# switching at zero cross/noiseless switching

Don Tillman don at till.com
Wed Nov 10 19:33:11 CET 1999

```   Date: Wed, 10 Nov 1999 09:03:26 -0800
From: Buck Buchanan <voltagecontrolled at home.com>

I'm wondering if there's anyone out there who can point me towards a
resource on switching audio at the zero crossing.  Or for that matter,
other techniques for noiseless switching at speed.  I'm open to using
specialized ICs.  Because of high speed, parts like the SSM2402 and
other "slow on/off" techniques won't work.

This is likely to be an interesting discussion...

Am I correct in thinking that even IF the switch itself adds no pop,
there will still be a pop since an instantanious switch at some
arbitrary point in the wave will cause a sharp transistion (fast rise
time) which itself will be the pop?

Yes, that's correct.

And furthermore, if I did manage to make it switch at the zero
crossing, wouldn't this be eliminated?

No it wouldn't.  You're thinking of delaying the switching until input
A crosses zero, right?  But then it's unlikely that input B is
crossing zero at that very moment and so switching to input B will
cause a click.  It would be no less noisey.

>> Proposal 2:
You could do a 2-step switch, like this:
Wait for input A to cross zero.
Switch from input A to mute.
Wait for input B to cross zero.
Switch from mute to input B.

>> Proposal 3:
But a much better approach is to ignore the zero crossing and simply
delay switching from input A to input B, waiting until the two inputs
are equal.

This might work well, I'd be interested in hearing it in a real synth
situation.

>> Proposal 4:
But as the clock frequency increases the delay might become
objectionable.  If you want to switch without a delay you could switch
immediately from input A to input B and then add in a decaying
exponential starting at the difference voltage.

A simple implementation would use a narrow pulse for switching from
input A to input B.  When the pulse goes high we charge up a capacitor
to inputA-inputB.  Then, when the pulse goes low the flip-flop changes
states and we switch from input A to input B and also add in the
capacitor voltage, and here the capacitor is no longer held to the
difference voltage, it just decays to zero.  You'd want to adjust
things so that the cap decay is close to zero by the time the next
switching event occurs.

-- Don

```