CD40106 random LFO ticking

[Morocotopo]

Hi fellows, some time since I´ve been here. Here´s my story: I made a (pseudo) random LFO based on R.G.´s pseudo random article using the CD40106, actually I breadboarded the Ken Stone Psycho LFO, which is almost the same thing. Coupled that to a state variable filter controlled with LED/LDR´s set on lowpass. The idea was to make a random wah.
Well, the thing worked, but I got ticking from every change of state of each Schmitt trigger in the 40106. So a kind of background chirping that´s almost an effect on its own, but not what I wanted. In the scope I could see that the power (9V) had a very sharp blip for every time one of the triggers changed state.
So, on to the usual problem solvers: decoupling the power to the 40106 with R´s and C´s, decoupling the rest of the circuit, star grounding, inductor decoupling, big power filter caps (went up to 1000 uF, no dice)...
Nothing worked. I could reduce it some by using 1K resistors to decouple, but the power voltage went wayyy down and the LFO amplitude became too low. So I gave up. I think that this random LFO is unusable, at least for a stompbox. The 40106 is a logic chip, maybe unsuitable to use in a circuit that has audio?
Any ideas to make it usable?

[anotherjim]

I'm looking at this page....
http://www.elby-designs.com/webtek/cgs/cgs03/cgs03_psycho_lfo.html
What you changed in it for single 9v power might matter.
It looks tricky to do to me. At the output, it's intended to output a CV varying positive from ground only. The 40106 is powered from the negative supply only. The inverting input of the opamp turns the oscillator negative going outputs into positive going as required. Messing with that the wrong way, or trying to keep it the same probably means the opamp is being driven outside the allowed input range.
My first idea about that is that the first opamp IC2A is not required for single supply and can be replaced just by the 100k Range pot. It should continue to use the final amp, still of the LM358 type as that can follow the CV close to ground.

The various 40106 oscillator stages look reasonable, however, make sure you don't have any resistor values wrong. If the feedback resistors are too low then the oscillators can take too much current - although that should be noticeable as the oscillators will be too fast. In principle, using small capacitors and large resistors is the best way to keep those circuits from interfering with the power supply.

[patrick398]

I think that this random LFO is unusable, at least for a stompbox. The 40106 is a logic chip, maybe unsuitable to use in a circuit that has audio?
Any ideas to make it usable?

I built one of these psycho LFOs which i use with a delay for random madness and it works very well.
I used a 7660S to give me +/- 9v and never experienced any noise issues. I also used a transistor as a variable resistor rather than LED/LDR though.

Here is the layout i made if it's any use to you. Not the smallest or prettiest but i works



Sorry i can't offer any explanations as to why it's ticking, but i can confirm the circuit works well.

[Morocotopo]

I'm looking at this page....
http://www.elby-designs.com/webtek/cgs/cgs03/cgs03_psycho_lfo.html
What you changed in it for single 9v power might matter.
It looks tricky to do to me. At the output, it's intended to output a CV varying positive from ground only. The 40106 is powered from the negative supply only. The inverting input of the opamp turns the oscillator negative going outputs into positive going as required. Messing with that the wrong way, or trying to keep it the same probably means the opamp is being driven outside the allowed input range.
My first idea about that is that the first opamp IC2A is not required for single supply and can be replaced just by the 100k Range pot. It should continue to use the final amp, still of the LM358 type as that can follow the CV close to ground.

The various 40106 oscillator stages look reasonable, however, make sure you don't have any resistor values wrong. If the feedback resistors are too low then the oscillators can take too much current - although that should be noticeable as the oscillators will be too fast. In principle, using small capacitors and large resistors is the best way to keep those circuits from interfering with the power supply.

Jim, I powered it from +9 / Gnd and replaced the gnd with 1/2V. I also tried it with Gnd/ -9V using a 7660S charge pump set up to give -9V. Pretty much the same results. Although it gave better wah sweep powering it from +9/Gnd. I tried it with 1uF and 10 uF caps, the speed changed of course with the different values. But I might investigate your suggestions, I think the idea of smaller caps and bigger resistors is intriguing. Thanks.

I built one of these psycho LFOs which i use with a delay for random madness and it works very well.
I used a 7660S to give me +/- 9v and never experienced any noise issues. I also used a transistor as a variable resistor rather than LED/LDR though.

Patrick, interesting. A PT2399 based delay? As I said, I tried also a 7660S to power it but it didn´t work well for me. The LED/LDR thing supposedly slows down the transients because of the response lag of the LDR´s, so it seems a good idea for avoiding noises. But you´re not passing audio through the transistor, right, so it doesn´t matter in your case?
I connected all unused ins and outs of the 40106 to ground, is that OK?

Here´s the schem that I made, to see if anyone sees any problems with it.

[bluebunny]

I connected all unused ins and outs of the 40106 to ground, is that OK?

Ken's original circuit held the first unused input high.  The output from this unused inverter then fed the second unused one (in the interests of balancing the universe, I imagine).  The second output is unconnected.  Given that these are inverters, tying the output to the same place as the input is probably gonna end up in a fight.  Just pick a logic level for the inputs and leave the outputs unconnected.

[anotherjim]

That's why it's always good to post your exact schematic. The inverter outputs tied to ground will mean it takes excessive current. That's why adding a series resistor in the power supply caused too much voltage drop. Leave them disconnected. Since the inverters are arranged in the chip in the same direction, pcb layout is easier to connect an unused input pin to the adjacent unused output. The input at the start of the chain is tied to the nearest supply trace, negative or positive which for pin1 might be to pin 14 opposite.

The rest of it looks pretty good to me. It was sensible to give the LFO separate power filtering.

You could have more variation of the oscillators if they were CD4093 2 input Nand gates, then you can select some out of use or have an output gating an input on & off. Another is to add a trimmer pot in place of R25/26 to supply a variable reference to U4A pin3 to tweak the bias on the LED drives if you find it is sweeping the effect too much one way or the other.

[Morocotopo]

Bluebunny, Jim, great suggestions. Will try them and report back.

[PRR]

> inverters, tying the output to the same place as the input is probably gonna end up in a fight.

Over-simplified: no, the whole thing biases to follow "the + input". In a simple CMOS inverter, this is the internal threshold, around half supply.

BUT a CMOS is a crappy inverter which draws no current when slammed to a rail and BIG current when held half-way. Not big enough to melt (in 9V systems). Big enough to sag a battery or sag-network much harder than expected.

Yeah, un-wanted CMOS leftovers should have inputs tied Hi or Lo, outputs left unconnected. (Sometimes the easy way to hold an input is to strap it to an idle adjacent output pin, not to balance the universe but to KISS.)