Anyone ever thought about using Analog Switches

[ExpAnonColin]

such as the Maxim DG series, in place of LDRs/OTAs?  They'd have to be totally square wave (unless you worked out some triangle wave conversion post switch), but it would work well for ring modulation and choppy tremolo, no?

-Colin

[ExpAnonColin]

Well, I just ordered some samples, so I guess we'll find out

-Colin

[Mark Hammer]

They can actually be used in a number of creative ways.

MXR used analog switches (4066, 4016) in place of LDRs and FETs in several of their products.  The most au courant one would be the Envelope Filter, where the duty cycle of a switch section (how much time it spends open vs closed) sets the effective resistance in a filter section and tunes the filter as the duty cycle is varied.

They also used such switching in place of LDRs or FETs for one of their later models of phaser (post Phase 45/90/100).  Finally, they also used such switching to produce a tracking filter for the green analog delay pedal (which Ross later cloned almost note for note).

The nice thing about using them in this way is that precision matching of *effective resistance* is a breeze.

[ExpAnonColin]

Cool, I had a feeling they would have been used before.  It's unfortunate that they aren't as ramp-flexible as OTA's (or are they?).  I hope I didn't order SMD again...

-Colin

[R.G.]

It's unfortunate that they aren't as ramp-flexible as OTA's (or are they?).

They are if you clock them at much greater than audio frequencies (say, 48kHz and up) and pulse width modulate the on-time for them.

You can also use a SPDT analog switch plus a capacitor clocked at above-audio and PWM'ed to fake a series resistor.

Be aware of the dangers hidden in analog switches:
- they are plagued by spiking with the clock. This is commonly referred to as "charge injection" in the data sheets. You absolutely must deal with this in some way. Getting over the audio band is one.
- CMOS switches are prone to SCR latchup if you ever have a voltage fed to the in/out and the power to the chip is off. Series resistors are usually mandatory.

As Craig Anderton points out, you can ramp the control waveform to the CD 4016 analog switch on the N-channel driver anyway. This helps with charge injection.

Most modern analog switches have charge-injection balancing techniques built in so that the high and low side switches balance out if the DC level of the signal is near the middle of the switch power supply. Hence the DC biasing resistors in "Bypassing with the CD4053" at GEO.

[ExpAnonColin]

Thanks RG!  I was worried about voltage spikes, and was wondering if there would be any... and I didn't even think of pulse width modulating!

-Colin