assafdarsagol
New member
I've been hot-rodding my Pro Filter and figured I'd share.
Disclaimer: I am not an expert in electronics, I am a DSP developer so I know my filters - but together with Claude Code and Kicad-Happy, I was able to iterate and improve the things that bothered me.
github.com
It started because I wanted to use it pre-distortion and currently it is too noisy for my liking. There were two types of noise i wanted to get rid of: a broad band white noise, and an oscillation at 17kHz.
To get rid of the broad band noise - I swapped both TL072s for OPA1656s, added 100nF decoupling right at the op-amp rails (six of them), bumped the charge-pump bulk caps to 100µF, That substantially improved the broadband noise situation and the noise dropped by about 10-12db - which is a LOT!
To take care of the oscillation I switched to a TC7660H to push the switching noise out of band. Once I did that The 17kHz oscillation was cleared.
However, I ran into a new issue: as you bring up the HP mix pot it chokes - probably due to high-frequency junk at the output mixer, and it takes a few seconds to clear.
It took me a while to track down: it's the two stages with pure resistive feedback (the summer/HP amp and the output mixer) going unstable up in the MHz range, somewhere the slower TL072 never reached. The fix is a small cap across each feedback resistor:
- 4pF across R2 (summer feedback)
- 82pF across R10 (mixer feedback)
The catch: R2 is inside the filter loop, so that cap has to be tiny. Bigger values kill the oscillation too, but they dump a pile of resonance into the top end. 4pF does the job and nothing more. R10 is outside the loop, so you can be generous there (82pF).
A couple of voicing mods while I was in there:
Flat resonance: it always bugged me that it resonates harder up high than down low. I matched the R2 cap with 4pF across R9 (LP feedback) and dropped C2 to about 10pF. That evens it out: now I get the same resonance across the whole sweep which sounds much more pleasing to the ear and refined.
Soft clipping: When filters are too linear, the resonance does not behave in a musical way. When it catches overtones it shoots in volume and sounds very digital and pokey. So I searched a way to tame it while keeping the filter sounding clean. For that I used antiparallel LEDs across the two integrator caps (C3 and C4) to round off the resonance when you push it.
I tried basically every diode and LED in my drawer, and the winner was a plain Tayda 3mm diffused red — not the water-clear ultrabright, those sound hairy. It keeps the resonance, but the louder peaks round off instead of spiking on whatever overtone they land on.
DC Blocker: I also retuned the output DC blocker for bass while I was at it — 470nF film and 27k, corner around 16Hz. Originally it was tuned to ~0.3Hz which I think lets a lot of knob turning junk through and takes too long to recover from dc offsets. This also recovers a lot of headroom from useless low frequencies.
I am still open to improving the mods and circuit so if you have suggestions or questions I'd be happy to discuss.
I will add some of the circuit board images in a later date as it is in the pedal boxes being used
Disclaimer: I am not an expert in electronics, I am a DSP developer so I know my filters - but together with Claude Code and Kicad-Happy, I was able to iterate and improve the things that bothered me.
GitHub - aklofas/kicad-happy: AI coding agent skills for KiCad electronics design. Works with Claude Code and OpenAI Codex. Analyze schematics, review PCB layouts, EMC pre-compliance, SPICE simulation, download datasheets, source components, and prep
AI coding agent skills for KiCad electronics design. Works with Claude Code and OpenAI Codex. Analyze schematics, review PCB layouts, EMC pre-compliance, SPICE simulation, download datasheets, sour...
github.com
To get rid of the broad band noise - I swapped both TL072s for OPA1656s, added 100nF decoupling right at the op-amp rails (six of them), bumped the charge-pump bulk caps to 100µF, That substantially improved the broadband noise situation and the noise dropped by about 10-12db - which is a LOT!
To take care of the oscillation I switched to a TC7660H to push the switching noise out of band. Once I did that The 17kHz oscillation was cleared.
However, I ran into a new issue: as you bring up the HP mix pot it chokes - probably due to high-frequency junk at the output mixer, and it takes a few seconds to clear.
It took me a while to track down: it's the two stages with pure resistive feedback (the summer/HP amp and the output mixer) going unstable up in the MHz range, somewhere the slower TL072 never reached. The fix is a small cap across each feedback resistor:
- 4pF across R2 (summer feedback)
- 82pF across R10 (mixer feedback)
The catch: R2 is inside the filter loop, so that cap has to be tiny. Bigger values kill the oscillation too, but they dump a pile of resonance into the top end. 4pF does the job and nothing more. R10 is outside the loop, so you can be generous there (82pF).
A couple of voicing mods while I was in there:
Flat resonance: it always bugged me that it resonates harder up high than down low. I matched the R2 cap with 4pF across R9 (LP feedback) and dropped C2 to about 10pF. That evens it out: now I get the same resonance across the whole sweep which sounds much more pleasing to the ear and refined.
Soft clipping: When filters are too linear, the resonance does not behave in a musical way. When it catches overtones it shoots in volume and sounds very digital and pokey. So I searched a way to tame it while keeping the filter sounding clean. For that I used antiparallel LEDs across the two integrator caps (C3 and C4) to round off the resonance when you push it.
I tried basically every diode and LED in my drawer, and the winner was a plain Tayda 3mm diffused red — not the water-clear ultrabright, those sound hairy. It keeps the resonance, but the louder peaks round off instead of spiking on whatever overtone they land on.
DC Blocker: I also retuned the output DC blocker for bass while I was at it — 470nF film and 27k, corner around 16Hz. Originally it was tuned to ~0.3Hz which I think lets a lot of knob turning junk through and takes too long to recover from dc offsets. This also recovers a lot of headroom from useless low frequencies.
I am still open to improving the mods and circuit so if you have suggestions or questions I'd be happy to discuss.
I will add some of the circuit board images in a later date as it is in the pedal boxes being used
