Super lowe parts count gate? (is this the right term?)

[midwayfair]

I'm looking for a minimal parts count way to make all signal larger than x bypass an amplification stage. Signal x or smaller would continue on its merry way.

I tried searching but I'm getting a lot of stuff that isn't what I'm looking for, like logic gates ...

[Jopn]

Sounds like you're either looking for a compressor, or the exact opposite of a noise gate.

[midwayfair]

Sounds like you're either looking for a compressor, or the exact opposite of a noise gate.

I might want to use it in a compressor, but that's not what I'm looking for. A compressor would take the entire signal and reduce the output. I want to do something with ONLY the part of the signal that exceeds a certain size.

A noise gate could theoretically work, but way too many parts for I want this for.

[samhay]

I guess you could rectify the signal then pass this to a comparator. The comaparator could switch e.g. a FET or 2 on or off, which could pass your signal or not. Would have to spend some time fine tuning, but seems reasonably achievable. If nobody comes up with a better idea, I'll try and flesh this idea out into something workable.

[midwayfair]

I guess you could rectify the signal then pass this to a comparator. The comaparator could switch e.g. a FET or 2 on or off, which could pass your signal or not. Would have to spend some time fine tuning, but seems reasonably achievable. If nobody comes up with a better idea, I'll try and flesh this idea out into something workable.

Hm, not a bad idea. I could maybe set up the FET as a voltage divider, too, so it just pans a certain amount according to the signal strength instead of just switching.

Would it also work to just split off the signal through a pair of diodes? Like clipping but it sends the overage signal to, say, a different transistor? (And would it still clip the "dry" signal?

[ashcat_lt]

Pretty sure you could split the signal two ways, with one going through back-to-back series diodes (like you suggested).  Both signals will be severely distorted when the diodes are conducting because it's dependent on instantaneous voltage.  If, for example, you run a +-1V sine wave through this (using si diodes) the "dry" signal will follow up to .6V and then stay there until the wave comes back down, the follow down to -0.6V...  The other side of the split will sit at 0 and then suddenly jump to 0.6V and then follow the sign till it come back down, then drop to 0 and sit there till the other diode opens.

An actual noise gate usually doesn't open and close that fast.  It averages the input over some time period.  While the average sits above the threshold, the gate is open and the output follows the entire wave cycle until the time average drops.

Which are you looking for?

[GGBB]

I guess you could rectify the signal then pass this to a comparator. The comaparator could switch e.g. a FET or 2 on or off, which could pass your signal or not. Would have to spend some time fine tuning, but seems reasonably achievable. If nobody comes up with a better idea, I'll try and flesh this idea out into something workable.

Isn't that essentially this: http://www.geofex.com/FX_images/vibmatic.pdf?  Jon - you used this in the Blue Warbler - is that not what you are after?

What you describe sounds a lot like upward compression.

[Jdansti]

>signal larger than x bypass an amplification stage. Signal x or smaller would continue on its merry way.

It sounds like the difference between this and a compressor is that a compressor increases the amplitude of small signals and decreases the amplitude of large signals to even everything out.  This would behave the same way for the small signals, but the large signals would not be limited. Correct?

Edited for typos.

[midwayfair]

Which are you looking for?

Not sure yet. I couldn't think a circuit that did what I described with a clean signal and was going to worry about how to use it later ...

[ashcat_lt]

Wait, what? 

Are you willing to accept massive distortion in this process or not?

[R.G.]

I think he wants downward expansion with an adjustable threshold.

A simple noise gate can be done as well. Noise gates always involve detecting when there is a signal bigger or smaller than X and then doing something. The simpler they are, the harder they are to adjust and get orthogonality in controls. And you rapidly run into the issues with low parts count setups not working very well.

[gritz]

I think he wants downward expansion with an adjustable threshold.

A simple noise gate can be done as well. Noise gates always involve detecting when there is a signal bigger or smaller than X and then doing something. The simpler they are, the harder they are to adjust and get orthogonality in controls. And you rapidly run into the issues with low parts count setups not working very well.

Yeah, this. As well as the normal 80/20 "rule" which suggests that 80% of the desired functionality will be attained with the first 20% of effort there's a flipside that generally dictates that below a certain amount of effort nothing worth a sniff will be achieved at all. This is especially true of anything approaching  transparent noise reduction.

A "gate" - as in a simple on / off switch is a compelling idea, but it's very complicated to make an effective one that works well with every signal that the user is likely to throw at it without making that user resort to twiddling controls every time he or she switches pickups, swaps from plectrum to fingerstyle, or engages a different fx pedal. You need to tell the gate's detection circuitry (the sidechain) how to differentiate between "good" and "bad" signal with filtering and it needs to respond at an appropriate speed so that good signal isn't lost and noise isn't let in. But it's far more complicated than that:

You have a gate and you twiddle the "threshold" knob to the appropriate level.It's important, because too low and your decaying notes are drowned in noise before the gate kicks in, too high and your notes are cut off prematurely. However, the output of a guitar pickup in response to a decaying note is a little unpredictable. A pckup is good at detecting string movement to/away from itself, but not so good at detecting when a string is oscillating in parallel with the pickup's face, so the signal strength may not decay smoothly with the note, but may fade in and out many times a second. The way that our ears work means that this few dB of variance is not terribly noticeable to us, but our gate will know - and this may cause the gate to chop in and out as the note level hovers around the gate threshold. Smarter gates will use hysteresis- the gate might open at say -40dB and shut off at -46dB so that signals that float around in the deadspace between those two levels don't alter it's state. But how ever non-low the parts count of your gate gets it's still going to struggle, because gates are really only suited to percussive sounds, like drums and palm muted guitar. Any sound that decays  slowly, or meanders, or is at all nuanced is a struggle. That's why better noise reduction systems use downward expansion (turning the volume progressively down depending on how far the signal is below the threshold), or filtering with downward expansion. The ISP Decimator uses " infinite expansion with knee" - as the signal drops below threshold the volume is wound down until at thresh -12dB (one quarter of threshold level) it's effectively silenced. But the Decimator uses a very clever (and high parts count) level detection regime which means that the volume reduction is quick but smooth. Try that with an ordinary envelope detector based on one diode and a capacitor and it'll either distort your decaying signal like mad or respond far too slowly.

There was a similar thread a few months ago and someone mentioned centre clipping (aka coring). This uses a pair of back to back diodes to block low level signals, but let higher amplitude signals through. It can be very effective if your signal is already heavily clipped and distorted, but it will sound like pish with a clean signal, because it's basically crossover distortion. Check out the Boss HM2 schematic and you'll see two germanium diodes (d6, d7). It can be a bit brutal - although a resistor in series with the diodes does make the effect a bit more progressive, at the cost of reinstating some noise.

It's complicated. Best thing is to eradicate noise at source as far as possible. Really.

[WaveshapeIllusions]

What I got from the OP was a gate that routed low-level signals one way, and high-level signals another. Basically amplitude based bypassing. The noise gate comments still apply though, as it is still a similar action.

I think using a schmitt trigger with a bit of hysteresis might be good for driving the switching circuit. One just has to properly get the signal amplitude into the right form to trigger it properly. Perhaps look into some of those PPM detectors?