4093 Oscillator noise

[Jaicen_solo]

Can anybody suggest any techniques for killing oscillator noise from a 4093 set up as an LFO?

I'm using the schem from Tonegod's Payback, which works ok. Unfortunately, it's introducing a fizzy square wave interference that I can't seem to filter out using conventional techniques.
I have a suspicion it's related to changes in current draw, but I have tried using the usual 100uF caps across the power pins on the chip itself, to no avail.

So, anybody have any suggestions before I switch to a TL072 relaxation osc?

[Jaicen_solo]

I've also got a digital noise bleeding through from the ISD chip itself. I'm not sure how much I should be expecting, but it's quite bad when it's in record mode. I'm sure it used to be dead quiet in record mode before, can't work it out.

EDIT: Forgot to mention, i'm working on a Payback using an ISD2560.

[earthtonesaudio]

If it was noise from the LFO it would be tick...tick...tick, not a constant hiss or buzz.  Probably coming from the big chip then.  Try killing the oscillator altogether to confirm.

[Jaicen_solo]

Actually, it was both.
I solved the noise from the ISD using massive capacitance across the digital supply rails.
The LFO is still ticking though. My thought is that it's slamming from rail to rail, and that's what's causing the ticking.
I've tried loads of things, but nothing works. I even tried the huge caps across the chips supply but that didn't help.

[earthtonesaudio]

An op-amp-based relaxation oscillator will have the same exact ticking problem.  The solution is to isolate the supply for the oscillator from the supply for the audio.  An extra battery is impractical so instead use RC filtering.

Here is a generic approach:


This looks a lot more complicated than it really is.  All I've added to your existing schematic is 100 ohms in series, plus a 10u cap to ground, between the battery and the positive power node on each subcircuit (oscillator and audio), and 100 ohms in series plus a 10u cap to ground from the negative power lead on the 4093.

The scope shots in this simulation illustrate that the tick in the audio is down about 30dB.  Making all the caps 100u will bring the attenuation down to about 80dB.

[Jaicen_solo]

Ok, that's certainly something to try.

Can you just clarify though; if i'm filtering a 4093, I put the resistor between +5v and pin 14, but which side of the resistor do I tie the capacitors to?

The side next to pin 14?

[earthtonesaudio]

The capacitors go from IC pins to a single star ground point, in all cases.

So for the 4093 one capacitor would go from pin 14 to ground, another capacitor would go from pin 7 to ground.  If your ground is awesome (very low impedance) you probably don't need the RC to pin 7, but in stompboxes the ground is in general not awesome.

[Jaicen_solo]

Ah, ok, but pin 7 is tied to ground. Why would I need to put a cap from ground to ground?

[earthtonesaudio]

Ah, ok, but pin 7 is tied to ground. Why would I need to put a cap from ground to ground?

You don't.  I misread your post.  In my previous schematic I put filtering on both sides of the 4093, so two resistors, two caps to ground for that IC alone.
If you're only filtering the pin 14 side you only need one resistor and one capacitor.

Also, in the simulation it looks like there is no change in performance for the version where only the positive rail is filtered.  Saves 2 components and works the same, so I'd call it better.
link

[Jaicen_solo]

Ok, i'll give it a go and i'll let you know how I get on.