IF you were going to try to reduce noise, which components would you look to?

[sjaltenb]

I was just glancing over my old boss CS-2 Compressor thats waiting to be wired into the board. it was never real noisy, but not silent. its got a load of similar resistor values, mostly 1m, 10k and 56k.

all are 5% carbon, the rest of the components are mostly ceramics and standard "greenie" films, lots of transistors, and a few electrolytics.

So I was thinking it would be easy, since the values are all so common to slap in all 1% metal film resistors... but would this do any good at all?

I'm just curious if there is any reason why this would reduce noise...or, in general, what components SHOULD one look to replace in this situation. Thought it might be a good topic...

Thanks

[BAARON]

Yes.  Metal film resistors produce significantly less noise than carbon film.  They are the quietest practical kind of resistor for pedals.  http://aikenamps.com/ResistorNoise.htm

Film capacitors will generally be quieter than ceramic caps.  Electrolytic caps Do introduce some distortion into the audio signal, but if you read this article at AMZ, I think you'll agree that they aren't worth replacing (unless they're really ancient, because then they start to deteriorate).  http://www.muzique.com/lab/distortion.htm

You might be able to find low-noise equivalents for the ICs and transistors in the pedal, but that takes a lot more investigation and research than straight up replacing the resistors and caps, and really, it might not be worth it... compressors are just inherently noisy!  They'll take any noise in your chain and make it worse.  Way worse.

[MikeH]

Metal film resistors and tantalum caps produce the least noise, however many say they sound 'brittle'.  Greenies are ok, but panasonic film caps might be a little quieter.

Looking up lower noise opamps is a good idea too.  A lot of big name pedals use noisier opamps because low noise types can be more expensive.

[svstee]

Correct me if I'm wrong, but isn't the benefit of metal film resistors over carbon strictly higher tolerance? The same for tantalum caps over electrolytics.

[BAARON]

Metal film resistors and tantalum caps produce the least noise, however many say they sound 'brittle'.  Greenies are ok, but panasonic film caps might be a little quieter.

Yeah, some people will say that about metal film resistors, and it's true to some extent, in certain situations...  For example, a carbon comp resistor Does add some 2nd order distortion to the audio signal (which makes it seem warmer/sweeter than a metal film resistor would) when the static voltage And the voltage swing is high enough, but the 9v environment we find in a pedal is nowhere near high enough to produce that nummy distortion... just add noise.  R.G. had something to say about this a few years back...
http://www.geofex.com/Article_Folders/carbon_comp/carboncomp.htm

RE greenies being okay, but Panasonic being better?  I'll agree there.  I use Panasonics in all my pedals.  Plus, the Panasonics are usually smaller than greenies, and small is good.

Correct me if I'm wrong, but isn't the benefit of metal film resistors over carbon strictly higher tolerance? The same for tantalum caps over electrolytics.

Sure, I'll correct you.     Regarding metal film, no, the benefit isn't entirely higher tolerance.  You can get 2% carbon film and 5% metal film resistors, as an example.  I have both sitting in front of me at my desk right now.

Metal film is quieter than carbon.  Try swapping the plate resistor on the first triode stage in a tube amp from carbon film to metal film and you'll see what I mean: the amp's hiss practically vanishes.  There's a reason Dumble style amps typically use Vishay-Dale metal film resistors for plate resistors: low noise... and revisiting the "metal film sounds brittle" theory, I don't know many people who would say that Dumble amps sound thin or brittle.

Tantalum higher tolerance?  Maybe sometimes, though again, you can get tantalum caps in ±5%, 10%, and 20%.  Their main advantage, AFAIK, is that they are tiny tiny tiny compared to electrolytic caps.  Their main disadvantage is that some people find they die easily.  People seem split on this issue.

[R.G.]

I'm just curious if there is any reason why this would reduce noise...or, in general, what components SHOULD one look to replace in this situation.

"Noise Reduction Techniques" by Ott.

Only resistors and active devices produce noise. Capacitors and inductors, to the extent they are ideal devices, do not create thermal noise at all. For instance, a polystyrene cap is practically noiseless. A leaky electro is noisy to the extent that the leakage looks like a resistor. Inductance does not create noise. The parasitic resistance of the inductor causes thermal noise, but this is usually quite small.

Stages which get a lot of amplification after then are noise critical, because of the gain amplifying any noise they make. So the first stage in most amplification chains determines the noise level. Correspondingly, work on low noise parts on the input stage first.

Read Geofex. See "When Good Opamps Go Bad" for a discussion of how transients can over time degrade input transistors and what to do about it.

Consider low noise bias techniques to avoid feeding noise into inputs.

OTAs are sometimes inherently noisy because they can only tolerate inputs of about 25mV with low distortion, so signal is divided down in front of them, and then amplified up, exposing any noise problems.

Film capacitors will generally be quieter than ceramic caps. 

...
You might be able to find low-noise equivalents for the ICs and transistors in the pedal, but that takes a lot more investigation and research than straight up replacing the resistors and caps, and really, it might not be worth it... compressors are just inherently noisy!  They'll take any noise in your chain and make it worse.  Way worse.
[/quote]
As I noted, capacitance does not cause thermal noise at all. Capacitance can have leaking, which makes noise, or microphonics, but the capacitance makes no noise at all. Compressors make noise worse and appear to be inherently noisy because they necessarly involve either (a) a gain stage which is then cut back, causing the gain to increase input noise or (b) an OTA. Careful design can make compressors quieter, but it is difficult.

[BAARON]

Capacitors do not introduce noise?  I'll keep that in mind.  Thank you for correcting me, R.G.!  Sometimes it's hard to sort out the audiophile myths from the real world.

[MikeH]

Aren't tantalums less leaky than typival electros?  Which would be why many people, myself included, think they are 'quieter', in that they don't produce as much noise due to leakage?

[petemoore]

  A stripped down compressor, tuned just right [to ~mimic CS comp characteristics], could offer a smaller platform to try to find and eliminate noise from. 
  I would start with a low noise opamp and pay particular attention around the input. Chances are just working with 5% parts and a good design is >90% of it, that's as far as a good hard compressing effect gets.
  I owned and used one, the hiss it made seemed to be part of the CS sound, as in only present with comp on = produced by CS circuit components, far be it from me to recommend anyone go in and try to dig that noise out of there.
  Still it was a cool compressor. All compressors set to 'hard'...rise the gain a lot when low/no input, I still use the 'fast foot stomp' method of noise reduction, using a Dynacomp with 'slow rise' mod. also not particularly noisy...lol...for a compressor.

[R.G.]

Capacitors do not introduce noise?  I'll keep that in mind.  Thank you for correcting me, R.G.!  Sometimes it's hard to sort out the audiophile myths from the real world.

That's one that doesn't make the radar much. Thermal noise only happens in resistors and active junctions. It's only the parasitics (lead resistance, leakage resistance, etc.) about caps that can make any noise. Audiophiles seem to enjoy not knowing the truth.

About compressors in general - a compressor is designed to have its highest gain when there is no other signal to mask it. This is practically a recipe for how to make something sound noisy. We did something in our latest compressors that seems to work. We put a very slight noise gate in so that when the signal drops below the gate threshold, it mutes the compressor output. This is a feedforward gate, which listens only to the input signal, so the input signal is always in control. It's a little tricky to get the threshold reliably set below the useful output of the compressor and before the noise gets obnoxious, but then there's some art to the process. 

[MikeH]

Huh- Now the Q&D has a noise suppression and a compression function.  Does anyone know if the noise suppression comes after the compressor?  And if so, is it for just this reason?

[sjaltenb]

THank you all so much for the replies as always. The nature of the compressor, and its place in the typical stompbox chain certainly makes for a noise-"producing" (amplifying) device.

This is great advice and information. As the CS2 and dynacomp are first in my chain, I think I will go ahead and replace all the resistors, and go from there. It will be an intersting project no matter the outcome and I will keep everyone informed in case of any interest.

I do know that my new dynacomp clone is pretty darn quiet, and it used all high grade, 1% metal films, etc etc (my old one did not and was NOT quiet!) So this is what sparked my initial curiosity. I would simply clone the pedal with new parts, leaving out the buffers and switching etc, but I think it will be impossible to find the IC...

The IC is Roland BA662A. THe rest of the componenents would probably be fairly easy to source. Any ideas for the IC?

**after a quick search i found that that ba662a is almost impossible to find and without substitute. I will stick with the original PCB and mod from there! thanks!

[Paul Marossy]

Audiophiles seem to enjoy not knowing the truth.

And being flat out ripped off. 

[MikeH]

The IC is Roland BA662A. THe rest of the componenents would probably be fairly easy to source. Any ideas for the IC?

**after a quick search i found that that ba662a is almost impossible to find and without substitute. I will stick with the original PCB and mod from there! thanks!

You could use the ross comp project at tonepad to build a dynacomp.  They are nearly identical, just a few part substitutions.  The project uses a 3080 I think, which are still available

[StephenGiles]

 

[Processaurus]

The IC is Roland BA662A. THe rest of the componenents would probably be fairly easy to source. Any ideas for the IC?

**after a quick search i found that that ba662a is almost impossible to find and without substitute. I will stick with the original PCB and mod from there! thanks!

You could use the ross comp project at tonepad to build a dynacomp.  They are nearly identical, just a few part substitutions.  The project uses a 3080 I think, which are still available

Roland had the BA662 chip custom made at probably great expense as an improvement over the lower quality 3080 (for filters and VCAs in their synths, more likely than their pedals); with it the CS-2 is, not meaning artistically neccessarily, but electronically, an improved version of the dynacomp...

[DDD]

As far as I remember Mr. Keen said some years ago: "A bad compressor is a good guitar compressor" since it gives more dynamics than hi-gain-very-precise-superoverengineered one. So "low-gain" compressor is better and less noisy. It's a good (optimum) recipe, isn't it?

[Mark Hammer]

1)  Wire.  Noise comes in many flavours and one flavour is EM/RFI.  Another flavour is leads wandering too close to where they shouldn't and picking up stray signals.  So, even if you stuff everything in a metal chassis that provides seemingly 100% shielding, it can help to have shielded leads from/to some places.  If they're really short, it may matter very little, but not all boxes are planned out that flawlessly, are they?

2) Match of op-amp to circumstance.  Some op-amps can have great noise specs in one circumstance, and surprisingly poor specs in another.  Get to know your op-amps and where they will shine best.

3) Some components pose greater risk of noise in one part fo the circuit than in another.  If I use noisier carbon film resistors at the input of a high gain circuit where their inherent nise may be amplified 500x or more, they will contribute more noise than if I stick them in the post-gain EQ circuit.

4) There is much to be said for optimizing S/N ratio.  This seems to be something that nobody even addresses anymore.  Those of us who grew up in the days of technology that was weak but as good as we could get (cheap tape recorders, vinyl, non-digital tuners, etc.) quickly learned all about feeding each subsequent stage with the highest amplitude signal possible, so as to optimize S/N ratio.  While I am not a huge fan of lots of gain in on-board preamps, a wee bit of gain (e.g., <3x) is very helpful for feeding the next thng in line with a decent signal to optimize S/N.

5)  As near as I can tell, the SSM2166 derives the envelope used for the downward expansion function from before the gain-reduction stage (i.e., it conveys the true dynamics of the input).  I don't think the downward expansion constitutes a separate "circuit", though. Rather, the same envelope is applied to the signal in a different way, depending on whether the signal is above some minimum (where gain-reduction occurs) or below some other minimum (where expansion takes place).  Downward expansion works VERY well when set right.  It can be near transparent if you can get both the decay-time, expansion amount, and threshold just right.

6) There is noise in pedals, and there is noise in compressors.  Compressors are generally the worst offenders of all pedals, simply because they treat no signal as if it is something which deserves a LOT of boost.  In those cases, you are essentially hearing the pure noise contribution of everythng preceding your compressor, as well as the front end of the compressor. This, again, presents a good example of why its not just the components, but where they are in the signal path, that makes the critical difference.

7) Said it before and I'll say it again:  Manufacturers design pedals without ANY foreknowledge of what you'll be feeding it.  Consequently, they often include more bandwidth at the top end than you might need.  In those cases, you can exert some additional degree of noise control by simply trimming top end with a suitably-valued, and suitably-placed, capacitor.

8 ) The problem with the 3080 has always been that it cannot handle signals of any reasonable amplitude without clipping. (http://www.oldcrows.net/~patchell/archives/ca3280_2.html).  As a result it contravenes EVERY principle of maintaining a good S/N ratio when people design with it.  You have to seriously attenuate any signals fed to it, and then boost the crap out of them afterwards.  I repeat, the 3080 itself is not noisy, but what you have to do with it makes all other normal sources of noise exaggerated in their impact.  Same is true of JFETs in phasers as well.  In a perfect world, the cumulative effect of all those phase shift stages and feedback (which only multiplies the cumulative effect) would be combatted by feding the phase-shift stages with a very hot signal, and then attenuating at the end.  But, since the JFETs will clip like crazy, we feed them timid signals and live with the consequences.  That's why LDR-based phasers are so wonderfully quiet - you can pummel them and they'll never complain.

[DDD]

And what about SMD resistors? What are they made of? Especially 100 kOhm - 2 MOhm ones, that are frequently used in the input stages?
Do they produce more noise than MF resistors with axial leads?   

[George Giblet]

About compressors in general - a compressor is designed to have its highest gain when there is no other signal to mask it. This is practically a recipe for how to make something sound noisy. We did something in our latest compressors that seems to work. We put a very slight noise gate in so that when the signal drops below the gate threshold, it mutes the compressor output. This is a feedforward gate, which listens only to the input signal, so the input signal is always in control. It's a little tricky to get the threshold reliably set below the useful output of the compressor and before the noise gets obnoxious, but then there's some art to the process.

That's really the best solution.

The OTA noise is generally the dominant noise source.  High quality OTA designs go through great trouble to keep the noise down.  While is some effects OTA noise isn't so evident, iin compressors it's a big deal - because of "highest gain when there is no other signal".

Regarding the CS-2 design, I'd drop the input 10k resistor R13 to something like 2k2.  Then perhaps for an almost trivial improvement: half R14, R15, R8 and double C6, C8.