Silencer - New Noise gate/Expander pedal

[Astronaurt]

Hey folks, here's a design I've been working on coming up with. What I've got here is pretty much a rough draft I drew up in MS paint pieced together from all my notebook skribblings! I haven't had a chance yet to test it, but I'll have a couple free days coming up that I'll have the quality time needed to really tweak out over a breadboard! I'm still not quite sure what op-amps I'd use, but I have some TL074s and 4558s lying around so I figured those...

Basically it goes something like this: Signal goes in the input buffer, and is sent off to both the Sidechain input and the Potential divider stage (which we'll get back to)... the Sidechain input will be a switched jack so when nothing is plugged in, the signal from the input buffer passes through to the precision rectifier stage. The rectified audio signal now has both halves going positive, but for the purposes of the proceding Schmitt Trigger stage they must be negative. So there's an inverting gain stage in between there. Now the Schmitt Trigger stage works because a Signal that goes positive over it's bias point will cause the output to swing to 9v, and this generates the control voltage for the "gate." if a Negative going signal coming from the inverting stage pushes the input voltage below the bias point, you get 0v. The bias point is controlled by the "Threshold" pot. The reason I opted for a Schmitt Trigger rather than taking any stock comparator circuit was to introduce some Hysteresis into the gating action. The sensitivity of the trigger to smaller or larger input voltages can be controlled with the "trigger sensitivity" trim pot. Out of the trigger it goes to the Attack and Release controls, which are Variable resistors in series and parallel with the 2 22uF caps. My hope is that these can be used either sparingly to keep the gating action smooth as opposed to choppy, or in exaggeration for some odd effects. From here it goes to the LED Driver stage, which drives the LED/LDR opto-isolator as well as an LED indicator for the face. The LDR forms a potential divider with the 100k Resistor above it, as well as with the 470k "Reduction" pot below it. When the LED turns on, the LDR drops to negligible resistance, thus the signal coming from the input buffer is here divided between the 100k Resistor and the value set by the Reduction pot. The "gate key" Momentary switch is to be a Footswitch on the face of the stompbox that would allow you to Gate the signal manually just by shorting the LDR and 1M resistor, with the Full Range switch choosing between whether the Gate Key switch follows the Reduction pot or simply mutes the signal altogether. From here the audio signal goes to the Output buffer for a little bit of make-up gain, through the Output Level trim to get it to the same level as bypassed, and sends your nice clean audio signal on it's way.  

Anyway, I wanted to post it up to get some kind of concensus from the forum... Does the design hold water? I'm by no means a pro at this sort of thing so I've been reading a bunch of stuff trying to refine my original idea, and I can only imagine there's something I've missed, or something that'll need more tweakin'! Is there anything that jumps out at someone? I figure if I can work some of the bugs out by way of Forum, then I won't have to make several trips for parts back and forth to the Local Electronics surplus Warehouse in seedy seedy Downtown. so thanks for the input fellows!

[PRR]

> Does the design hold water?

It may work.

Some of it may be soggy.

I think the Attack and Release pot designations are swapped?

> The rectified audio signal now has both halves going positive, but for the purposes of the proceding Schmitt Trigger stage they must be negative. So there's an inverting gain stage

Flip the diodes, the rectifier makes negatives.

Also: AC-coupling rectified wave-shapes works wacky. Cap-coupling eventually finds the "average" of the wave.

"Precision" rectifiers have limits. I suppose TL072 and (costly) 1N34 will rectify guitar precision-enough at raw guitar level. But you are really looking for the bottom edge of guitar, possibly under 1mV. That demands heavy gain in the opamp.

Noise gates should usually look for midband signal, discount low hums and high hiss.

Why "precision"? It's just a yes/no decision and user-adjustable by-ear.

Usually a noise-gate side-chain starts with large (>1000X) simple audio gain with filtering. Discount hum/hiss, bring mV up to Volts, a simple non-precision half-wave rectifier works fine.

> divided between the 100k Resistor and the value set by the Reduction pot

The minimum loss 100K:470K, not even 2db, hardly seems worthwhile. You usually (IMHO) want 10db-20db, meaning reduction resistor 50K-10K, which is pretty far down on a 470K pot.

Oh, wait. Your output buffer is an inverter with 47K input impedance. This really fouls your 100K-LDR-470K network. While such a thing can be figured, and values computed, you have added an overall inversion and some-one will whine about that. Non-inverting hi-Z input is just a couple more parts, a lot less computation, and keeps polarity nominal.

[Astronaurt]

hahaha, soggy is better than sunk I suppose! Good feedback though. I didn't realize about the simple diode flip in the rectifier! That means I can take out the stupid inverting stage altogether, raise the 2 10k's in the feedback loop of the first op-amp in the rectifier and get the gain from there. Also perhaps filter lows going in, and filter highs with a small Cap in the feedback loop of that op-amp? then go straight from the Rectifier to the Schmitt Trigger without the need for AC coupling. I'm all for simplifying where possible!

Also, the Attack and Release controls are right I'm fairly sure, although thinking about it I suppose you could label them either way. I think of a gate opening as the attack time, and the speed it clamps down as Release, i.e. releasing the signal. For the output buffer section, it looks like the suggestion of a non inverting stage makes a lot more sense than whatever the hell I was thinking!

But to address some of your other points tho PRR, I'd say much of it is very purposeful. To be very frank, I AM NOT A FAN of Noise Gates. My biggest problem with them in the past is the unavoidable choppiness and plain unusability you get from putting in a particularly noisy signal, which seems rather backwards because that's what the thing is supposed to be designed to help right?! Rather than sit on my hands and just accept it, I'd rather do something. My fundamental goal is to make more of the parameters user variable so that you get a smoother and more useful noise control solution. Most of the time you don't see these controls at all on commercial stomp boxes, but there are hardly any Studio grade Gate/Expanders I've seen without them. I don't want a strict no-nonsense ON/OFF, signal/noise, dichotomy because I feel that the sound you hear isn't really adhering to that dichotomy anyway. Hence the emphasis on the Attack/Release/Reduction controls which are there to blur the lines a bit more. I also foresee it having use outside of just noise control, as fiddling with wonky exaggerated attack and release times on a gate can always be fun.  I also wanted to add the "gate key" momentary switch, which serves as a manual gate control. You could use it as a panic button if feedback has ran away and won't let the gate close, or for a buckethead style stutter effect, or for whatever else really. I just think this pedal has the potential to be really fun, so I'm gonna revise my schematic and get around to testing it!

[liquids]

I just want to throw an idea out there that I've had (and been confused with noise gate designs in general) for a while now.

Your clean guitar input signal is the best 'detector' for when you'd want to mute or unmute a signal for noise gate purposes, not your post-processed signal.

You're post-processed signal (last before amp input, or end of effect loop, maybe before time-delay effects) is where you'd want the signal to mute.

IMO, a noise gate should have two inputs.  One for the clean dry input signal.  And one for the signal to be muted post distortion etc.

[Astronaurt]

Hmmm, that's actually a really good point. It'd necessitate a Splitter box right from the guitars output so that you can send one chain to effects with the noise gate at the end, and the other right to the Noise Gate's sidechain control input. Good ideas like that are why I wanted to put a "Sidechain Input" on a switched jack on this design. So thanks for it! Now I'm gonna have to make a splitter box... I've also got a revised schematic I'm working off of to test the circuit tonight. We'll see what comes of it!

[gritz]

Hello Astronaurt, I'm rather interested in this as I'm about to embark on a similar project. What optocoupler will you be using, or will you be rolling your own from an LED and LDR? The reason I'm asking is that your LDR forms the bottom half of the potential divider and so will need to have a pretty fast (a few milliseconds) release time to allow note attacks through and most are in the region of 100mS (give or take an order of magnitude!). I was planning to use a resistive optocoupler too, but in the top half of the divider - my reasoning being that LDR attack times are plenty faster than release times and the attack times can be tamed with a capacitor and resistor somewhere...

I have a couple of Silonex NSL-32SR3 optos that I am going to try. The specsheet gives attack / release times of 5 / 10 milliseconds, although the actual behaviour will depend on the amount of current it's fed and the value of the resistor that forms the other half of the divider, so I'm looking forward to some marathon breadboard sessions!

A good point was made about filtering the input to the detector, I'll be doing that too. I'll probably start with 6dB / oct bandpassing at mebbe 250Hz to 1kHz. My gate is going to be more of a "Dimebag" effect, but I'd like it to be subtle-ish as well, if possible.

I've not done any work on mine yet (apart from some "back of an envelope" scribbles), but I did have a couple of thoughts:

1) I too thought about putting my distortion pedal "in a loop". I may make a hybrid circuit - the level detector is driven by the clean guitar signal and level control is applied pre and post distortion. That development is probably a long way off yet though!

2) I may try feeding back a small portion of the gate output back into the detector to give some hysteresis, i.e. gate starts opening at say -30dB and starts closing at -36dB. Might be a bit too abrupt, but won't be too hard to try anyway.

I did run an analysis on the gate built into my PodXT modeller (which works ok apart from the attack being a bit slow) and found it wasn't a gate, but a 2:1 downward expander with a soft knee, so that's given me some pointers. It's attack time is fixed at 10ms (too long!) and release varies from 10 ms to about 300 ms. The fact that it's an expander explains the lack of choppiness at least.

Anyways, I'll look forward to seeing your progress!

[Astronaurt]

I actually made my own Opto-isolators out of an LED and an LDR out of this "Sack O' Photoresistors" I got from my friend who works at an Electronics surplus store. Unfortunately, I measured their Resistance reacting to light, and they are depressingly slow. Where do you get some of these Silonex units ehh?   

[gritz]

I actually made my own Opto-isolators out of an LED and an LDR out of this "Sack O' Photoresistors" I got from my friend who works at an Electronics surplus store. Unfortunately, I measured their Resistance reacting to light, and they are depressingly slow.

I found exactly the same when I rolled my own with off-the-shelf LDRs. I figured I could maybe use one in a really gentle compressor, but I was looking for something a lot quicker.

Where do you get some of these Silonex units ehh?   

I ordered mine from UK Farnell (I'm in Britain). They stocked three different speeds, so I ordered a couple of each as I wanted to experiment with compression and gating - they worked out at about $3 US each. The Silonex website lists worldwide representatives at http://www.silonex.com/reps.html and distributors at http://www.silonex.com/company/distributors.html. Elsewhere on the same site is quite a lot of application data, including some schematics. They also look like the kind of company that would try to help if you shot them an email.

Hope this helps. Merry Christmas bud!