One knob, one chip 'Coracle' tremolo: new design for your constructive criticism

[anotherjim]

The biggest noise source will be the inverters. They have non-linear gain and it's highest at the zero crossings, lowest at the peaks, so the device noise gets boosted most by the gain when it should be silent. This effect can only be reduced by picking resistor values that give only modest gain.
How much gain does this tremolo need?  Only as much as it needs to sound similar in volume to the bypassed level.

You can reduce series resistor noise by choosing metal film resistors.

[Radical CJ]

Here is an updated schematic implementing some of the improvements suggested above.
Thanks everyone for your help!





Signal Path

Addressing the problematically low input impedance, I have implemented the simplest and most space saving solution, which is simply to increase the input series resistor and feedback resistor on the first inverter in the signal path (using the same values as the input for ROG's UBE Screamer, as per Soggybag's suggestion).

If I understand correctly, this should increase the input impedance to about 470K (which is mostly provided by the input resistor). While more input impedance would be better (as would be an op amp buffer), this keeps the design as simple and small as possible, and achieves an input impedance greater than the popular commercial CMOS inverter based pedals that I'm aware of, such as the Red Llama with around 100k, and MXR envelope filter with 200K.

This change also fixes the high pass filter on the input, which I think now has a cut off of around 34Hz passing all guitar frequencies and then some? Or does this calculation need to also account for the pulldown resistor in some way?

The decoupling cap between the inverter buffers/stages in the guitar signal path has been removed as I can't see why this is necessary, and some comparable tremolo designs similarly omit the decoupling cap between op amp buffers.

Output volume is now determined by a trimmer, and the output capacitor should pass all guitar signal? However, as with the input, I'm a little unclear on the exact formula for calculating the cut off frequency when both a series resistor and pulldown resistor are involved.

LFO

As per Anotherjim's suggestion I have increased the value of the rate pot and changed the value of the integrator cap accordingly to reduce current kick and any associated noise. (I didn't have a 330n cap, so the LFO's range is now a bit different, but perhaps actually better). 

The suggestion made by Andy/Fancy Lime to put small caps between input and output of inverters 1 and 2 made a marked difference in removing the tick/flutter, with 10n being quite effective but 1n not that effective.

With the flutter gone there is still some noise. More like a pulsing hum at a frequency that sits somewhere in the guitar's mids range. But it is fairly quiet.

From here I think I'll try a perfboard version and check how it performs with different gear combinations. Although we're still in a COVID lockdown here, so I probably won't be able to test with any friend's rigs for a while. I'll also record some sound samples.

Cheers everyone, 

Joseph

[Steben]

The biggest noise source will be the inverters. They have non-linear gain and it's highest at the zero crossings, lowest at the peaks, so the device noise gets boosted most by the gain when it should be silent. This effect can only be reduced by picking resistor values that give only modest gain.
How much gain does this tremolo need?  Only as much as it needs to sound similar in volume to the bypassed level.

You can reduce series resistor noise by choosing metal film resistors.

Side note:
Not so long ago I was a bit tempered here in my noise remarks. But MOSfet based stuff like inverters do have an issue. High supply and high input signals will give better ratio's. The supply voltage will influence the clipping though. EDIT: low supply will render harsh clipping, high supply might give too soft muddy clipping, but some feedback might help. It is a trade off, but as I mentioned recently: stacking inverters does not help. One should IMHO try to keep the MOSfet stages for clipping, not so much for gain.Just an idea.
A drawback of a CMOS inverter is the fact it draws most current at center bias if used as an amplifier stage. The opposite of a class AB amp. It is a bad block to use as current drawing source to simulate sag. Which brings us back to good old opamp design ...

[Vivek]

http://www.runoffgroove.com/ubescreamer.html

Lovely name for a Tube screamer made out of CD4049UBE

[Radical CJ]

http://www.runoffgroove.com/ubescreamer.html

It's so good I think they must have come up with the name first, and then designed the circuit to go with it.

Lovely name for a Tube screamer made out of CD4049UBE