audiobipolarmultiplexerpower
I'm looking to use a CD4051B multiplexer as a digitally-controlled switcher for audio signals.
The audio is coming from the HP / BP output of a voltage-controlled filter in a synth. The signal output is about 3v peak-to-peak. The mux select will be controlled by 5v TTL logic.
I only have supply voltages of 5v, 12v and -12v in my synth.
I'm finding the datasheets for the 4051 rather vague – has anyone had any success powering these chips with +/-12v bipolar power? (the -12v supply going to Vee)
Add capacitors between each power rail and ground. They'll act like short circuits to the high-frequency noise. I see that you have one between the -9 V rail and ground-- that's a good start.
The type of capacitor you use can make a difference. Electrolytic caps, for example, are relatively cheap, but they only deliver their rated capacitance at low frequencies, so they're not as good for filtering out high-frequency noise. I've had good luck with tantalum capacitors in this kind of application, but be careful that you don't plug them in backwards, as they will explode. Be double careful with the negative rail, because "backwards" might be the reverse of what you think. For both rails, you want the plus on the capacitor connected to the more positive terminal, which is ground for the -9 V rail.
You could start from a classic adjustable LDO application:
I picked the LP3874 because it has a very low dropout voltage of 24mV. By switching R2 between two values you can have 3.3V or 5V at the output, even with 5V in.
Calculations
If we pick 4700\$\Omega\$ for R1, then R2 must be 2700\$\Omega\$ for 3.3V out, and 1500\$\Omega\$ for 5V out. So for R2 place 1500\$\Omega\$ in series with 1200\$\Omega\$ and place a switch parallel to the 1200\$\Omega\$.
A BJT as switch is maybe not such a good idea, because of the \$V_{CE(SAT)}\$, but a small logic MOSFET is possible. Even a common BSS138 (with an \$R_{DS(ON)}\$ of 6\$\Omega\$) is usable.
edit
Not everybody seems to be convinced of the BSS138, but its resistance actually brings the voltages closer to their theoretical value than 0\$\Omega\$: 0.22% instead of 0.52% for the 5V, and 0.85% instead of 1% for the 3.3V.
Also it only has a \$V_{GS(th)}\$ of 1.5V maximum for \$I_D\$ = 1mA, which is about what flows through the resistor divider.
Best Answer
I suggest to have the audio signal centered at 6V and power the 4051 with GND and 12V (it works with voltage supply up to 15V) and use a simple level shifter from 5V TTL to 12V (either inverting as shown in the diagram below or non-inverting; whatever suits better) for the logic signals.
simulate this circuit – Schematic created using CircuitLab