Still, the main issue is an awful output impedance, of the order that it's not only bad on paper, but giving volume jumps even when switching pedals with decent input impedances after it, and cable capacitance influence causing a varying amount of treble loss. I had solved this by adding an output buffer, which when traces became available I found out it's what the OCD v2 does too (I had also added a capacitor before it to emulate the treble loss of a normal cable). Another issue was the linear volume control, and then in general I thought it could have used a cleanup and standard E6 values for not critical parts. So here it is:
- Opamp. TL072, pretty much the standard for guitar inputs with the low current noise. Why use the slightly noisier 082?
- Input stage. This is pretty much verbatim, I just made the input impedance a standard 1Meg, but also removed the input resistor because its noise gets amplified by everything, altough small in value. It's a judgement call, noise and parts saving against current limiting and protection/reliability: I still haven't hit my face against issues caused by ESD on the input, so until then i choose the former. Gain stopper was removed to make this into an even cleaner boost if wanted: the original can already be very clean, why not go all the way and make the input stage unity gain. Gain control law doesn't suffer, my spreadsheet tells me.
- Clipping. Switched to plain 4148 diodes because the mosfets were used for the body diodes and there's not any justifiable reason to use them (a previous research I did didn't show differences at least in the VI curves), I had even made a comparison of the two and didn't hear much of a difference. They still go to vref, not because it's special but because it saves me a cap and a resistor.
- Tone. Moved between stages so it doesn't affect output impedance. "Hey why are the resistors the same, don't you have to subtract the 10K?". As a friend pointed out, when the diodes are conducting, they're in parallel with the 10K, so that resistor doesn't count much, exactly when the tone control it's most needed. This does mean in theory that the tone control cutoff varies dynamically with the amount of clipping, which i don't think it's that bad of a thing. Both simulations and side by side comparisons say that it's pretty much like the original. Log does give a better sweep but lin works well too.
- Output stage. Just a low gain clean boost with common values and low impedances to bring the unity gain at about noon on the volume pot and buffer the output. The gain has been chosen by having in mind headroom and an output amplitude roughly equal to guitar levels with volume at noon, starting from the assumption of a worst-case ~1.2Vpp output from the clipping stage.
This stage has the same bass treble rolloff frequencies as the original "recovery" stage, but has been redone for low impedances and common values.
- Volume. 10K log pot for a lower output impedance than you'll ever need and no weird interactions. Log for a smoother control.
Extra: I was working on the plastic timbre, a Plexitone workalike, at the same time as this. Since I enjoyed that circuit but didn't want to build it separately, recognizing the similarities, I've devised a mod for the C. OD to make it sound similar with little effort.