Simple noise-reduction circuit

[Skreddy]

Wouldn't an expander circuit with a bit of a clean boost effectively drop the noise level way below that of the straight signal?  Nobody but metalheads likes gates, but every live performer wants less noise.  I am missing something obvious, or does there not seem to be very many noise-reduction circuits based on this idea on the market right now?

[Meanderthal]

 There's lotsa downward expander units out there now, but they live in racks... Not a bad idea to put one in a pedal!

[Mark Hammer]

The Q&D compressor, based on the SSM 2166 chip includes a downward expander.  Works great.  The chip itself is a little pricey as chips go, but the few extra dollars it costs saves you a mountain of headaches and yields a ridiculously simple circuit for what it does.

[Skreddy]

I'm thinking a really cool challenge would be to make a SSM 2166-based expander/boost built into the body of a plug (using an oversized speaker plug for maximum room).  Might be possible with surface-mount... Wow, but I guess a 9v wouldn't fit.  I really want this to be small and self contained and unobtrusive.  Most pros use a Nady wireless, and I really don't want to add an extra clip-on unit to the guitar rig.  I guess I could think about other power sources.  Mmmm...

[g3rmanium]

Wouldn't an expander circuit with a bit of a clean boost effectively drop the noise level way below that of the straight signal?

Uhm, why the boost? Anyway, there's a National chip I recently stumbled upon that does noise reduction. Let's see if I can find it...

[g3rmanium]

Anyway, there's a National chip I recently stumbled upon that does noise reduction. Let's see if I can find it...

LM1894 -- that didn't take very long

[StephenGiles]

"I really want this to be small and self contained and unobtrusive."

Why?

[Nasse]

http://freespace.virgin.net/ljmayes.mal/comp/philips.htm I googled this circuit, if anybody remembers Philips DNL system.


Don´t know if Graig Anderton Electronic Projects for Musicians book noise gate circuit is online somewhere, but it is quite simple. The book is worth buying.


IMHO most noise reduction systems sound bad. Software based sound-processing systems can do wonders sometimes.

[Mark Hammer]

The common factor in many level-based noise reduction systems is their design around anticipated signal levels.  Every such unit is predicated around the idea that noise doesn't matter quite so much when there is signal to attract your attention, but matters a lot more when the only thing audible is noise.  As such, the unit has to be able to tell the difference between when there is signal to distract the listener, and when there isn't.  It does this based on average level.  An exquisitely responsive envelope follower is very helpful in carrying out the noise-reduction activity (i.e., near-zero lag), as is an exquisitely responsive control element (i.e., something that tracks the envelope in perfect sync).

But few such devices are in a great position to detect the envelope and adjust treble or all content over a broad range of potential input signals.  The "best" ones are usually predicated on seeing line-level signals, and have the gain in the envelope follower and amount of attenuation adjusted accordingly.  The Philips circuit linked to by Nasse, while potentially terrific, anticipates an input hovering around 775mv P-to-P.  In a perfect world, I suppose, the user would have a "magic" switch that adjusted several parameters at once so that the noise reduction device could cope with anything from simple dynamic voice mic signals of, say 15-20mv, to things right up to the several-volt range of synth outputs.

The perennial problem with gates is that since they have to "evaluate" the signal level in order to make gating decisions, there is always the risk of being either too conservative or too liberal in that evaluation and letting the wrong things turn the gate on.  If too conservative, the gate will certainly turn off before the noise has a chance to re-enter the picture, but the onset of notes will often be chopped and corrupted in an unnnatural manner.  If too liberal, the user will not miss the onset of notes, but the tails will linger much longer than one would like.  In a sense, what you want is a gate that is just liberal enough at the onset, and just conservative enough at the tail.  My guess is that DSP-based noise gating is probably more adept at functioning in such a manner and behaving the way the user ideally wants.

Generally speaking, expansion is better than gating, and expansion coupled with dynamic filtering is better than mere expansion alone.  Where gating operates in all or none manner, expansion operates in a graded manner.  True, one can buy/build gates where the amount of attenuation is adjustable such that the non-note interval doesn't suddenly go dead, but rather gets quieter.  However, the amount of attenuation is applied in a global manner below a certain threshold (i.e., anything below this level gets cut this much), in contrast to downward expansion where the adjustment is proportional (i.e., below this point, all content is "multiplied" by this constant), hence less abrupt.  The SSM2166 chip, referred to earlier, allows the user to adjust the point at which the downward expansion sets in, however it does not readily permit adjustment of the amount/ratio of expansion.  And just so we're clear about this, "expansion" means that greater contrast is created, such that what is a lesser change/drop in volume at the input, is translated into a greater change/dropin volume at the output.  In upward expansion, the loud parts get much louder.  In downward expansion, the quiet parts get much quieter.

Dynamic filters target hiss specifically, where gates address both hum and hiss.  With dynamic filters, the assumption is that residual hiss becomes more noticeable at lower input signal levels, so some gentle lowpass filter is applied as the signal level declines.  By this method, if attenuation of signal is exclusive to the high end, then enough of the remainder of the signal is in place such that notes won't appear to die suddenly as they do with gates.  Of course, key to the palatability of such filtering is that it be perfectly flat up to the corner frequency and nonresonant.  It simply wouldn't do to have a slight resonant peak at the corner frequency emphasizing some of the signal content and moving up and down like an autowah.

When downward expansion and dynamic filtering are applied, the two can lean on each other somewhat.  If the overall signal is going to be strategically attenuated by the circuit at some point, then the burden doesn't rest on the filtering and it can be gentler in both the amount applied and the input signal levels it is applied to.  Conversely, if some of the objectionable hiss is already removed, then any downward expansion can also be gentler, since it won't take much attenuation for that pre-filtered signal to seem really really quiet and noise-free to the listener.

As you can imagine, it is so much easier to design a device to take on 60hz hum OR high-end hiss, than it is to design something to take on both.  Perhaps the ideal tool is something that splits the signal at input and applies one strategy to the low end (e.g., a deep notch filter and some gentle gating), and a second strategy (e.g., dynamic filtering and some gentle downward expansion) to the mids and highs, and recombines them to achieve a pristine output that retains all the natural qualities of the input....minus the crap.

The constant challenge is for those who wish to achieve a decent quality signal from a single-coil equipped guitar.  One strategy for achieving a better guitar output is to use an instrumentation amp configuration on-board, whereby the two leads from the pickup are treated the same way the two leads from a balanced voice mic are, with a separate ground.  The leads go to inverting and noninverting inputs of an op-amp, and any audio crap acquired by the pickup is cancelled out in the op-amp.  Of course, the reason why one doesn't see that more frequently is because you need a separate op-amp for each pickup to accomplish it, and need to have the balanced processing and noise-cancellation occur prior to pickup switching/combining and level-setting.  In other words, it more or less demands that the onboard wiring be fairly complex and inflexible in its signal routing.  It also can make it cumbersome (though far from impossible) for those wishing to change pickups.

Finally, as has been discussed many a time, the other challenge is that noise comes from so many sources along the way, including imperfect cables, noisy pedals, and such.  Any noise-reduction system geared towards (and situated in) the path between guitar and first pedal is going to ignore some fairly serious potential sources of additional noise.  Any system geared towards cleaning up all the accumulated noise up to the point where you plug into an amp is going to have to be so heavy-handed that the odds are good that it will leave its imprint on the signal, as opposed to being invisible.

Since hum is going to be a much greater risk coming from the guitar than from subsequent pedals, any gating (gentle, please) and 60hz filtering should ideally be applied between guitar and pedals.  And any hiss-related filtering might yield more benefit if added AFTER all pedals.

[Skreddy]

Awesome.  Thank you, Mark and G3rmanium.  Great work, getting my creative juices flowing, especially about the 60-cycle notch tidbit and all the cool suggestions and scenarios.

StephenGiles; the background of this whole direction is my recent visit backstage as the guest of Aerosmith's Brad Whitford's guitar tech.  I was having a pre-concert conversation with him in the catering/dining room and asked him point blank what problems do you have that are not already solved (in the area of guitar effects).  Pretty much the only dragon that hasn't been definitively slain is that of noise, especially with all the lighting and video screens on stage nowdays.  I noticed that Aerosmith uses Nady wireless systems.  The best place to tackle hum and hiss in this case (as Mark pointed out) is directly out of the guitar.  So if I ever want to make a solution in this case, it would have to be something that plugs right into the guitar and is carried around with the artist during the performance, along with the wireless transmitter.

[Meanderthal]

 Oh, like a compander in a wireless phone or a dbx encode/decode type system? That would require a double-ended system, and really, the new wireless stuff is pretty quiet and probably already includes a compander circuit.
Single ended noise reduction like a downward expander would be best put to use as Mark suggested, at the end of the signal chain. That way you'd whack the pickup noise, any radio hiss or whatever, the noise created by any effects, and the cumulative noise of all patch cords. If ya just run the wireless receiver into an amp, put the downward expander between them and you'll get the guitar, radio, and patch cord noise all at once...

[Skreddy]

Yeah; I'm certain wireless transceivers have companders.  I don't want compression, though; only expansion.  All I want to do is get rid of hum and hiss from the guitar itself.  I don't just want to lower the noise floor; I also want to raise the signal level a tiny bit so that everything that follows it will have a clearer signal and won't need to have a pristine noise floor itself.  I realize you can always shield your guitars and put in noiseless pickups, but that's not a good option for vintage equipment sometimes, and it would be much more efficient to have a small noise-killer that can plug into any guitar on the fly.  Dealing with hum and hiss at the pedalboard/rack/amp layer is basically a function of existing good practices and isn't as problematic as noise coming from the source instrument itself.

I really like Mark's idea of having the expander zap 60 cycle hum with a notch filter (but only when the signal falls below a certain level, and in a gradual way, not like a gate) plus work on the hiss too (again only when the signal is low).  The 60 cycle notch filter could also have a switch to select other international power-supply ac frequencies (assuming others exist and are common enough to bother).

A completely custom enclosure is likely needed to accommodate the 9v battery and keep everything tiny, light, unobtrusive, and easy to maintain.  I mean, you could cover it in duct tape, but that would make changing the battery less than fun. 

[Meanderthal]

 I wonder if you could do this with an external dummy coil for 'pseudo' humbucking? Seems like the orientation of the dummy coil would be a problem, and a walco/eh style plug in box won't work for a strat anyway... but if it worked it would at least be passive...if...

[Mark Hammer]

Done it, and it works...up to a point.  I stuck a dummy coil in the control cavity of a buddy's Tele, and the hum was reduced.  Imperfectly, as you'd expect, but noticeably and pleasantly lower than before.  Interested parties should probably check out the thread in OT/Lounge about the Suhr system using the tremolo backplate.

[bancika]

Anyway, there's a National chip I recently stumbled upon that does noise reduction. Let's see if I can find it...

LM1894 -- that didn't take very long

Anyone has some circuit with this chip which could be used with guitar, I tried circuit from data sheet but it only induced more noise.