odd (?) CMOS switch behaviour

[tempus]

Hey all;

I was experimenting with a MAX4622 (http://datasheets.maxim-ic.com/en/ds/MAX4621-MAX4623.pdf)analog switch today, and I found that I needed to connect a 1M resistor to ground to get the thing to switch back to its default position when a control signal was applied to the control pin (i.e., pin 15 from the pinout diagram on p.1). Is this normal? I know that it's necessary when switching MOSFETs, but I didn't know that it applies (if it does) to CMOS switches. I didn't notice any in RG's page about using 4053s either. And strangest of all, I experimented with these a while back, and don't recall having to use one there either, although I may have had one in there anyway in an attempt to ramp down the control voltage (can't remember now).

Thanks

[composition4]

Yes, that's normal, CMOS inputs don't have a "default" state.

Also, leaving CMOS control inputs floating (i.e. no pull-up or pull-down resistors) is a bad idea due to the high impedance of CMOS inputs.  If you don't pull up or down it can easily pick up stray voltages and oscillate between states... leading to erratic behaviour at best, CMOS death at worst

Jonathan

[slacker]

Depends how you're switching the control pin. If it's connected to something that either gives 0 volts or the full supply voltage, like the output of another logic chip for example, then you shouldn't need a pull down.
Like Jonathan said though if it can float at any time then it will need a pull down or pull up resistor.

[tempus]

Depends how you're switching the control pin. If it's connected to something that either gives 0 volts or the full supply voltage, like the output of another logic chip for example, then you shouldn't need a pull down.

Right now I remember. I was switching it from a flipflop the last time I messed around with it, so it would have had 2 definite states.

Thanks for clearing that up.