switch topologies.

I once made a sample-and-hold, quite good at drift, using a MOSFET (BS170,
which appears close enough to 2N7000, etc.). Of course having three pins on
the MOSFET means it has a reverse diode, so I told the diffamp to saturate
to the negative rail when holding. It made one hell of a recovery transient
(all the way to +V), but it only lasted a microsecond. ;-)

You can also use diodes in similar fashion, though leakage is worse
(1N4148's cause visible drift (for periods of seconds) on the hold cap).
Speed is of course excellent, especially if you use, say, RF mixer
schottkies...

Most things of this sort will cause an asymmetrical pop when switching, but
some can be built balanced. You can use a bridge rectifier as a switch by
forward- or reverse-biasing the DC terminals, using the AC terminals as the
switch leads. Bias current must be greater than peak signal current, and
needs to be at the same potential as the signal (obviously, being locked
within a diode drop of it!), and should present little impedance to the
signal (implying CCS's). A possible solution presents itself using +/-
power supplies with equal CCS's to forward-bias the diodes, and a pair of
open collectors to pull the reverse bias. Because the capacitance, reverse
recovery (if any), current and etc. is all equal and opposite, there is
(ideally) no offset current or charge injection. Moreover, the turn-off
slope affects a quickly-moving input signal the same way, whether rising or
falling at the instant of turn-off; there is no skew of the slew rate when
disconnected from a fast-rising input.

Tim

--
Deep Fryer: A very philosophical monk.
Website @ http://webpages.charter.net/dawill/tmoranwms

"Jon Slaughter" wrote in message
...
Audio mute switch