Pedal Noise? (a very in depth question)

[Joe Hart]

What are the worst offenders in the noise department? I know that IC's and transistors can be noisy, but some just sound better despite the noise! Does anyone have any "numbers" on what makes noise?

I mean, how much more noise does a carbon comp resistor have compared to a metal film? Twice the noise? Seven times the noise?

How many metal film resistors would cancel out the noise benefit? Say I had to get an odd value and had three metal film resistors that added up to the correct value, and also had one carbon film resistor that was correct -- which would be less noisy? If space wasn't an issue?

And what about caps?

Any ideas? I know that you use your ears, but say that you were really obsessive compulsive (not me -- a friend of mine) and needed to know? Where should I -- uh... I mean, my friend -- draw the line?

Thanks.
-Joe Hart

[shtgoosephour20]

i wish i was wanted the only one responding. i get the idea that ic chips are a variable like pickups or body wood that contributes to the qualitys of the sound. the resistors seem to matter as well, maybe like strings and the caps seem to be adjusting the tone, higher or lower.
i want someone to flat out tell me how to make my pedals better. what little things to i need replace to make it seriously rock. compressors, overdrive, wah...

[Johnny G]

well from personaly experience i think that electrolytic caps in the signal path add noise. i modded my DS-1 with the mods from fuzz central and while it could be a combination of metal film caps and resistor changes, i think it was morethe replacement of those electrolytics

fraid i cant really give any more info than that tho.

[MartyMart]

I guess you "draw the line" under how much cash you want to spend !
I've built about 15 stompers now and have used the best quality parts from the beginning, after reading about components here first !
I dont have a single "noise" issue other than my "Blue Clipper" which is a bit "hissy", I think thats the nature of that circuit.
I've posted before about "quality" and "sound" from good components, so from my fairly limited but "picky" experience, subbing a couple of parts for "best" quality seems pointless, however making the whole build from quality parts does make for a good "tone" and quiet build.
I only use "electro" caps when there is a big value, otherwise its all Tants, Metal/Poly film caps and 1% metal film resistors.
I have not seen "carbon comp" resistors around for a long time and dont see the point of that "mojo"   ie: the "noise mojo"  -- at all !

Cheers,
Marty.

[Joe Hart]

Okay, so electrolytic caps can be bad. When and where can I replace them with metal film (aka non-polarized caps)? I know that the general consesus is "almost always," but what is the "almost" part?
-Joe Hart

[vdm]

hey,
sorry im no big help, but ive been toying with my bluesbreaker and put it back in box this evening.

I just love the tone that pedal spits out, but as soon as i turn the tone up with some treble it gets so noisy. i've tried shortening the wires and moving them to different places but i just cant get a good sound. tried all different IC's as well.

Maybe I'll build another sometime soon with high quality parts... if only i could get them here in australia easily :s... i guess smallbear is the best option.

trent

[Torchy]

crp

[R.G.]

What are the worst offenders in the noise department?

1. Resistors have a fixed amount of noise which depends on their value and their temperature. The lower the value and lower the temperature, the less thermal noise. Some resistors approach the theoretical minumum (wirewound and metal film) and some have excess noise above this amount (carbon film to some extent and carbon comp a lot). The fix? Don't use carbon comp in input stages.
2. Reactances (inductance and capacitance) do not create noise. None, nada, zip. Only to the extent that they incorporate non-reactances (that is, equivalent series resistance) do they have noise.
3. Arcs make noise, even tiny ones. Electrolytics leak a little current, and that makes for some noise. In general, this is incredibly miniscule, except for electros that have been heat or mechanically damaged, or which have dried out and become leaky. This noise is from micro-arcs inside doing the leakage. The fix for cap noise is to use good caps that don't leak. Electros are usually NOT a problem.

Bad soldering **is** a problem, as it's inherently a leaky, arc-y situation. So are dirty, contaminated PCB surfaces.
4. All semiconductors make noise which depends on - yep, their composition and temperature. Leakage is noisy, and surface leakage on older point contact and alloyed germaniums is truly ugly. Modern passivated bipolars are incredibly quiet by comparison.
5. FETs of all stripes have lower noise than bipolars at high impedances because their gates do not allow input currents to creat noises. They also do not have recombination noise because they have no minority carriers to recombine. FETS usually beat bipolars for noise at impedance levels over about 10K to 100K.
6. Bipolars have significant input noise and recombination noise. However, their low input impedance allows them to be used at very low impedance levels which makes them less noisy for certain low-impedance sources, like perhaps moving coil phono cartridges. Bipoolars usually beat FETs for noise below 10K impedance levels.
7. The input stage of any circuit essentially determines the noise for circuits with any significant gain that does not contain excess noise resistors or faulty electros, arcs, or faulty semiconductors. All of the noise at an input stage is amplified by the rest of the circuit and therefore has a "noise advantage" by the rest of the circuit's gain. **Noise is an input stage problem** unless there's a defect in the rest of the circuit.
8. Distortion devices make noise worse because signal is clipped to a certain level, while noise, being usually below clipping, is amplified linearly by the whole (usually huge) gain of the circuit. So noise is amplified selectively.
9. Compressors are noisy by nature because they have their greatest gain when signal is lowest. This is inherently a noise-boosting condition.

i want someone to flat out tell me how to make my pedals better. what little things to i need replace to make it seriously rock. compressors, overdrive, wah...

Sorry. If it was that easy, I or someone else would already have made the best sounding stuff and we would not build our own. Moreover, your ears don't hear "better" the same way other ears do. So all you can do is learn and change things a bit to suit you. There are no pat answers.

well from personaly experience i think that electrolytic caps in the signal path add noise.

Defective or leaky electros do. Good quality, well soldered ones don't.

I have not seen "carbon comp" resistors around for a long time and dont see the point of that "mojo"

Read http://geofex.com/Article_Folders/carbon_comp/carboncomp.htm
for what the mojo is and isn't. If your experience is limited to low voltage effects, no, you can 't hear it. If you play with high voltage vacuum tube amps, maybe.

Okay, so electrolytic caps can be bad. When and where can I replace them with metal film (aka non-polarized caps)? I know that the general consesus is "almost always," but what is the "almost" part?

Whenever you can afford the generally more expensive film resistors and whenever you can physically fit them on the board and inside the box. Electros are incredibly volume efficient for the amount of capacitance, and they are made from cheaper materials than other types of caps.

Notice that in addition to electros being bad sometimes, soldering can be bad and make noise, and so can transistors, ICs, resistors, and you can have cracked resistor leads, and board cracks to make noise. Also notice that a noise defect that you can't hear on the output stage will drive you mad if the same problem is on an input stage. For how transistors and ICs can make noise, see

http://geofex.com/circuits/when_good_opamps_go_bad.htm

. It's entirely possible to damage perfectly good parts by installing them with high heat or poor soldering, and it's possible that the **circuit design** damages the parts a little in terms of noise performance every time it's turned on by abusing the inputs.

[R.G.]

Whenever you can afford the generally more expensive film resistors

...capacitors, capacitors, capacitors, not resistors...

Someday I'll learn to think while I type.

[RDV]

More gain = more noise. Even if you do your soldering correctly and you use the lowest noise parts that you can, if it's got lots of gain, then its got a lot of noise. We can only hope to minimize it, but you can't eliminate it. Listen to the clips on my site, the more gainy the device, the more noise is generated. The worst is the Easyvibe/BSIAB2 clip, good old whooshing hum. I try to not let it bother me.

RDV

[Mark Hammer]

There is internally generated noise, there is "exported" noise, and there is "imported" noise.

Since the lion's share of people here seem to have an obsession with high gain and distortion, all of these apply equally.

If you have a high gain stage (and let's be ridiculously arbitrary and declare that anything with any internal gain over 20 is "high" gain), then it not only amplifies whatever noise is produced internally, but it also amplifies whatever noise it imports by that same factor.

If one pays careful attention to component selection and type, then internal noise can be kept to a minimum.  If the output stage is carefully designed and goes to a well-designed input stage, then little of that internally-generated noise is exported to the next stage.  

A good case in point is any sort of BBD-based device.  These things generate all manner of audio garbage, but if the pedal is well-designed, including both circuit, layout and component choices, and well set up, very very little of that potential audio garbage flying around on the inside ever makes it to the outside.

That the pedal takes care of ITSELF, however, is separate from whether it can successfully eliminate or exclude any noise it imports from other devices.  Again, the best illustration of this is a compressor.  It is assumed that any commercial compressor will be capable of processing a reasonably wide bandwidth so that it can process most instruments and be sold to the widest audience.  Acoustic guitars, drums, and synths can have several octaves more bandwidth than many electric guitars and basses, so there is very little ceiling on the high end that most commercial stompbox compressors will accept.

Unfortunately, by default, compressors work by cranking up the gain on everything until otherwise directed to do so.  If the signal they get is chock full of hiss and hum, AND that hiss and hum level is sifficiently low enough to be lower than a normal instrument signal, then that stuff will be amplified by the maximum gain of the compressor when you stop playing.  No matter how good the compressor is, and how much you've spent buying the finest caps, chips, and resistors you could locate via Absolute Sound magazine, if it amplifies a lousy input signal by a factor of 20 or 40, it will be a noisy pedal and export that noise elsewhere down your signal chain unless there is some means for gating or filtering it out.

A context where all these things meet, and can be beaten by good design, is the typical high gain distortion pedal.  "Distortion", from the musician's standpoint, is simply added potential harmonic content which we ration out by means of our finger/palm muting, picking style, electric adjustments on the guitar, string choices, and note choices, so that there is sometimes more and sometimes less.  

The harmonic content that is potentially added is not ALL musically useful or desirable, though.  Mostly, we want a number of lower-order harmonics, and in certain proportions to each other.  We are NOT indiscriminate about which harmonics we want, or about their weightings.  In fact, there is a large segment of potential harmonic content we are decidedly NOT interested in.  That's good, because it means that can be filtered out.  This happens in a few ways.  For instance, if I *know* I'm going to be adding lower-order harmonics of notes, do I really need to take in more than a few khz of bandwidth, or can I safely filter out pretty muich all the top end?  After all, why do I need to produce harmonics of harmonics, and if the clipping is going to give me back all those frequencies in the 7khz and up range anyways, why do I need to retain them as part of my input signal?

That's great, because it means that any acquired hiss that comes along for the ride as part of the input signal can be safely filtered out right at the input of the pedal.  Don't need it.  Don't want.  Can easily live without it.

What about the output part of this hypothetical pedal?  How much bandwidth do we need or want?  In most instances, not all that much.  Indeed, we have even have a name for unwanted excess bandwidth.  We call it "fizz".  Because full bandwidth is not needed, we can actually safely filter out the excess at the output stage of the device.  One of the side benefits of this is that whatever imported noise that has been amplified can be reduced, and whatever internally generated noise that may have arisen because of component choices (or limitations) and high gain, can also be reduced.  

Fuzz boxes CAN be very very noisy, but they don't have to be, and don't have to trade off character for noise specs, nor do you have to be a scientist or rare-component hunter to accomplish it.  Hell, in a lot of cases, sometimes all you need to do to make things far more manageable is stick a cap across the input and ground of the output-level pot, or identify the last op-amp (if there are any) in a device and stick a small value cap across the feedback path to reduce the gain for high end.  

It is important to realize that many commercial devices are (or were) designed by folks who know an awful lot about audio and design in general, but perhaps much less about musical needs of guitarists.  I've seen plenty of things where feedback caps were used that rolled off high end at maybe 33khz to stop layout-related oscillations.  Okay, fine, but of what earthly use is that content between 8khz and 33khz and why export it?  Beats me.

When trying to beat the noise game in your rig, it is worth knowing a bit about which components MAY matter with respect to generating noise internally, but it also pays to know and consider issues of bandwidth and what you do and don't need to either take in from earlier stages, or export to others.

[petemoore]

This may sound 'stupid' but here's the 'answer' I use.
 Bypass switch for noisy effects.
 I build many high gain devices, [sometimes I build a type more than  twice] then 'race' them to see which get the gain/noise ratio to sound I like. Every one has a bypass switch, stomper preferrably, and I may even hit that [to eliminate noise during pause] for a short pause in a music passage then again when the Lead resumes.
 Some circuits have bigger noise issues than others, using high end parts to build the ones with alot of noise [selectively at input or noise areas, or throughout the chain] can't hurt, seems to make a difference, before the 1rst active stage in the chain being a good place to start.

[Davide]

Always think to resistors like they are an ideal resistor and a little generator (of tension or current.. as you want to see it).. caps cant be noisy, at all.

[Joe Hart]

Wow! My posts always seem to pull out a lot of intelligent discussion -- and for that I thank you -- it's always very informative.

So, I was asking for a couple of reasons. First, for building effects (of course). But also for modding existing ones. I have some Boss pedals that are fairly noisy. A digital delay (but I can just use a bypass box for that one, because I don't use it all the time), and a chorus (CE-5). Unfortunately, I opened up the chorus and it's all those pesky, tiny surface mount components. What a bummer!

I usually run my guitar into a Crybaby Wah, a Phase 90, a DOD 250 clone, a stereo chorus (as a splitter as well as a very subtle chorus), a stereo volume pedal, and a stereo digital delay, then into two high gain amps. The 250 and the chorus are on all the time, the Phase 90 is on a lot, and I use the wah a lot as a filter. So, as you can see, there is the potential for a lot of noise. I was hoping that I could replace some stuff (or even a lot of stuff) and cut down on some noise. And I don't much like noise gates because I use my guitar volume a lot and use the natural sustain of my guitar a lot, and gates tend to cut me off too early.

Thoughts?
-Joe Hart

[Mark Hammer]

Here's the next pedal challenge for folks.  We have seen many contributions for noise-making pedals here.  Maybe now is the time for noise-reducing ones.

There are schematics posted around for noise gates, but these tend to be unsatisfying because they generate there own quirky artifacts like abruptly ending note decays and such.

The ideal solution, and something which rarely gets discussed here, is a dynamic noise filter.  This is essentially an auto-wah at its core, except that it  functions only in lowpass mode and filters out very high end rather then lower treble and mids.  The idea is that you don't want to kill the audio signal outright, just the annoying hiss part of it.  That still leaves the hum part, but there may be less objectionable ways of killing that too, without eroding the attack and decay of note envelopes.

The original version of this is the National Semiconductor LM1894 - http://www.national.com/pf/LM/LM1894.html -  though other devices like Rocktron's proprietary HUSH chip - http://pdf.searchdatasheets.com/pdf/832/832.pdf - accomplish the task for guitars more effectively and suitably.  Of course, there is no requirement to use those chips, but any design based upon easily obtainable chips that has a low parts count will be appreciated by many, and could become as common a part of their arsenal as a compressor or clean booster.

I might also point out that some sort of time delay can also allow gates to function more effectively.  For example, a brief fixed delay of a few milliseconds (MN3207 plus simple CMOS clock) can permit a device to sense the onset of the note/chord in a more articulate and reliable way and open the gate in time to avoid cutting off any part of it.

Finally, the SSM2166 used in the AMZ Q&D Compressor - http://www.technicalaudio.com/pdf/semi_IC/AD-SSM2166.pdf - also has a downward expander function which is sort of the complement to a limiter and is quite effective in eliminating noise.  A limiter functions by restricting how much output there is, relative to input, over a certain threshold.  Everything UP TO that level makes it to the output on a 1:1 basis, but things over a certain level may only be a smidgen louder despite large increases in input signal level.  In a downward expander, everything at the input makes it to the output in a 1:1 ratio, but everything BELOW that level is substantially quieter than its input level (i.e., small differences in input level make for large differences in output level)

Just some ideas.

[NaBo]

wow... that would be GREAT, Mark!  ... ... ... you had better get to work!   :wink:

[Paul Marossy]

I agree with RDV. The higher the gain, the higher the noise floor is, typically. Yet, there are some really high gain pedals that do not have a lot of noise/hiss. I would have to say that part of noise control lies in the design of the circuit as well as the shortcomings of the parts themselves.

[Mark Hammer]

And I'm going to maintain that at least *part* of them comes from dumping bandwidth that you don't need at the earliest possible opportunity.  Do you have a design with multiple cascaded gain stages?  Do you expect to generate some additional harmonic content along the way?  Well then why the heck hang onto that stuff above 5khz after the first gain stage, and amplify something that is going to be even harder to eliminate once gain is applied?  Why not just dump it up front?  When I made my adapted Anderton Tube Sound Fuzz, I made sure to get rid of high end in the op-amp front end by means of a slightly larger feedback cap, and slightly larger caps in subsequent invertor stages.  It's by no means the sort of high gain device that, say, a Davisson Blackfire is, but there is some pretty substantial output coming from it, maybe even more than my BMP.  Yet the hiss at the output is surprisingly low.  Why?  Because I didn't amplify it 200 times before trying to get rid of it.

Admittedly, I know bupkes about discrete devices or what sort of resistance value should go to Vref in op-amp stages, so I don't doubt that there are ways of designing discrete gain stages that create less noise.  But what I do know is that even in bonehead op-amp based things, it doesn't take a whole lot to make the hiss situation easier on ones self.

[vseriesamps]

Joe you might think about using a noise reducer - maybe Mark Hammer's so to be announced specialty circuit if we all luck out here - together with a selected culprit inside a loop. Some pedal where you're not doing as much tricky tapering out. Sometimes a delay pedal is good for fudging taper (though they can be noisy as well, haha).

Good luck everybody. K

[bwanasonic]

This may sound 'stupid' but here's the 'answer' I use.
 Bypass switch for noisy effects.

Another handy thing is a volume pedal at the end of your signal chain as a manual noise gate/panic switch.

Kerry M