load and that load will be effectively reduced by the impedance of anything following it.

R.G. pointed this out in his write up on the SRPP - the gain would drop (by half if I recall correctly) with even a 1M load following the stage.

[shredgd]

The cable between your fuzz and whatever comes after it has capacitance.  Together with the 500k output pot, this creates a low-pass filter which reduces harsh high frequencies.  Using a shorter cable followed by a buffer minimizes this effect, so you get the full force of all those brittle trebles on your poor eardrums.  You could probably add a small-ish cap from the wiper of the output level control to ground and make everything right.

I can confirm that this is the real issue, not impedance differences between buffers and amps (which are roughly the same).
I experimented with small caps from signal to ground after the fuzz circuit (but before the buffers), and I can now get the exact same tone as I was using an all-true-bypass arrangement. You can tune the exact value of the cap to taste: in my case my 5 meters cable from pedalboard to amp had a measured 600pF impedance, so I could get the same tone by using 470pF + 150pF in parallel to get a similar value.

It's incredible to me that no modern fuzz pedal exists with this feature (i.e. a switch to choose between standard - for true bypass pedalboards or people using just the fuzz pedal - and cap to ground at the end of the circuit - for people using one or more buffers after the fuzz)! Or does it? 

Giulio

[bool]

I used the RC LP at the end of discrete circuits like fuzz or a booster since 80s: it's very simple to re-calculate the RC constants and to add the filtering cap.

It's also easy - but dosent work nice for all of the circuits - to change the output ("level/volume") pot to smaller value, such as 25k or 50k and scale the output cap accordingly.