Post
by Albert H » Wed Apr 19, 2017 1:09 am
The 2N5459 is used as a voltage-controlled resistor - as you vary the bias on the gate, the resistance from source to drain varies. A tiny amount of audio is also applied to the gate - this is a distortion reduction trick that was discovered way back in the 70s. Contrary to what Mr Krakatoa asserts, FET attenuators are the preferred audio control method in virtually all commercial compressors and limiters. They have very low noise and distortion, so are widely used in studio audio processors.
Another option is to use an optical control circuit - there are some very well-regarded audio processors that use this approach. If you use an LED to illuminate a Light Dependent Resistor, you can make a low noise gain controller or attenuator. However, the LDRs react quite slowly, so a limiter of this sort will let through transients, and will recover quite slowly. These are great for use in a studio (my preferred guitar compressor uses this technology because it retains the attack of each guitar note), but are not really much use for a limiter. I have used them in a slow-acting AGC in radio broadcast chains that have several sources that can have wildly differing levels.
The next kind of audio control is the OTA - the Operational Transconductance Amplifier. This is a modified type of op-amp, which has a current input pin that allows gain control. There's only really one or two of these devices on the market these days: the LM13600 and 13700. They're practically identical for our purposes. They can be used in series with the audio, which is OK, but they tend to add a bit of noise and distortion, or a better approach is to use them to control the negative feedback around an op-amp. These devices are OK, and for pirate use they're probably good enough. There are two OTAs in each IC, and they're very well matched, so stereo limiters are easily realised.
The final type of voltage-controlled attenuator is the specialist chip, like the "THAT" series of high quality audio processing components. I've used the 2018 in my commercial limiters with great success, but the ICs are over £10 each. They are also susceptible to supply noise and they really don't like RF about! These ICs are capable of really superb audio performance, but the sidechain often limits their speed of response, necessitating a clipper after the limiter, which rather defeats the point of a low distortion gain control stage!
My latest commercial audio processor uses the MN3007 clocked so as to eliminate any possible beat products with the 19kHz and introduces about 1.5ms of delay into the audio path. This gives the sidechain time to operate, so this limiter has (effectively) zero attack time and also no overshoot. There is no need for a clipper, and the use of high quality audio attenuator ICs, guarantees fidelity. It allows you to be the loudest thing on the band, without nasty clipping or pumping. It's NOT cheap or simple, but if you want the best, it's the only game in town. It makes BW's processor sound nastily distorted, and it outperforms Orban's offerings too! We're shipping lots of them to the 'States and South America, and many station engineers describe our processor / coder combination as their "secret weapon" !
"Why is my rig humming?"
"Because it doesn't know the words!" 
That looks like a nice limiter you advertised and im sure it works very well!