Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!spool.mu.edu!uwm.edu!bionet!agate!ucbvax!dog.ee.lbl.gov!pasteur!zabriskie.berkeley.edu!spp From: spp@zabriskie.berkeley.edu (Steve Pope) Newsgroups: sci.electronics Subject: Re: CMOS Analog Switches Message-ID: <11728@pasteur.Berkeley.EDU> Date: 6 Mar 91 18:05:41 GMT References: <5292@umbc3.UMBC.EDU> Sender: news@pasteur.Berkeley.EDU Reply-To: spp@zabriskie.berkeley.edu.UUCP (Steve Pope) Distribution: na Organization: U.C. Berkeley -- ERL Lines: 52 In article <5292@umbc3.UMBC.EDU> hack@umbc5.umbc.edu () writes: > > I am tring to build a audio amplifier and I wish to have >it electronically switched (as few mechanical switches as >possible). What I was concidering is using CMOS bilateral >switches to handle this. What I am concerned is they might >produce a great deal of distortion to the signals. Now I know >that audio signals are typically 20-20kHz, still I don't know >if that would be a problem. > > I would like to know if there is a better way, or if >the CMOS chips are right for the jobs which ones are better >for this type of application. > >Thank you in advance, >J. Hack Don't use CMOS switches, use a single depletion-mode discrete JFET and observe the following: (1) place the switched FET(s) in series with the signal(s) at the summing node (- input) of an op amp. (+ input) of the op amp should be at ground. The input signal is fed through a resistor (say 10 Kohm), through the FET, into the summing node. (2) Use 0 V on the gate to switch ON. (3) Use a large enough voltage (negative for an N-channel FET) to turn the FET off so that any signal swing does not start to turn it back on. For eample, with a -2V pinch-off voltage, a signal that swings down to -12V, and a -15V voltage on the gate, you have 1V of margin which should be adequate to keep the FET off. (4) (this is important) couple the signal controlling the FET to the FET via an RC filter with 100 msec or so time constant. This will keep the circuit from clicking when you turn it on or off. (This is why you don't want to use a CMOS switch such as a 4016, they click). You should measure the distortion of each circuit built -- it should be well under .01% (in fact about .001%), if it is too high replace the FET. The trick is that by putting the FET at the summing node, with its on resistance being very small compared to the resistance in series with it, there is never any appreciable voltage across it, therefore the nonlinear I/V characteristic of the channel does not result in appreciable distortion. steve