Will this noise gate do?

[stonerbox]

Morning!

I am looking for a noise gate with a microscopic footprint, to add to a big project I am working on. I plan to place a clamper fairly early in the chain to lightly dampen pickup noise. It does not have to be exceptional but it needs to be transparent.
How would this very simple gate fair? One thing I am thinking is that it could cause impedance problems with the 'gate' set to low resistance.

[blackieNYC]

https://www.diystompboxes.com/smfforum/index.php?topic=119340.20
See Kipper's entry 32. One little buffer.

[stonerbox]

Thank you blackieNYC for the link!

I already have a mega high (literally) impedance stage in place so I will just use that. Next thing to tackle, for me at least, is to try to remove the swelling in effect of that gate (as heared in Richie's video) and fine tune the decay some. Also be omitting Q4 and all components relating to that LED indicator. It is just an indicator, right?

I wonder if the results without that opamp will do or if it gets much too finiky to adjust threshold with only a transistor (less components)?

[Mark Hammer]

The gate initially shown has fixed time constants.  You probably want to think about what sort of onset/offset of gating provides the least intrusion/disruption to what you want it to be married with, so that the end result is improved.

You might also want to consider making D1 a Schottky diode.  Its role is to shunt off a half-cycle of whatever the input signal is, reducing ripple.  A lower Vf diode, like a Schottky, will shunt off more of the unwanted half-cycle, leaving a lower-amplitude ripple to contend with.

[amptramp]

If you want a microscopic footprint and that is your main criterion, the design in the first post may fit but I doubt you would be happy with the performance.  You have no control over the attack and decay times and the presence of R2 means you do not have a cutoff, you just have a change in gain from 1.0 going through to 1/11 or 9,09 % with the BC547 turned on.  If you are driving this circuit from a low-impedance source, the low input impedance with the gate control set to low resistance may not be that much of a problem but a lot of pedals have a 1K series output resistor just to handle short circuits at the output and driving from one of those pedals may cause a problem at a low-resistance setting.

The AQR design in the previous post overcomes a lot of the problems but one idea you might want to look at is a photocell in series with the signal lit up by an LED that is driven by the average signal.  There would also be a capacitor to ground at its output side.  When the signal goes down, the resistance of the photocell increases and the RC time constant gets longer, reducing the bandwidth of the signal and reducing the treble so the cutoff sounds seamless.  Another idea along the same line is the Magic Monitor circuit where the diode rectifier drives a reactance stage that connects to the signal after the series resistor.  The reactance stage increases in capacitance to ground when the rectifier bias goes down, limiting the treble.  RCA used this circuit in the 1940's as a means of reducing the hiss from records when the input was low, like between record cuts.  It was a little bulkier:



but you could easily do a transistorized version with a JFET reactance stage.  It might be what you want rather than a traditional noise gate.

[FSFX]

I am looking for a noise gate with a microscopic footprint, to add to a big project I am working on.

You won't get much more microscopic than this SSM2167 or the SSM2166 chip which you can use as a compressor and noise gate.
https://www.analog.com/media/en/technical-documentation/data-sheets/SSM2167.pdf

[blackieNYC]

I have to say I hated the 2166 as a compressor.  Maybe as a gate only it would be good. 
The compression was great, but there is so much noise that they had to include a noise gate.  You hear a little burst of it at the end of each note. 

[Kipper4]

Looks like I  made a demo too. Who knew.....lol

https://youtu.be/4X7san9gGA4

Have fun Stonerbox.

[stonerbox]

I am overwhelmed and humbled by the feedback, thank you all! So much to dig into and I am especially interested in diving into the Magic Monitor someday.


Here is what I got so far. It works wonders in dampening the noise (without completely cutting out the signal) but how could I make adjust the decay time? It closes in about 800ms-1s now and I would like to add maybe 500ms more.

[Kipper4]

What happens if you make c33 bigger?

[Mark Hammer]

Take a peek at just about any effect that uses a sidechain, like you use, and decay time will be set by providing a resistive path to ground in parallel with C33, though usually with a larger value of C33, like perhaps 1-10uf.  The Decay control woud be a large-enough range of resistance values to bleed off the voltage stored on C33 quickly or more slowly.  A suitable range might be a 22k resistor in series wit a 500k-1M variable resistance.

Since Q18 is being used as a voltage-controlled resistance to ground, several options are possible for consideration.  One is the addition of a (modest) resistance between C39 and the collector of Q18, to reduce the amount of attenuation produced, if it acts too much like a gate.  The other is the addition of a cap between C39 and Q18's collector to provide a gated/triggered treble cut, rather than full bandwidth attenuation.  Neither may be necessary, but both are worth experimenting with for sonic options.

[stonerbox]

Klipper and Mark, great suggestions!


I am blown away how good this latest version works with the ridicules hyper-like-fuzz I've built. The gate makes it play and operate perfect, even better than before. I can now get full blown acid dripping fuzz or smooth cleans simply by picking soft or hard on the guitar. If I roll back the guitar volume just a little the guitar gets super clean.

Now riddle me this!

Why this configuration helps to not cut the signal fully is somewhat clear to me (hFE+R73?) but why does it make the gate super smooth when going from dampened to fully open? It reacts near perfect to different picking strengths, it almost feels natural in the way it operates now.
Piggybacking transistors lowers hFE but if I substitute the 2n9304 for a low hFE 2N2369 it silences the signal (when gating) and also nukes the smooth touch sensitive function of the gate. It would be economical (minuscule, but still) to be able to have one NPN there instead of three.
I also noticed that in removing R73 effectively raises the sensitivity of the gate opening.

Second question is to Mark, when inserting a 1-2Meg resistor/pot from C33 to ground the decay does not get extended. If I place it in between D2 (at C33) and Base of Q17 I can extend it but it still eventually closes too quick. In order to get a smoother decay I need a minimum of 47uF or preferably a 100uF. Despite the high uF values the cutoff comes off as an anti log curve. In a perfect world I would have a log instead but I guess that is asking a lot from such simple circuit.

Any Ideas?

Edit: I realized that lowering (or even omitting) R73 lets me use one 2n3904 or even a 2n5088. It is however not as natural (in response to transients) as with the triple setup. Edit 2: If I scale down the size of C26 I can get a bigger range in picking clean before the gate opens up and the fuzz starts spitting acid again. Neat!

[stonerbox]

Terribly sorry about the bump but clarification on this would really help me out and maybe others too (check the updated text down below). Also another thing that is not clear to me, what caps determine attack and decay? As far as my experiments show both C33 and C38 effects both the decay and swell/attack of the gate, correct?

Lastest version

Now riddle me this!

Why this configuration helps to not cut the signal fully is somewhat clear to me (hFE+R73?) but why does it make the gate super smooth when going from dampened to fully open? It reacts near perfect to different picking strengths, it almost feels natural in the way it operates now.
Piggybacking transistors lowers hFE but if I substitute the 2n9304 for a low hFE 2N2369 it silences the signal (when gating) and also nukes the smooth touch sensitive function of the gate. It would be economical (minuscule, but still) to be able to have one NPN there instead of three.
I also noticed that in removing altering R73 effectively raises changes the sensitivity (opening) of the gate.

In order to get a smoother decay I need a minimum of 47uF or preferably a 100uF. EDIT: Putting resistance in between Caps and Bases solved the issue with decay, of course. Letting the voltage drop slowly instead of being dumped through the transistors. Despite the high uF values the cutoff comes off as an anti log curve. In a perfect world I would have a log instead but I guess that is asking a lot from such simple circuit.

Any Ideas?

Edit: I realized that lowering (or even omitting) R73 lets me use one 2n3904 or even a 2n5088. It is however not as natural (in response to transients) as with the triple setup, it has much more range to play around with before it blows open.. If I replace R73 with a 50k pot it effects the sensitivity a great deal when increased.

[Mark Hammer]

D11and D2 turn the audio signal into a positive-going DC voltage.  That voltage/current is stored and "averaged" by C33.  Resistance between D2 and C33 determines how quickly C33 charges up, and that cap's value determines how long it would take for it to be fully charged up to capacity.  Small C values mean that the time between being fully charged and fully discharged, in response to any incoming DC, is shorter/faster.  But discharge time can be expedited by providing a path for current to drain off, via a resistance in parallel with C33.

[stonerbox]

D11and D2 turn the audio signal into a positive-going DC voltage.  That voltage/current is stored and "averaged" by C33.  Resistance between D2 and C33 determines how quickly C33 charges up, and that cap's value determines how long it would take for it to be fully charged up to capacity.  Small C values mean that the time between being fully charged and fully discharged, in response to any incoming DC, is shorter/faster.  But discharge time can be expedited by providing a path for current to drain off, via a resistance in parallel with C33.

Excellent, Thank you!

Now for my last issue. I need to boost the signal a little after the gate but if I hook up a NPN boost it does not increase the volume noticably. How come? Something to do with impedance I would assume? The follow up stage (post gate/boost) has a 3nF in parallel with a 500k pot into a 1Uf, the base bias of 1M from 4.5v and a 150k to ground. Collector 68k and emitter 100k pot, unorthodox stage, I know, but it has to be that way for following stages to work properly.
If I put a buffer after it works but instead of a boost I generate -200mdB loss in volume. In my hunt for minimal footprint I would rather not follow up the gate with both a buffer and a boost or the other way around but perhaps there is no other way?

[stonerbox]

Anybody?



Edit: Think I found the answer. My stage after the boost is around 3-2Ω (2n5088. Base 4.5v->1Meg+150k ⏚. Collector 68k. Emitter 100k Pot) and my boost stage have an output of 1-1.5M. Should I even bother to redesign the output impedance to 1/10 of 2Ω?

.

Way back during my technician-in-training days (1967-1970), one of the subjects was transmission line theory, where we were taught that load impedance should be equal to source impedance.  In the years since, and in further recent studies, this rule has held up well.  As I see it, a pickup has very non-linear frequency response, and a load mismatch helps compensate for this response.  Wouldn't it be technically more correct to design for a matching impedance and use filtering to shape the response?  One of the potential issues I can see with high load (pedal input) impedance is that the effect of cable capacitance between the guitar and pedal will be more significant than it would with a load impedance that matches pickup impedance.

Matched source and load impedances are the way to get maximum power transfer. For RF transmission lines, that is the ideal case, as matched impedances also keep the transmission line from having standing waves and reflections.

In general, that's not what we want for audio signals. We want maximum voltage transfer, not power transfer. For maximum voltage transfer, the load impedance should be at least ten times the source impedance. If we used current, not voltage amplifiers, we would want input impedances less than one-tenths of the source impedance to suck in every last bit of current the source could put out. But for the voltage amplifiers we use, the input impedance acts like a voltage divider. A matched input impedance immediately loses half the input signal to loading just by ohm's law.

Edit 2:

Got it working by moving the boost from the back to upfront. However the bass gets squashed now similar, but not as violent, as to how a Fuzz Face reacts to a buffer. Overloading input of Q1? Will I ever get this working properly...
[click to enlarge]

[antonis]

My stage after the boost is around 3-2Ω (BC549c. Base 4.5v->1Meg+150k ⏚. Collector 68k. Emitter 100k Pot) and my boost stage have an output of 1-1.5M. Should I even bother to redesign the output impedance to 1/10 of 2Ω?

Could you plz post a schematic of what you're saying..??

[stonerbox]

My stage after the boost is around 3-2Ω (BC549c. Base 4.5v->1Meg+150k ⏚. Collector 68k. Emitter 100k Pot) and my boost stage have an output of 1-1.5M. Should I even bother to redesign the output impedance to 1/10 of 2Ω?

Could you plz post a schematic of what you're saying..??

Just did while you answered! Up above. 
Now, I know the setup of Q1 is quite unorthodox but I need to squeeze every possible dB out of that 2N5088, around +35dB.

[stonerbox]

Swapped R65 for a 8.2k just to give it a little push.

Done some A/B testing with both guitars and drum machine/synths. The gate circuit muffles the attack too much. Tried to tackle on a buffer after but that did  not improve response. I guess Q1 (far right one in schem.
) is extremely picky with buffers and boosters. It prefers pickups and line levels.

[stonerbox]

Replacing the input boost stage's input cap from 1uF to a 10n did the trick in restoring the snappy attack of all signals (guitars/synths). Turns out the boosted signal and foremost its bass smashed Q1 and destroyed/overtook the transients of the midrange.
I did not know how but I feared the gate itself distorted the transient response somehow but sometimes the answer to a seemingly "advanced problem" is as easy as an input cap!

Also added a unity gain stage after the gate which helped to preserve tone and dynamics.
A boost a gate and unity gain stage and it's done, so much for microscopic footprint!