Striker plexi drive schematic

[Digital Larry]

i was talking about throwing away frequencies in general. it would seem to my uneducated posterior, in theory and in limited practice, that once you attenuate frequencies to ground potential, they, and the area they once occupied would be gone, even from a sine wave.  if you're merely "turning them down" with an adjustable device on a recorded (not live) media, yah, turn it back up and no harm no foul.

If you have a series capacitor going into a shunt resistor, that forms a "high-pass" filter.  It's said that it "rolls off" frequencies below the corner frequency, e.g. 6dB/octave.  But that doesn't mean it makes them disappear completely.  If you have a resonant notch filter, like shall we say in a Varitone, you could achieve infinite attenuation if all of the components had zero resistance.  In the real world this does not happen and so the notch might be deep, but not truly infinite.  Also that deepest part of the notch is at just one frequency and anything to either side is not attenuated as much.  But if you notch something out and then boost it to try to recover, you will be boosting a bunch of noise, as someone already mentioned.  And while you can conceivably recover the amplitude of a signal via a boost following a cut, each step will introduce phase shift (frequency dependent time delay) that cannot be cancelled out.  This is the "no free lunch" clause in action.  Fortunately our ears are not very sensitive to phase shift, unless for example you mix the shifted signal back in with the dry signal which causes cancellations in amplitude.

About 10 years ago I studied the details of MP2 and MP3 compression as I was writing a program to transfer WAV backing tracks into a Korg PXR4 digital recorder.  So I know a little about it but not a lot.  

First your audio gets digitized and sliced into frames which are 1/75th of a second long.  Then it goes through a frequency analysis stage called an FFT, so you have a bunch of numbers representing the spectrum of that 1/75th of a second of audio.  Then the "perceptual encoder" goes through and says "hey there's a signal at -25 dB at 440 Hz, but it's right next to a signal at -3 dB at 469 Hz, therefore nobody's ever going to hear the 440, so let's throw it away".  That's right, 440 Hz doesn't get encoded at all for that 1/75th of a second.  This is in essence a computer program making decisions about what you can or cannot hear and completely throwing away the bits it doesn't think you'll be able to notice missing.  Or, care enough to complain to iTunes that "Bon Jovi and Barbra Streisand cover the songs of Slayer" has something messed up and you want your money back.

On playback the audio signal is reconstructed frame by frame using frequency and phase information that was stored in the compressed file.

Later on, if you decide "ah, I never really liked 469 Hz anyway" and take pains to remove it, your 440 Hz isn't going to rise out of the muck because it simply isn't there (at least for the 1/75th of a second in question).  The lower your bitrate, the more aggressive (and audible) these decisions to throw away parts of the spectrum become.

[Striker Amplification]

Thank ya Jimi!

[R O Tiree]

Leakage is nearly-constant current source at micro-micro levels. It provides current, but at an incremental impedance that usually way higher than 1M. This was the substance of the Millenium Bypass, and my recently-posted idea to better-fake germaniums by adding a reverse biased germanium diode from collector to base of a silicon transistor to up the leakage to germanium levels.

You may need the current to bias, but its AC impedance is much higher than a resistor.

Now that makes perfect sense. Thanks, RG.

Some Schottkys have appreciable leakage as well... BAT46, perhaps? From the datasheet, at the voltages and temperatures we're typically going to be looking at, about 10-15µA? Might need 2 in parallel to get to the typical 30-odd µA quoted above by DigitalLarry.

Edit: made a mistake on the exponents values. Ignore that bit about Schottkys