Noise Reduction DIY >not gated< DIY?

[orson]

Hello everyone, i've been searching for ages for a noise reduction project. I can't seem to find any. I don't mean a noise gate, i have one of those. I'm looking for a project like an old DBX or dolby noise reduction schematic to get rid of hissing and other annoyances in realtime. The simpler the better, as i have never had anything to do with noise reduction before.

Thanks In Advance

[David Schwab]

Neither DBX or Dolby work in real time. DBX uses a "compansion" (compression/expansion) scheme, where it compresses the signal when recording, and then using downward expansion when playing back. This pushes the noise floor down below the audible level.

Dolby works using emphasis/de-emphasis, whereas it boosts the highs when recording (emphasize), and cuts the highs by the same amount during playback. That's probably over simplifying what Dolby does, as there are a few types of Dolby noise reduction systems, and they probably combine that with some type of compansion. This is why a cassette tape which was not recorded with Dolby, will sound muffled when played back with Dolby on.

The only type of noise reduction that works in real time is some type of noise gate, which might also combine low pass filtering with gating.

[orson]

Oh well, i guess i'm going to have to find some other way of doing it.

[orson]

DBX has a new graphic EQ range out with type iii noise reduction

http://www.dolphinmusic.co.uk/page/s...roduct_id/6859

[Mark Hammer]

I'll offer a different opinion. They all work in "real time". The question is whether they are single or double-ended. Double-ended requires that some sort of encoding be adopted, which is then "de-coded" after the stage or stages that presumably introduce noise or distortion or both.

In the case of tape and vinyl albums and film, yes, the material is encoded in some manner which is then decoded on playback, and the decoding results in useful noise-reduction. But such double-ended noise reduction is regularly used with lots of things. For instance, many delay lines, and even chorus, phaser, and flanger pedals use companding (which is a cousin to dbx) to reduce noise. The delay chip and delay process introduces risk of clipping and several noise sources, so the input signal is compressed, passes through the delay circuitry, and then gets expanded again, with protection against unwanted clipping and the acquired noise drastically reduced. That all happens in "real-time".

Alternatively, there are single-ended solutions where a circuit can reduce noise even though the program material (signal) has not been pretreated in any way. Probably the most common these days is the H.U.S.H. noise filter system marketed and used by Rocktron. This uses a combination of "downward expansion" and adaptive filtering. The idea is that if you exaggerate small differences in level below some threshold, and then filter than content as well, you can avoid the irritation of hard gating, plus you can take out the hiss when it matters most, and adopt a more relaxed attitude when you won't notice the hiss quite so much.

So, at and below a certain signal level, a 1db change in volume is translated into, say, a 3db difference in what comes out, without any change in the signal above that level. The sound is never completely chopped, so you get all the initial attack and all the decay/sustain , but if it gets quiet, it gets REAL quiet. At the same time, content juuuuuust under that threshold may have just a wee bit of treble shaved off. Since there is so much signal anyways, you won't hear the hiss much. As the signal level declines, though, more and more of the treble is shaved off the top, as the puny signal starts to compete with the noise/hiss superimposed on top.

This type of approach won't salvage the worst material, but it can do a very nice job on material which is only modestly challenged. The SSM2166 mic compressor/processor chip has such downward expansion (without the adaptive filtering) built in, and does a terrific job turning what is all too often a noisy effect (compressor) into something that is pleasingly quiet, all without seriously hampering the onset and tail of sounds.

You can buy the SSM2166 from Small Bear Electronics (smallbearelec.com) and you can find the datasheet and a sample circuit online (http://www.analog.com/UploadedFiles/...SM2166_b_.pdf). Though it includes a compressor, you don't HAVE to use it.

The SSM2000 HUSH chip can also be gotten, but not quite as cheaply and the resulting circuit is vastly more complicated than what you'd make with the SSM2166. Well worth checking out.

[orson]

Yeah thanks for that, I came accross Rocktron products earlier, i didn't even nkow they existed. The reason i'm doing this is to give me somthing to do as i am off work sick atb the moment & reduce the hissing made by my mesa amp's preamp section. It has an effect send/return after the preamp and i placed a graphic EQ in the chain and it made a huge difference, my amp sounded like crap but at least i knew where the problem was coming from. I contacted mesa and they basically told me that it was a noisey amp at lower levels, and i've now had the chance to talk to other owners and they basically say the same thing. I have swapped around valves etc and found that an electroharmonics valve at the first preamp stage with the rest being mesa tubes, for some reason increases the volume but reduces the hiss slightly. But the amp is still useless for recording at lower levels (which i need intolerant neighbours ).

[orson]

This has caught my eye, in fact it's caught both;

http://www.muzique.com/ssm2166.gif

[Mark Hammer]

It's a pretty damn simple circuit for what it does, isn't it? I have one of these in my pedal arsenal and it works like a charm.

The only caveat I would offer is that the chip itself is designed for mic processiing, so the input impedance of the chip itself is around 175k, apparently. The circuit shown does nothing to offset that. If the unit is insrted into the signal path AFTER some other pedal which has a low-impedance output, then the low input impedance of the SSM2166 does not present a problem. Most players, however, myself included, prefer to stick a compressor right at the head of the signal chain after the guitar. In which case it is advisable to have some sort of unity-gain buffer stage added to the input of the circuit shown, just to make sure the guitar isn't loaded down and lose its sparkle.

[David Schwab]

A simple buffer amp would fix that if you wanted to have it right after the guitar.