Electronic characteristics of an unpowered IC

[Taylor]

People sometimes find it useful to swap different ICs into a socket. There's all the tube screamer opamp stuff, which doesn't interest me that much, but there are other situations in which it can be interesting to swap compatible but different chips into a circuit. In The Tone God's "Christine", which is an inverter-based fuzz, an unbuffered chip is normally used, but I find some very interesting sounds by swapping in a buffered inverter chip.

The next step then, is to make a switch to swap internally. But switching all the pins is a drag.

Can we parallel, for example, 2 hex inverter chips, except for the power pin, and then select with an SPDT which one is powered up? Will the unpowered chip, whose inverters are all in parallel with the powered chip's, have any effect on the operation of the circuit? Or is an unpowered chip electronically invisible?

[blooze_man]

I think you should be good. Beavis's Screamer Lab had 3 or 4 IC's in the way you described.

http://www.beavisaudio.com/projects/ScreamerLab/

[PRR]

> is an unpowered chip electronically invisible?

Not very "invisible". Imagine diodes from every pin to both rails. In normal use, with + rail higher than - rail, these diodes are back-biased, no-effect. With + rail collapsed, you can't go more than a few tenths volt without hitting a diode.

There are ZIF sockets.

For the circuits you cite, duplicate circuits and a signal switch may be cheaper than a ZIF socket.

[Taylor]

Hmm. Interesting info. I already have a socket, and the insertion force is not an issue for me, but it's obviously more convenient to have a switch than to have to take the back off and swap chips. Not very performance-friendly.

I suppose a second box with the different chip is a better way to go, especially since in this circuit the different chip completely changes the sound in every way, not just golden ears tubescreamer stuff.

[acromarty]

You don't need to switch all the pins, you only need to select the output from one or other chip. The inputs can stay connected in parallel, with power applied to both.

As the previous posters have said, leaving an unpowered chip connected in circuit at best will give you an unwanted near short circuit load on the driven signals, and at worst might blow the unpowered chip.

[R.G.]

You don't need to switch all the pins, you only need to select the output from one or other chip. The inputs can stay connected in parallel, with power applied to both.

As the previous posters have said, leaving an unpowered chip connected in circuit at best will give you an unwanted near short circuit load on the driven signals, and at worst might blow the unpowered chip.

+1.

Or the powered chip.

Or both. 

The details of the chip internals matter, as Paul says, and the details that matter, such as where the substrate diodes go, what kind of isolation tubs exist under circuit sections, whether there is an internal bias voltage generator, and more, are somewhere between often and usually not mentioned or documented in spec sheets.

[Nasse]

I have faint memory of few hobby circuits authors telling that some chips can sometimes fail if fed signal when not powered. It was perhaps 7555 and the other some cmos logig used "wrong way" or for other purpose that it was designed

[JKowalski]

With the unbuffered inverter, can't you just put three in series to make your "buffered" inverter? Use the same chip but switch between the one stage and the three stage outputs.